I Had Codex Generate 310 Images with the Buzzworthy Qwen-Image 2.1, Then Had It Critique Them Brutally

Bottom line first: it is very strong at making images. But do not place too much trust in it yet for long text, precise local edits, or OpenPose reliability.

This time, I had Codex handle everything—from prompt creation and generation to image review and scoring—and tested Qwen-Image 2.1 across 310 images. I did not cherry-pick only the good results. Every failed image is included as-is, too.

Test Conditions

• Model: Qwen-Image 2.1 INT8 version (qwen-image-2.1-int8)
• Runtime environment: PurrStudio API
• Base settings: 40 steps, guidance scale 1.0, one image per job
• Standard T2I: 100 images
• Standard IT2I: 100 images
• Anime T2I: 50 images
• Anime IT2I: 50 images
• Three references—person, OpenPose skeleton, and background: 5 photorealistic images and 5 anime images
• IT2I reference images: up to 4 images
• Scoring: five criteria, each scored from 1 to 5; the 25-point total was multiplied by four to convert it to a 100-point scale
• Generation time: from started_at to finished_at, excluding queue wait time

The average score across all 310 images was 84.2/100. The recorded pure generation time averaged 21.0 seconds, with a median of 18.8 seconds. Initial loading time and server restart time during the run are not included.

Brutally Honest Overall Assessment

Qwen-Image 2.1 INT8 is quite powerful if all you want is the ability to create a single image normally. Photorealism, anime, period photography, lighting, and background art are consistent. Short text and logos were also surprisingly accurate.

On the other hand, long text broke down in Japanese, English, and Croatian alike. It can make something that looks like a magazine, but it is not yet suitable for uses where people need to read the body text. IT2I is strong at color changes and style changes, but edits like “change only this part and preserve everything else perfectly” regularly leak into other areas. OpenPose looks natural when it succeeds, but it also produced failures where the skeletal lines were drawn directly into the finished image.

Additionally, a CUDA driver error occurred with five references, so the final evaluation was standardized to a maximum of four images. With four images, it remained stable to the end.

Below are the results for all 310 images.

Standard T2I: 100 Images

T2I-001｜Everyday Smartphone Photo｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.61 seconds
Brief review: The intent of “everyday smartphone photo” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-002｜High-Resolution Portrait｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.46 seconds
Brief review: The intent of “high-resolution portrait” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-003｜Studio Portrait Shoot｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.52 seconds
Brief review: The intent of “studio portrait shoot” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-004｜Fashion Magazine Cover｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.4 seconds
Brief review: The intent of “fashion magazine cover” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-005｜Documentary Photograph｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.44 seconds
Brief review: The intent of “documentary photograph” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-006｜Street Snapshot｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.35 seconds
Brief review: The intent of “street snapshot” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-007｜Product Catalog｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.47 seconds
Brief review: The intent of “product catalog” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-008｜Architectural Photograph｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.49 seconds
Brief review: The intent of “architectural photograph” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-009｜Macro Photography｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.58 seconds
Brief review: The intent of “macro photography” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-010｜Film Still｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.52 seconds
Brief review: The intent of “film still” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-011｜Cel Animation｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.39 seconds
Brief review: The intent of “cel animation” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-012｜Shojo Manga｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.37 seconds
Brief review: The intent of “shojo manga” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-013｜Shonen Manga｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.45 seconds
Brief review: The intent of “shonen manga” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-014｜Western Comic｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.38 seconds
Brief review: The intent of “Western comic” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-015｜Watercolor Painting｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.42 seconds
Brief review: The intent of “watercolor painting” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-016｜Thickly Painted Oil Painting｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.28 seconds
Brief review: The intent of “thickly painted oil painting” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-017｜Colored Pencil｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.36 seconds
Brief review: The intent of “colored pencil” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-018｜Woodblock Print｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.39 seconds
Brief review: The intent of “woodblock print” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-019｜Papercut Art｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.32 seconds
Brief review: The intent of “papercut art” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-020｜Clay Animation｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.52 seconds
Brief review: The intent of “clay animation” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-021｜1850s Wet Plate｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.46 seconds
Brief review: The intent of “1850s wet plate” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-022｜1920s Black and White｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.51 seconds
Brief review: The intent of “1920s black and white” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-023｜1940s News Photography｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.46 seconds
Brief review: The intent of “1940s news photography” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-024｜1960s Color｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.55 seconds
Brief review: The intent of “1960s color” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-025｜1970s Instant Photo｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.69 seconds
Brief review: The intent of “1970s instant photo” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-026｜1980s Family Photo｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.54 seconds
Brief review: The intent of “1980s family photo” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-027｜1990s Compact Camera｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.52 seconds
Brief review: The intent of “1990s compact camera” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-028｜2000s Digital Camera｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.5 seconds
Brief review: The intent of “2000s digital camera” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-029｜VHS Still｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.5 seconds
Brief review: The intent of “VHS still” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-030｜Security Camera｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.42 seconds
Brief review: The intent of “security camera” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-031｜Front View｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.59 seconds
Brief review: The intent of “front view” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-032｜Perfect Side View｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.47 seconds
Brief review: The intent of “perfect side view” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-033｜Rear View｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.43 seconds
Brief review: The intent of “rear view” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-034｜45-Degree Angle｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.47 seconds
Brief review: The intent of “45-degree angle” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-035｜Low Angle｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.52 seconds
Brief review: The intent of “low angle” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-036｜Extreme Low Angle｜88/100

Evaluation: Prompt adherence 4/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.46 seconds
Brief review: The intent of “extreme low angle” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-037｜High Angle｜92/100

Evaluation: Prompt adherence 5/5 · Image quality 5/5 · Natural structure 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.49 seconds
Brief review: The intent of “high angle” is immediately clear. The details have some softness, but for T2I, this is a very straightforward output.

T2I-038｜Top-down view｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.51 seconds
Brief comment: The intent of “top-down view” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-039｜Ground level｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.47 seconds
Brief comment: The intent of “ground level” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-040｜Over-the-shoulder｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.5 seconds
Brief comment: The intent of “over-the-shoulder” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-041｜Ultra-wide angle｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.44 seconds
Brief comment: The intent of “ultra-wide angle” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-042｜Fisheye｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.5 seconds
Brief comment: The intent of “fisheye” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-043｜Standard 50mm｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.53 seconds
Brief comment: The intent of “standard 50mm” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-044｜Telephoto compression｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.48 seconds
Brief comment: The intent of “telephoto compression” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-045｜Macro｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.5 seconds
Brief comment: The intent of “macro” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-046｜Shallow depth of field｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.56 seconds
Brief comment: The intent of “shallow depth of field” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-047｜Deep focus｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.51 seconds
Brief comment: The intent of “deep focus” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-048｜Motion blur｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.5 seconds
Brief comment: The intent of “motion blur” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-049｜Long exposure｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.51 seconds
Brief comment: The intent of “long exposure” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-050｜Tilt-shift｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.52 seconds
Brief comment: The intent of “tilt-shift” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-051｜Overcast｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.63 seconds
Brief comment: The intent of “overcast” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-052｜Midday｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.49 seconds
Brief comment: The intent of “midday” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-053｜Golden hour｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.54 seconds
Brief comment: The intent of “golden hour” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-054｜Blue hour｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.51 seconds
Brief comment: The intent of “blue hour” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-055｜Window light｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.55 seconds
Brief comment: The intent of “window light” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-056｜Rembrandt lighting｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.51 seconds
Brief comment: The intent of “Rembrandt lighting” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-057｜Backlighting｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.55 seconds
Brief comment: The intent of “backlighting” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-058｜Red and blue neon｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.61 seconds
Brief comment: The intent of “red and blue neon” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-059｜Candlelight｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.54 seconds
Brief comment: The intent of “candlelight” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-060｜Volumetric lighting｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.62 seconds
Brief comment: The intent of “volumetric lighting” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-061｜Natural smile｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.44 seconds
Brief comment: The intent of “natural smile” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-062｜Big laugh｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.47 seconds
Brief comment: The intent of “big laugh” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-063｜Holding back tears｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.48 seconds
Brief comment: The intent of “holding back tears” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-064｜Crying intensely｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.48 seconds
Brief comment: The intent of “crying intensely” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-065｜Anger｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.46 seconds
Brief comment: The intent of “anger” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-066｜Surprise｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.42 seconds
Brief comment: The intent of “surprise” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-067｜Anxiety｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.56 seconds
Brief comment: The intent of “anxiety” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-068｜Suspicion｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.49 seconds
Brief comment: The intent of “suspicion” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-069｜Looking at the camera｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.52 seconds
Brief comment: The intent of “looking at the camera” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-070｜Eyes looking to the side｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.43 seconds
Brief comment: The intent of “eyes looking to the side” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-071｜Pixie cut｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.58 seconds
Brief comment: The intent of “pixie cut” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-072｜Long braid｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.43 seconds
Brief comment: The intent of “long braid” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-073｜Two-tone left and right｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.48 seconds
Brief comment: The intent of “two-tone left and right” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-074｜Complex braiding｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.43 seconds
Brief comment: The intent of “complex braiding” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-075｜Clear glasses｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.49 seconds
Brief comment: The intent of “clear glasses” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-076｜Three-layer winter outfit｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.47 seconds
Brief comment: The intent of “three-layer winter outfit” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-077｜Specified uniform｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.53 seconds
Brief comment: The intent of “specified uniform” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-078｜Mixed materials｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.49 seconds
Brief comment: The intent of “mixed materials” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-079｜Facial features｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.46 seconds
Brief comment: The intent of “facial features” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-080｜Three generations｜88/100

Assessment: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 88/100
Generation time: 15.49 seconds
Brief comment: The intent of “three generations” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-081｜Color order｜92/100

Assessment: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.43 seconds
Brief comment: The intent of “color order” comes across at a glance. The details are somewhat rough, but this is a very straightforward output for T2I.

T2I-082｜Exactly Three Items｜100/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 100/100
Generation time: 15.49 seconds
Brief review: The intent of “exactly three items” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-083｜Exactly Seven Items｜88/100

Evaluation: Instruction adherence 5/5 · Image quality 4/5 · Structural naturalness 4/5 · Text and quantity accuracy 5/5 · Practicality 4/5 → 88/100
Generation time: 15.43 seconds
Brief review: The placement of the seven birds is quite impressive. However, the outlines of small objects are a little soft.

T2I-084｜Above, Below, and Behind｜80/100

Evaluation: Instruction adherence 4/5 · Image quality 4/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 4/5 → 80/100
Generation time: 15.61 seconds
Brief review: The intent of “above, below, and behind” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-085｜Left and Right Hands｜92/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.5 seconds
Brief review: The intent of “left and right hands” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-086｜Three-Layer Space｜92/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.47 seconds
Brief review: The intent of “three-layer space” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-087｜Mirror Reflection｜76/100

Evaluation: Instruction adherence 4/5 · Image quality 4/5 · Structural naturalness 3/5 · Text and quantity accuracy 4/5 · Practicality 4/5 → 76/100
Generation time: 15.59 seconds
Brief review: It looks like a mirror, but a strict reading of the reflection relationship still raises doubts.

T2I-088｜Glass Refraction｜96/100

Evaluation: Instruction adherence 4/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 96/100
Generation time: 15.48 seconds
Brief review: The intent of “glass refraction” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-089｜Water Reflection｜92/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 4/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 92/100
Generation time: 15.43 seconds
Brief review: The intent of “water reflection” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-090｜Handshake｜100/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 100/100
Generation time: 15.53 seconds
Brief review: The intent of “handshake” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-091｜Short Japanese Text｜96/100

Evaluation: Instruction adherence 5/5 · Image quality 4/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 96/100
Generation time: 45.85 seconds
Brief review: The intent of “short Japanese text” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-092｜Short English Text｜100/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 100/100
Generation time: 27.38 seconds
Brief review: The intent of “short English text” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-093｜Short Croatian Text｜100/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 100/100
Generation time: 26.18 seconds
Brief review: The intent of “short Croatian text” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-094｜Long Japanese Text｜68/100

Evaluation: Instruction adherence 3/5 · Image quality 4/5 · Structural naturalness 5/5 · Text and quantity accuracy 2/5 · Practicality 3/5 → 68/100
Generation time: 26.12 seconds
Brief review: The long text merely captures the general feel and cannot be said to match the original exactly. The gap from short text was stark.

T2I-095｜Long English Text｜68/100

Evaluation: Instruction adherence 3/5 · Image quality 4/5 · Structural naturalness 5/5 · Text and quantity accuracy 2/5 · Practicality 3/5 → 68/100
Generation time: 26.28 seconds
Brief review: The long English text also contained omissions and paraphrasing. Proofreading is essential before using it as readable copy.

T2I-096｜Long Croatian Text｜56/100

Evaluation: Instruction adherence 2/5 · Image quality 4/5 · Structural naturalness 5/5 · Text and quantity accuracy 1/5 · Practicality 2/5 → 56/100
Generation time: 26.13 seconds
Brief review: The long Croatian text shows conspicuous spelling breakdowns. It looks polished, but is unsuitable for body-copy use.

T2I-097｜Trilingual Poster｜96/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 4/5 · Practicality 5/5 → 96/100
Generation time: 26.21 seconds
Brief review: The intent of “trilingual poster” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-098｜Japanese Logo｜100/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 100/100
Generation time: 26.44 seconds
Brief review: The intent of “Japanese logo” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-099｜Croatian Logo｜100/100

Evaluation: Instruction adherence 5/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 5/5 · Practicality 5/5 → 100/100
Generation time: 25.99 seconds
Brief review: The intent of “Croatian logo” is immediately clear. Some details are a little soft, but as T2I output, it is quite straightforward.

T2I-100｜Magazine-Style Trilingual Long Text｜68/100

Evaluation: Instruction adherence 3/5 · Image quality 5/5 · Structural naturalness 5/5 · Text and quantity accuracy 1/5 · Practicality 3/5 → 68/100
Generation time: 26.29 seconds
Brief review: The magazine-like layout is strong. But the body text contains a great deal of pseudo-text, making it unusable for long-form content meant to be read.

100 Standard IT2I Tests

IT2I-101｜Change Only the Clothing Color｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference integration 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 19.56 seconds
Brief review: “Change only the clothing color” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-102｜Change Only the Eye Color｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference integration 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.74 seconds
Brief review: “Change only the eye color” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-103｜Change Only the Hair Color｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference integration 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.76 seconds
Brief review: “Change only the hair color” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-104｜Change Only the Hairstyle｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference integration 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.83 seconds
Brief review: “Change only the hairstyle” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-105｜Change Only the Expression to a Smile｜88/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 5/5 · Reference integration 4/5 · Lack of artifacts 5/5 · Practicality 4/5 → 88/100
Generation time: 18.89 seconds
Brief review: “Change only the expression to a smile” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-106｜Add Glasses｜68/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 5/5 · Reference integration 2/5 · Lack of artifacts 5/5 · Practicality 3/5 → 68/100
Generation time: 18.79 seconds
Brief review: The intent to add glasses was weak, with the result leaning too heavily toward preserving the original image. The edit was overly conservative.

IT2I-107｜Remove the Hat｜64/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 5/5 · Reference integration 2/5 · Lack of artifacts 5/5 · Practicality 2/5 → 64/100
Generation time: 18.92 seconds
Brief review: The original image did not have a clearly defined hat, so this was effectively a no-op. It also loses points due to the test design.

IT2I-108｜Replace the Wristwatch｜64/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 5/5 · Reference integration 2/5 · Lack of artifacts 5/5 · Practicality 2/5 → 64/100
Generation time: 18.94 seconds
Brief review: Localized replacement of the wristwatch did not succeed. This exposed weaknesses in subject recognition and localized editing.

IT2I-109｜White Background Only｜64/100

Evaluation: Edit fidelity 3/5 · Preservation of unedited areas 3/5 · Reference integration 3/5 · Lack of artifacts 4/5 · Practicality 3/5 → 64/100
Generation time: 18.71 seconds
Brief review: “White background only” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-110｜Change the Label Color｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.77 seconds
Brief review: “Change the label color” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-111｜Change the Outfit｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.78 seconds
Brief review: “Change the outfit” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-112｜Age the Subject｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.94 seconds
Brief review: “Age the subject” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-113｜Make the Subject Younger｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.94 seconds
Brief review: “Make the subject younger” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-114｜Make the Hair Long｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.76 seconds
Brief review: “Make the hair long” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-115｜Change the Expression While Preserving Everything Else｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.78 seconds
Brief review: “Change the expression while preserving everything else” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-116｜Change the Pose｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.88 seconds
Brief review: “Change the pose” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-117｜Change to a Seated Position｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.85 seconds
Brief review: “Change to a seated position” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-118｜Change the Angle｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.81 seconds
Brief review: “Change the angle” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-119｜Anime Conversion｜88/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 4/5 · Reference integration 5/5 · Lack of artifacts 4/5 · Practicality 4/5 → 88/100
Generation time: 18.72 seconds
Brief review: “Anime conversion” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-120｜Convert to the 1920s｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.68 seconds
Brief review: “Convert to the 1920s” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-121｜Nightscape Background｜92/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 5/5 · Practicality 5/5 → 92/100
Generation time: 19.01 seconds
Brief review: “Nightscape background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-122｜Forest Background｜92/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 5/5 · Practicality 5/5 → 92/100
Generation time: 18.86 seconds
Brief review: “Forest background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-123｜Make It Rainy｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.81 seconds
Brief review: “Make it rainy” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-124｜Day to Night｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference integration 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.81 seconds
Brief review: “Day to night” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details together.

IT2I-125｜Summer to Winter｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.88 seconds
Brief comment: “Summer to Winter” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-126｜Warm Lighting｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.83 seconds
Brief comment: “Warm Lighting” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-127｜Background Blur｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.74 seconds
Brief comment: “Background Blur” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-128｜Remove Passersby｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.86 seconds
Brief comment: “Remove Passersby” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-129｜Transparent Background｜52/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 3/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 52/100
Generation time: 18.82 seconds
Brief comment: “Transparent Background” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-130｜Black and White Conversion｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.74 seconds
Brief comment: “Black and White Conversion” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-131｜Old Photo Restoration｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.8 seconds
Brief comment: “Old Photo Restoration” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-132｜Colorization｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.85 seconds
Brief comment: “Colorization” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-133｜Contrast Adjustment｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.79 seconds
Brief comment: “Contrast Adjustment” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-134｜Face Restoration｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.78 seconds
Brief comment: “Face Restoration” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-135｜Fill Missing Areas｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.85 seconds
Brief comment: “Fill Missing Areas” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-136｜Left Outpainting｜56/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 4/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 56/100
Generation time: 18.87 seconds
Brief comment: Although left and right expansion was requested, there is no dedicated canvas control, so this is not effectively outpainting.

IT2I-137｜Landscape Expansion｜56/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 4/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 56/100
Generation time: 18.89 seconds
Brief comment: Despite the request for a landscape format, the composition remains largely unchanged. The lack of dedicated controls on the API side hurts performance.

IT2I-138｜Portrait Expansion｜56/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 4/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 56/100
Generation time: 18.72 seconds
Brief comment: Vertical expansion is weak. Controlling the canvas through prompting alone is impractical.

IT2I-139｜Add Detail｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.7 seconds
Brief comment: “Add Detail” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-140｜Center-Preserved Panorama｜56/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 4/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 56/100
Generation time: 18.83 seconds
Brief comment: A center-preserved panorama was not achieved. It fails to both retain the original image and extend it.

IT2I-141｜Sign Word Replacement｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.88 seconds
Brief comment: “Sign Word Replacement” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-142｜Change Product Name｜84/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 5/5 · Practicality 4/5 → 84/100
Generation time: 18.92 seconds
Brief comment: “Change Product Name” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-143｜Change Date｜52/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 3/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 52/100
Generation time: 18.89 seconds
Brief comment: “Change Date” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-144｜Change Price｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.91 seconds
Brief comment: “Change Price” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-145｜Japanese-to-English Translation｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.95 seconds
Brief comment: “Japanese-to-English Translation” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-146｜English-to-Japanese Translation｜48/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 2/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 48/100
Generation time: 18.87 seconds
Brief comment: It was supposed to replace the English with Japanese, but retained both languages. It failed to follow the “only” in the editing instruction.

IT2I-147｜Change Font Only｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 19.03 seconds
Brief comment: “Change Font Only” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-148｜Change Text Color Only｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.81 seconds
Brief comment: “Change Text Color Only” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-149｜Add Heading｜100/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.87 seconds
Brief comment: “Add Heading” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-150｜Remove Text Only｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.9 seconds
Brief comment: “Remove Text Only” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-151｜Clothing Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 27.9 seconds
Brief comment: “Clothing Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-152｜Hairstyle Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.93 seconds
Brief comment: “Hairstyle Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-153｜Glasses Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.94 seconds
Brief comment: “Glasses Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-154｜Makeup Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.74 seconds
Brief comment: “Makeup Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-155｜Material Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.85 seconds
Brief comment: “Material Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-156｜Color Palette Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.84 seconds
Brief comment: “Color Palette Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-157｜Art Style Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.64 seconds
Brief comment: “Art Style Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-158｜Background Transfer｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.76 seconds
Brief comment: “Background Transfer” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-159｜Animal Costume｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.57 seconds
Brief comment: “Animal Costume” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-160｜Add Furniture｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.8 seconds
Brief comment: “Add Furniture” captures the core editing intent. However, as more references are added, it becomes less effective at keeping their roles separate and tends to blend details together.

IT2I-161｜Upright Pose｜92/100

Evaluation: Edit fidelity 5/5 · Preservation of unedited areas 4/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 4/5 → 92/100
Generation time: 27.75 seconds
Brief comment: It faithfully retained the skeletal lines and naturally shifted to an upright pose. A success as an OpenPose reference.

IT2I-162｜One-Arm-Raised Pose｜88/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 5/5 · Reference adherence 4/5 · Lack of artifacts 5/5 · Practicality 4/5 → 88/100
Generation time: 27.51 seconds
Brief comment: It captured the direction of the one-arm-raised pose, but prioritized preserving the person, so the pose change was restrained.

IT2I-163｜Seated Pose｜84/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Lack of artifacts 5/5 · Practicality 4/5 → 84/100
Generation time: 27.56 seconds
Brief comment: The conversion to a seated pose is natural. It opts for visual consistency over an exact skeletal match.

IT2I-164｜Running Pose｜36/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 2/5 · Reference adherence 2/5 · Lack of artifacts 2/5 · Practicality 1/5 → 36/100
Generation time: 27.45 seconds
Brief comment: The skeletal lines have bled into the person. It interpreted them as elements to draw rather than as pose references.

IT2I-165｜Twisting Pose｜48/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 3/5 · Reference adherence 2/5 · Lack of artifacts 3/5 · Practicality 2/5 → 48/100
Generation time: 27.41 seconds
Brief comment: Skeleton adherence is weak, and it was pulled toward the composition of the original image.

IT2I-166｜Arms-Crossed Pose｜48/100

Evaluation: Edit fidelity 2/5 · Preservation of unedited areas 2/5 · Reference adherence 2/5 · Lack of artifacts 4/5 · Practicality 2/5 → 48/100
Generation time: 27.56 seconds
Brief comment: The person was turned into a mannequin. Shape imitation wins out over understanding the meaning of the skeleton.

IT2I-167｜Hand and Finger Pose｜28/100

Evaluation: Edit fidelity 1/5 · Preservation of unedited areas 2/5 · Reference adherence 1/5 · Lack of artifacts 2/5 · Practicality 1/5 → 28/100
Generation time: 27.73 seconds
Brief comment: A complete failure case that drew the OpenPose lines directly.

IT2I-168｜Sideways Pose｜28/100

Evaluation: Edit fidelity 1/5 · Preservation of unedited areas 2/5 · Reference adherence 1/5 · Fewest artifacts 2/5 · Usability 1/5 → 28/100
Generation time: 27.59 seconds
Brief review: Skeleton lines remained over the insect. The correspondence between the subject and the skeleton has broken down.

IT2I-169｜Contour Guide｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Fewest artifacts 4/5 · Usability 4/5 → 80/100
Generation time: 26.61 seconds
Brief review: “Contour Guide” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-170｜Depth Guide｜80/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 4/5 · Fewest artifacts 4/5 · Usability 4/5 → 80/100
Generation time: 26.63 seconds
Brief review: “Depth Guide” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-171｜Person · Clothing · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 30.13 seconds
Brief review: “Person · Clothing · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-172｜Person · Hair · Outfit｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 30.0 seconds
Brief review: “Person · Hair · Outfit” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-173｜Person · Pose · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 29.88 seconds
Brief review: “Person · Pose · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-174｜Product · Material · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 30.15 seconds
Brief review: “Product · Material · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-175｜Character · Art Style · Pose｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 30.08 seconds
Brief review: “Character · Art Style · Pose” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-176｜Face · Body · Outfit｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 29.98 seconds
Brief review: “Face · Body · Outfit” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-177｜Two People + Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 30.05 seconds
Brief review: “Two People + Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-178｜Building · Season · Light｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 30.05 seconds
Brief review: “Building · Season · Light” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-179｜Room · Sofa · Lighting｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 30.0 seconds
Brief review: “Room · Sofa · Lighting” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-180｜Dish · Tableware · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 29.95 seconds
Brief review: “Dish · Tableware · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-181｜Person · Clothing · Shoes · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.44 seconds
Brief review: “Person · Clothing · Shoes · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-182｜Person · Hair · Hat · Glasses｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.48 seconds
Brief review: “Person · Hair · Hat · Glasses” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-183｜Person · Outfit · Pose · Art Style｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.44 seconds
Brief review: “Person · Outfit · Pose · Art Style” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-184｜Two People · Pose · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.4 seconds
Brief review: “Two People · Pose · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-185｜Product · Label · Material · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.5 seconds
Brief review: “Product · Label · Material · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-186｜Room · Desk · Chair · Lighting｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.3 seconds
Brief review: “Room · Desk · Chair · Lighting” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-187｜Front · Side · Back · Pose｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.33 seconds
Brief review: “Front · Side · Back · Pose” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-188｜Car Body · Paint · Wheels · Location｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.17 seconds
Brief review: “Car Body · Paint · Wheels · Location” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-189｜Building · Landscaping · Sky · Time｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.28 seconds
Brief review: “Building · Landscaping · Sky · Time” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-190｜Main Character · Supporting Character · Prop · Location｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.45 seconds
Brief review: “Main Character · Supporting Character · Prop · Location” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-191｜Person · Jacket · Pants · Shoes · Bag｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 36.97 seconds
Brief review: “Person · Jacket · Pants · Shoes · Bag” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-192｜Person · Hair · Makeup · Glasses · Hat｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.97 seconds
Brief review: “Person · Hair · Makeup · Glasses · Hat” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-193｜Person · Outfit · Pose · Background · Art Style｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.68 seconds
Brief review: “Person · Outfit · Pose · Background · Art Style” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-194｜Product · Logo · Material · Pedestal · Background｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 34.56 seconds
Brief review: “Product · Logo · Material · Pedestal · Background” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-195｜Three People · Background · Composition｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 34.15 seconds
Brief review: “Three People · Background · Composition” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-196｜Main Character · Companion · Vehicle · Background · Light｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 34.42 seconds
Brief review: “Main Character · Companion · Vehicle · Background · Light” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-197｜Front · Side · Back · Expression · Pose｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 34.15 seconds
Brief review: “Front · Side · Back · Expression · Pose” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-198｜Room · Sofa · Desk · Lighting · Decor｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 33.99 seconds
Brief review: “Room · Sofa · Desk · Lighting · Decor” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-199｜Dish · Plate · Cutlery · Tabletop · Light｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 34.35 seconds
Brief review: “Dish · Plate · Cutlery · Tabletop · Light” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

IT2I-200｜Five-Person Group Photo｜76/100

Evaluation: Edit fidelity 4/5 · Preservation of unedited areas 4/5 · Reference adherence 3/5 · Fewest artifacts 4/5 · Usability 4/5 → 76/100
Generation time: 34.29 seconds
Brief review: “Five-Person Group Photo” captures the core editing intent. However, as more references are added, it becomes less effective at separating their roles and tends to blend details.

50 Anime T2I Images

ANIME-T2I-001｜Modern Cel Anime｜92/100

Evaluation: Prompt fidelity 5/5 · Anime image quality 5/5 · Natural structure 4/5 · Genre reproduction 4/5 · Usability 5/5 → 92/100
Generation time: 15.99 seconds
Brief review: “Modern Cel Anime” is highly polished as anime, with cohesive linework, color, and screen composition. It also distinguishes the specified genre well.

ANIME-T2I-002｜Shōjo Manga｜92/100

Evaluation: Prompt fidelity 5/5 · Anime image quality 5/5 · Natural structure 4/5 · Genre reproduction 4/5 · Usability 5/5 → 92/100
Generation time: 15.6 seconds
Brief review: “Shōjo Manga” is highly polished as anime, with cohesive linework, color, and screen composition. It also distinguishes the specified genre well.

ANIME-T2I-003｜Shōnen Manga｜92/100

Evaluation: Prompt fidelity 5/5 · Anime image quality 5/5 · Natural structure 4/5 · Genre reproduction 4/5 · Usability 5/5 → 92/100
Generation time: 15.76 seconds
Brief review: “Shōnen Manga” is highly polished as anime, with cohesive linework, color, and screen composition. It also distinguishes the specified genre well.

ANIME-T2I-004｜1990s OVA｜92/100

Evaluation: Prompt fidelity 5/5 · Anime image quality 5/5 · Natural structure 4/5 · Genre reproduction 4/5 · Usability 5/5 → 92/100
Generation time: 15.71 seconds
Brief review: “1990s OVA” is highly polished as anime, with cohesive linework, color, and screen composition. It also distinguishes the specified genre well.

ANIME-T2I-005｜1980s Sci-Fi｜88/100

Evaluation: Prompt fidelity 4/5 · Anime image quality 5/5 · Natural structure 4/5 · Genre reproduction 4/5 · Usability 5/5 → 88/100
Generation time: 15.84 seconds
Brief review: “1980s Sci-Fi” is highly polished as anime, with cohesive linework, color, and screen composition. It also distinguishes the specified genre well.

ANIME-T2I-006｜Anime Film Background｜92/100

Evaluation: Prompt fidelity 5/5 · Anime image quality 5/5 · Natural structure 4/5 · Genre reproduction 4/5 · Usability 5/5 → 92/100
Generation time: 15.7 seconds
Brief review: “Anime Film Background” is highly polished as anime, with cohesive linework, color, and screen composition. It also distinguishes the specified genre well.

ANIME-T2I-007｜Chibi Character｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.62 seconds
Brief review: “Chibi Character” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-008｜Dark Fantasy｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.64 seconds
Brief review: “Dark Fantasy” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-009｜Sports Anime｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.68 seconds
Brief review: “Sports Anime” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-010｜Slice of Life｜88/100

Evaluation: Prompt Adherence 4/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 88/100
Generation time: 15.71 seconds
Brief review: “Slice of Life” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-011｜Magical Girl｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.69 seconds
Brief review: “Magical Girl” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-012｜Mecha｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.7 seconds
Brief review: “Mecha” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-013｜Cyberpunk｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.67 seconds
Brief review: “Cyberpunk” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-014｜Japanese-Style Fantasy｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.69 seconds
Brief review: “Japanese-Style Fantasy” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-015｜School｜88/100

Evaluation: Prompt Adherence 4/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 88/100
Generation time: 15.67 seconds
Brief review: “School” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-016｜Isekai｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.67 seconds
Brief review: “Isekai” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-017｜Horror｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.74 seconds
Brief review: “Horror” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-018｜Cooking｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.77 seconds
Brief review: “Cooking” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-019｜Idol｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.78 seconds
Brief review: “Idol” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-020｜Historical Drama｜88/100

Evaluation: Prompt Adherence 4/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 88/100
Generation time: 15.61 seconds
Brief review: “Historical Drama” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-021｜Face Close-Up｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.63 seconds
Brief review: “Face Close-Up” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-022｜Full-Body Character Sheet｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.62 seconds
Brief review: “Full-Body Character Sheet” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-023｜Profile View｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.74 seconds
Brief review: “Profile View” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-024｜Rear View｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.73 seconds
Brief review: “Rear View” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-025｜Bird's-Eye View｜88/100

Evaluation: Prompt Adherence 4/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 88/100
Generation time: 15.74 seconds
Brief review: “Bird's-Eye View” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-026｜Low-Angle Shot｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.71 seconds
Brief review: “Low-Angle Shot” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-027｜Running Action｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.71 seconds
Brief review: “Running Action” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-028｜Sword Fight｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.68 seconds
Brief review: “Sword Fight” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-029｜Dance｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.62 seconds
Brief review: “Dance” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-030｜Ensemble Cast｜76/100

Evaluation: Prompt Adherence 4/5 · Anime Image Quality 4/5 · Structural Naturalness 3/5 · Genre Reproduction 4/5 · Practicality 4/5 → 76/100
Generation time: 15.65 seconds
Brief review: “Ensemble Cast” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-031｜Sunrise｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.84 seconds
Brief review: “Sunrise” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-032｜Moonlight｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.7 seconds
Brief review: “Moonlight” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-033｜Neon｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.68 seconds
Brief review: “Neon” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-034｜Backlighting｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.82 seconds
Brief review: “Backlighting” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-035｜Rain｜88/100

Evaluation: Prompt Adherence 4/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 88/100
Generation time: 15.7 seconds
Brief review: “Rain” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-036｜Snow｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.65 seconds
Brief review: “Snow” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-037｜Flames｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.73 seconds
Brief review: “Flames” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-038｜Underwater｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.71 seconds
Brief review: “Underwater” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-039｜Magic Effects｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.64 seconds
Brief review: “Magic Effects” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-040｜Speed Effects｜88/100

Evaluation: Prompt Adherence 4/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 88/100
Generation time: 15.69 seconds
Brief review: “Speed Effects” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-041｜Expression Sheet｜100/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 5/5 · Genre Reproduction 5/5 · Practicality 5/5 → 100/100
Generation time: 15.77 seconds
Brief review: “Expression Sheet” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-042｜Three-View Drawing｜100/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 5/5 · Genre Reproduction 5/5 · Practicality 5/5 → 100/100
Generation time: 15.63 seconds
Brief review: “Three-View Drawing” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-043｜Costume Design｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.6 seconds
Brief review: “Costume Design” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-044｜Prop Sheet｜92/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 4/5 · Genre Reproduction 4/5 · Practicality 5/5 → 92/100
Generation time: 15.62 seconds
Brief review: “Prop Sheet” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-045｜Four-Panel Comic｜100/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 5/5 · Genre Reproduction 5/5 · Practicality 5/5 → 100/100
Generation time: 15.63 seconds
Brief review: “Four-Panel Comic” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-046｜Anime Poster｜100/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 5/5 · Genre Reproduction 5/5 · Practicality 5/5 → 100/100
Generation time: 15.7 seconds
Brief review: “Anime Poster” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-047｜English Title｜96/100

Evaluation: Prompt Adherence 5/5 · Anime Image Quality 5/5 · Structural Naturalness 5/5 · Genre Reproduction 4/5 · Practicality 5/5 → 96/100
Generation time: 15.67 seconds
Brief review: “English Title” is highly polished as anime, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-048｜Croatian Title｜100/100

Rating: Instruction fidelity 5/5 · Anime image quality 5/5 · Structural naturalness 5/5 · Genre reproduction 5/5 · Practicality 5/5 → 100/100
Generation time: 15.7 seconds
Brief review: “Croatian Title” is highly polished as an anime image, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-049｜Japanese Logo｜100/100

Rating: Instruction fidelity 5/5 · Anime image quality 5/5 · Structural naturalness 5/5 · Genre reproduction 5/5 · Practicality 5/5 → 100/100
Generation time: 15.62 seconds
Brief review: “Japanese Logo” is highly polished as an anime image, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

ANIME-T2I-050｜Magazine Cover｜100/100

Rating: Instruction fidelity 5/5 · Anime image quality 5/5 · Structural naturalness 5/5 · Genre reproduction 5/5 · Practicality 5/5 → 100/100
Generation time: 15.72 seconds
Brief review: “Magazine Cover” is highly polished as an anime image, with cohesive linework, color, and composition. It also strongly distinguishes the specified genre.

50 Anime IT2I Images

ANIME-IT2I-001｜Hair Color Change｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 19.42 seconds
Brief review: “Hair Color Change” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-002｜Smiling Expression｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.87 seconds
Brief review: “Smiling Expression” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-003｜School Uniform Outfit｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.99 seconds
Brief review: “School Uniform Outfit” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-004｜Night Scene Conversion｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.93 seconds
Brief review: “Night Scene Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-005｜Rainy Weather Conversion｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.86 seconds
Brief review: “Rainy Weather Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-006｜Transparent Background｜28/100

Rating: Edit fidelity 1/5 · Character preservation 2/5 · Reference adherence 1/5 · Lack of artifacts 2/5 · Practicality 1/5 → 28/100
Generation time: 18.84 seconds
Brief review: Background transparency failed, leaving a purple background. It is unusable for RGBA applications.

ANIME-IT2I-007｜Chibi Conversion｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.98 seconds
Brief review: “Chibi Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-008｜1990s Style Conversion｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.95 seconds
Brief review: “1990s Style Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-009｜Watercolor Anime Conversion｜96/100

Rating: Edit fidelity 5/5 · Character preservation 4/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 96/100
Generation time: 18.74 seconds
Brief review: “Watercolor Anime Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-010｜Poster Conversion｜96/100

Rating: Edit fidelity 5/5 · Character preservation 4/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 96/100
Generation time: 19.0 seconds
Brief review: “Poster Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-011｜Hairstyle Change｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.93 seconds
Brief review: “Hairstyle Change” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-012｜Add Glasses｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 19.02 seconds
Brief review: “Add Glasses” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-013｜Remove Weapon｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.93 seconds
Brief review: “Remove Weapon” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-014｜Winter Season｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 19.05 seconds
Brief review: “Winter Season” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-015｜Backlighting Conversion｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.94 seconds
Brief review: “Backlighting Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-016｜Landscape Conversion｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.84 seconds
Brief review: “Landscape Conversion” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-017｜Angry Expression｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 19.02 seconds
Brief review: “Angry Expression” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-018｜Color Tone Change｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.98 seconds
Brief review: “Color Tone Change” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-019｜Background Simplification｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 18.83 seconds
Brief review: “Background Simplification” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-020｜Line Art Extraction｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 18.86 seconds
Brief review: The line-art conversion is clear, and the contours are well organized. The result is easy to use as an edit.

ANIME-IT2I-021｜Outfit Transfer｜92/100

Rating: Edit fidelity 5/5 · Character preservation 4/5 · Reference adherence 5/5 · Lack of artifacts 4/5 · Practicality 5/5 → 92/100
Generation time: 27.52 seconds
Brief review: The separation between the roles of the outfit and character is relatively clean. However, it does not achieve a perfect match in every detail.

ANIME-IT2I-022｜Pose Transfer｜92/100

Rating: Edit fidelity 5/5 · Character preservation 4/5 · Reference adherence 5/5 · Lack of artifacts 4/5 · Practicality 5/5 → 92/100
Generation time: 27.16 seconds
Brief review: “Pose Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-023｜Art Style Transfer｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 26.76 seconds
Brief review: “Art Style Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-024｜Background Transfer｜88/100

Rating: Edit fidelity 4/5 · Character preservation 5/5 · Reference adherence 4/5 · Lack of artifacts 5/5 · Practicality 4/5 → 88/100
Generation time: 27.34 seconds
Brief review: “Background Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-025｜Hairstyle Transfer｜80/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 4/5 · Practicality 4/5 → 80/100
Generation time: 26.97 seconds
Brief review: “Hairstyle Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-026｜Expression Transfer｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 27.44 seconds
Brief review: “Expression Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-027｜Accessory Transfer｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 27.51 seconds
Brief review: “Accessory Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-028｜Color Palette Transfer｜100/100

Rating: Edit fidelity 5/5 · Character preservation 5/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 100/100
Generation time: 27.35 seconds
Brief review: “Color Palette Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-029｜Composition Transfer｜96/100

Rating: Edit fidelity 5/5 · Character preservation 4/5 · Reference adherence 5/5 · Lack of artifacts 5/5 · Practicality 5/5 → 96/100
Generation time: 27.54 seconds
Brief review: “Composition Transfer” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-030｜Lighting Transfer｜88/100

Rating: Edit fidelity 5/5 · Character preservation 4/5 · Reference adherence 4/5 · Lack of artifacts 5/5 · Practicality 4/5 → 88/100
Generation time: 27.61 seconds
Brief review: The composition with multiple characters and moonlight works. However, the identities of the individual characters became less distinct.

ANIME-IT2I-031｜Character, Outfit, and Background｜76/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 3/5 · Lack of artifacts 4/5 · Practicality 4/5 → 76/100
Generation time: 30.7 seconds
Brief review: “Character, Outfit, and Background” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-032｜Character, Pose, and Light｜76/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 3/5 · Lack of artifacts 4/5 · Practicality 4/5 → 76/100
Generation time: 30.55 seconds
Brief review: “Character, Pose, and Light” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-033｜Character, Art Style, and Background｜76/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 3/5 · Lack of artifacts 4/5 · Practicality 4/5 → 76/100
Generation time: 30.66 seconds
Brief review: “Character, Art Style, and Background” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-034｜Character, Hair, and Outfit｜76/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 3/5 · Lack of artifacts 4/5 · Practicality 4/5 → 76/100
Generation time: 30.56 seconds
Brief review: “Character, Hair, and Outfit” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-035｜Mecha, Location, and Light｜76/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 3/5 · Lack of artifacts 4/5 · Practicality 4/5 → 76/100
Generation time: 30.7 seconds
Brief review: “Mecha, Location, and Light” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-036｜Swordsman, Pose, and Snow｜76/100

Rating: Edit fidelity 4/5 · Character preservation 4/5 · Reference adherence 3/5 · Lack of artifacts 4/5 · Practicality 4/5 → 76/100
Generation time: 30.38 seconds
Brief review: “Swordsman, Pose, and Snow” was edited without compromising the anime style. With multiple references, it is good at combining elements, but do not expect perfect separation.

ANIME-IT2I-037｜Magical Girl・Effects・Night｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 30.39 seconds
Brief review: “Magical Girl・Effects・Night” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-038｜Idol・Costume・Stage｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 30.46 seconds
Brief review: “Idol・Costume・Stage” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-039｜Chef・Cooking・Lighting｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 30.36 seconds
Brief review: “Chef・Cooking・Lighting” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-040｜Three People・Background・Composition｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 30.46 seconds
Brief review: “Three People・Background・Composition” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-041｜Person・Costume・Pose・Background｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.77 seconds
Brief review: “Person・Costume・Pose・Background” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-042｜Person・Hair・Glasses・Light｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.44 seconds
Brief review: “Person・Hair・Glasses・Light” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-043｜Mecha・Color Scheme・Location・Weather｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.08 seconds
Brief review: “Mecha・Color Scheme・Location・Weather” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-044｜Two People・Costumes・Background・Composition｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 32.78 seconds
Brief review: “Two People・Costumes・Background・Composition” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-045｜Character・Art Style・Pose・Effects｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 32.82 seconds
Brief review: “Character・Art Style・Pose・Effects” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-046｜Person・Hair・Costume・Pose｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.53 seconds
Brief review: “Person・Hair・Costume・Pose” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-047｜Person・Expression・Accessories・Light｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.3 seconds
Brief review: “Person・Expression・Accessories・Light” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-048｜Mecha・Color Scheme・Composition・Location｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.47 seconds
Brief review: “Mecha・Color Scheme・Composition・Location” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-049｜Three People・Background・Light｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.54 seconds
Brief review: “Three People・Background・Light” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

ANIME-IT2I-050｜Four-Person Group｜76/100

Evaluation: Edit Fidelity 4/5・Character Retention 4/5・Reference Incorporation 3/5・Low Artifact Rate 4/5・Practicality 4/5 → 76/100
Generation time: 33.51 seconds
Brief review: “Four-Person Group” was edited without breaking the anime style. With multiple references, it is good at combining elements, but it is best not to expect complete separation.

10 Images with Three References: Person + OpenPose + Background

POSE-BG-001｜Standing Upright・Nighttime City｜80/100

Evaluation: Subject Retention 4/5・Pose Fidelity 4/5・Background Fidelity 4/5・Low Artifact Rate 4/5・Practicality 4/5 → 80/100
Generation time: 31.49 seconds
Brief review: It was generally able to keep the three roles—subject, upright skeleton, and background—separate. However, facial identity slipped slightly.

POSE-BG-002｜One Arm Raised・Harbor at Sunset｜88/100

Evaluation: Subject Retention 5/5・Pose Fidelity 4/5・Background Fidelity 4/5・Low Artifact Rate 5/5・Practicality 4/5 → 88/100
Generation time: 30.58 seconds
Brief review: The background was incorporated well, and the upright posture was natural. It prioritized subject retention over strict adherence to OpenPose.

POSE-BG-003｜Seated・Watercolor Seaside｜84/100

Evaluation: Subject Retention 4/5・Pose Fidelity 4/5・Background Fidelity 4/5・Low Artifact Rate 5/5・Practicality 4/5 → 84/100
Generation time: 30.78 seconds
Brief review: The subject and background were blended naturally. Adherence to the seated skeleton was moderate.

POSE-BG-004｜Running・Seaside Architecture｜28/100

Evaluation: Subject Retention 1/5・Pose Fidelity 2/5・Background Fidelity 1/5・Low Artifact Rate 2/5・Practicality 1/5 → 28/100
Generation time: 30.79 seconds
Brief review: Skeleton lines appeared over the background. It completely failed to separate the roles of the references.

POSE-BG-005｜Twist・Theater Lights｜28/100

Evaluation: Subject Retention 1/5・Pose Fidelity 2/5・Background Fidelity 1/5・Low Artifact Rate 2/5・Practicality 1/5 → 28/100
Generation time: 30.78 seconds
Brief review: Skeleton lines remained on the photographic background, making it unusable as a finished image.

POSE-BG-006｜Standing Upright・Rainy Station｜92/100

Evaluation: Subject Retention 5/5・Pose Fidelity 4/5・Background Fidelity 4/5・Low Artifact Rate 5/5・Practicality 5/5 → 92/100
Generation time: 30.71 seconds
Brief review: It cleanly integrated the anime character, pose, and station background. A successful example with three references.

POSE-BG-007｜One Arm Raised・Neon City｜96/100

Evaluation: Subject Retention 5/5・Pose Fidelity 4/5・Background Fidelity 5/5・Low Artifact Rate 5/5・Practicality 5/5 → 96/100
Generation time: 30.05 seconds
Brief review: The boxer’s identity and the neon background were retained while also achieving the pose. Very strong.

POSE-BG-008｜Seated・Floating City｜84/100

Evaluation: Subject Retention 4/5・Pose Fidelity 4/5・Background Fidelity 4/5・Low Artifact Rate 5/5・Practicality 4/5 → 84/100
Generation time: 30.58 seconds
Brief review: The background and character look natural. Precision in following the seated instruction is slightly weak.

POSE-BG-009｜Running・Moonlit Bamboo Grove｜28/100

Evaluation: Subject Retention 1/5・Pose Fidelity 2/5・Background Fidelity 1/5・Low Artifact Rate 2/5・Practicality 1/5 → 28/100
Generation time: 30.64 seconds
Brief review: A clear failure, with skeleton lines remaining over the character.

POSE-BG-010｜Twist・Rainy Street｜32/100

Evaluation: Subject Retention 1/5・Pose Fidelity 2/5・Background Fidelity 1/5・Low Artifact Rate 3/5・Practicality 1/5 → 32/100
Generation time: 30.49 seconds
Brief review: The character was retained, but skeleton lines were exposed. Handling of pose-control images is unstable.

Final Conclusion

The safest use case is single-image photorealistic or anime artwork created from short natural-language prompts. It is quite strong with style, lighting, camera angle, and mood. Short titles and logos are also within the practical range.

By contrast, use caution with long-form typography, pixel-level preservation, precise outpainting, and reliable OpenPose adherence. In particular, skeletal references are supplied as ordinary image references rather than “control inputs,” so the lines can sometimes leak through even when instructed to remove them. Looking only at successful images can easily lead to overconfidence.

My personal conclusion: Qwen-Image 2.1 INT8 excels as a “model that produces a finished image in one shot.” But when viewed as a “model that makes precise edits like a design tool,” it still has some rough edges.