Each profile stays connected to visible outputs, test conditions, limitations, and blind preference data where available.
Rendered every requested string in the poster and product prompts. Its first outputs were clean and legible, though the poster moved away from the requested condensed type and the product shot omitted the reflective pool.
Matched the largest share of AI Photo's five briefs, including the serum layout, direct-flash portrait style, and directional motion snapshot. It still failed the strict risograph object test.
The most reliable finisher in AI Photo's 30-row Seedream portrait review, with particularly strong close-face detail and anatomy. It did not win every visual comparison.
A strong portrait option when composition and candid camera feel matter. In AI Photo's broad saved-output review, Lite won more direct editorial calls than Pro despite one missing output.
Still competitive on prompt composition and scene completeness. It remains useful as a baseline because newer Seedream routes do not uniformly replace its strengths.
A capable general image route with strong instruction following in AI Photo's retained comparisons. Its visible behavior should be judged task by task rather than from a single overall score.
Produced the strongest poster layout in AI Photo's five-prompt first-output test and clean anatomy in the diner portrait. It missed the exact clock time and crate count in the constrained illustration.
A strong reference-combination and editing route with useful evidence in AI Photo's mirror-selfie benchmark. Its task-level completion and moderation behavior remain visible row by row.
Compare skin texture, anatomy, camera logic, lighting, and the difference between a polished image and a believable photograph.
Open task guide →Workflow guideCharacter ConsistencyEvaluate whether the same authorized person or fictional character remains recognizable across changes in pose, wardrobe, crop, and environment.
Open task guide →Task guideAI Image EditingJudge edits by what changed, what stayed fixed, and how many attempts were required—not by whether the final image looks attractive in isolation.
Open task guide →Platform guideAI Companion ImagesSeparate the quality of generated media from chat quality, memory, character control, privacy, and the cost of getting a usable image.
Open task guide →Evidence guideAI Image ModerationCompare documented outcomes from the same permitted adult-fashion prompts without extrapolating a platform-wide policy from one request.
Open task guide →