Ideogram 4
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.
View tests →AI Portrait Realism
Compare skin texture, anatomy, lighting, and camera realism.
What to check
- Does the face remain structurally believable at full size?
- Do hands, jewelry, reflections, and clothing survive inspection?
- Does the camera style match the requested device and lighting?
Tests and guides
ComparisonsWhat Portrait Benchmarks Actually Measure: 22,000 Human Ratings Explained
Charts from Rapidata's 35-prompt face benchmark, compared with DreamBench++ identity testing and demographic identity-drift research.
Read article →ComparisonsIdeogram 4 vs Krea 2 vs Grok: 15 First Outputs ComparedFive same-prompt tests for typography, product photography, portraits, motion, and constrained illustration. One first output per model, with no rerolls.
Read article →ComparisonsFive AI Image Models, One Input: 27-Output Controlled Portrait TestFive shared portrait prompts across Seedream 4.5, GPT Image 2 Medium, Grok, Seedream V5 Lite, and V5 Pro—plus one deduplicated Lite-versus-Pro extension row.
Read article →ComparisonsNano Banana 2 vs Seedream vs GPT Image 2 vs Grok: 14-Prompt Portrait ComparisonFourteen adult portrait prompts with side-by-side outputs from Nano Banana 2, Seedream 4.5, GPT Image 2 Medium, Grok, and Seedream V5 Lite.
Read article →ComparisonsBest AI Models for Realistic Mirror Selfies: 15 Same-Input ComparisonsFifteen visible comparisons from 18 tested mirror-selfie prompts: 39 retained outputs and 15 recorded blocks across Nano Banana, Nano Banana 2 Lite, and Nano Banana 2.
Read article →ComparisonsSeedream V5 Pro vs V5 Lite vs 4.5: Same-Input TestSeedream V5 Pro, V5 Lite, and 4.5 compared on five exact prompts with one frozen Mila input, verified prompt hashes, fixed framing, and all 15 outputs retained.
Read article →Models
Krea 2 Medium
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.
View tests →Seedream V5 Pro
Earned 7 of 15 editorial points in the historical Seedream review, including facial-detail and anatomy comparisons. All 15 prompts returned an output.
View tests →Seedream V5 Lite
Earned 4 of 15 editorial points in the historical Seedream review, including clothing detail and scene composition. One prompt returned no output.
View tests →Seedream 4.5
Earned 4 of 15 editorial points in the historical Seedream review. It retained the cafe props and bedroom pose, but added an extra leg in the kneeling portrait.
View tests →GPT Image 2 Medium
Returned all five controlled portrait outputs. In the separate three-prompt moderation test, it returned the studio portrait and blocked two adult-fashion prompts.
View tests →Grok Imagine
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.
View tests →Nano Banana 2
Returned 15 of 18 mirror-selfie prompts, compared with 12 each for Nano Banana and Nano Banana 2 Lite. Completion does not measure visual quality.
View tests →