This is an AI image generation comparison for
reference-to-image 
prompt:
This subject presented as a premium collectible figurine on a designer’s desk. Scene: • Center: a realistic 1/6-scale figurine of the subject on a clear round stand, natural museum pose one foot slightly forward. Keep their current outfit, hair, and accessories exactly as in the source. • Right: an upright glossy retail box showing the same subject as box art matching outfit and look, with brand AI Creators Tools and model Gina; clean typography and a small authenticity sticker. • Left: a widesc...
Log in to see full prompt.
Tested: October 11, 2025
Nano Banana knows exactly what you're after here. Perfect.
Tested: October 11, 2025
Realistic, preserves likeness. Main issue is text. Note that only newly introduced text is a problem, model correctly copies the text present on reference image.
Tested: October 11, 2025
So the model struggles with this prompt. The model is realistic, even if you drop in 'toy' in with a 'figurine'. Also completely changes outfit. Gave it 5 attempts. Not one produced same subject on box, screen and figurine.
Tested: October 11, 2025
Qwen definitely knows what's going on here. Successdully executed. Only downside is lack of photorealism.
Tested: October 11, 2025
Great work by Wan 2.5 on the first attempt
Tested: October 11, 2025
Reve's not bad with this concept.
Tested: October 11, 2025
I'd say Runway has potential with this prompt.
Tested: October 12, 2025
Seedream does great job with this prompt.
Tested: October 31, 2025
Text and finer facial features like eyes & nostrils remain challenging for this model in somewhat complex prompts, it seems.
Tested: December 1, 2025
An improvement over Q1 that the box isn't much smaller than the figurine itself. Her facial features aren't too crisp, likely could be helped with an additional reference image with her closeup portrait.
Tested: December 1, 2025
So yes, if you upload an additional clear headshot of your character it improves facial features clarity. Box size is now too small again (because model is huge)
Tested: December 4, 2025
Tested: December 5, 2025
Crisp facial features (except eyes) + accurate text, not to mention knowing the whole concept of this figurine on designer desk scene. Ok, 1 mistake, extra letter 'a' in Creators (out of 8 words, still impressive)
Tested: December 5, 2025
This didnt go so well. Uploaded subject and style. Second time uploaded 2 subject images (HD closeup face and full body - was even worse). It won't allow to simply give a subject reference, need at least 2 images. On FAL, which permits a single image, this prompt still makes cartoonish childlike-dolls
Is the 1/6-scale figurine centered on a clear round stand?
Does the figurine maintain the subject’s exact outfit, hair, and accessories from the source?
Is the pose natural and museum-like, with one foot slightly forward?
Does the figurine’s scale appear consistent with a real collectible (relative to keyboard, monitor, etc.)?
Is the retail box positioned to the right, standing upright and clearly visible?
Does the box art depict the same subject, outfit, and look as the figurine?
Is the brand name “AI Creators Tools” and model name “Gina” clearly printed on the box?
Is there a visible authenticity sticker on the box?
Does the monitor on the left show a grayscale 3D sculpt or turntable of the same subject?
Does the 3D sculpt on-screen match the figurine’s form and pose accurately?
Are desk props (keyboard, mouse pad, notes) neatly arranged and minimal?
Does lighting appear as bright natural daylight with soft, realistic shadows?
Are materials (plastic, box print, desk surface) rendered photorealistically?
Are there no mismatches or duplicates between figure, box, and monitor representations?
Check out the results from Freepik (Gemini 2.5 Flash) vs Vidu AI (Vidu Q1 Image) vs InfiniteYou (ImagineYou) vs Fal AI (Qwen Image Edit) vs Wan (Online Platform) (Wan2.5 Image-Edit) vs Reve (Reve Image 1.0) vs Sora (GPT‑4o) vs Freepik (Runway's Gen-4 Image) vs Freepik (Seedream 4.0) vs Fal AI (ChronoEdit) vs Vidu AI (Vidu Q2 Image) vs Vidu AI (Vidu Q2 Image) vs Pruna AI (P-Image-Edit) vs Fal AI (Seedream 4.5) vs Kling AI (Kling O1 Image) for similar or identical prompts side-by-side.
Cute pet figurine christmas tree ornament
