Outfit Transfer Challenge

Image Editing Photorealism

5 models were given the same image and edit instruction, and the community voted blind on which outputs looked best. How it works

This is extremely practical (think e-commerce, fashion, virtual dressing rooms). It’s visually very obvious when it fails, and success looks impressive. It tests a model’s understanding of clothing physics, body shape, and lighting.

Blind Vote This Challenge

#1 — Wan 2.7

Edit instruction

“Use Image 1 as the base person. Dress them in the exact elaborate outfit from Image 2 (including all layers, accessories, jewelry, and shoes). Carefully adapt the clothing to the body shape and pose in Image 1 while maintaining realistic fabric behavior, correct proportions, and perfect lighting/shadow matching. Keep the person’s exact face, hair, and background completely unchanged.”

Voters were asked to judge by Clothing Accuracy Fit & Fabric Realism Lighting & Shadow Consistency No Side Effects

Challenge Rankings

5 models

#	Model	Price	¢/img	Speed	Elo
1	Wan 2.7 Alibaba	$$	3¢		1095
2	GPT Image 1 Mini OpenAI	$	0.5¢		1082
3	Qwen Image Edit Latest Alibaba	$$$	7.5¢	—	1062
4	FLUX.2 [klein] 9B Black Forest Labs	$	0.6¢		1059
5	Grok Imagine Image xAI	$$	2¢		996

Alibaba’s Wan 2.7 leads the leaderboard with an Elo of 1095 and a 62.5% win rate, narrowly outpacing OpenAI's GPT Image 1 Mini. Notably, the budget-friendly GPT model (1082 Elo) delivers a similar 62.5% win rate at one-sixth the price per image compared to Wan 2.7.