Pose & Character Mashup

Image Editing

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

This prompt is challenging because it forces models to do two hard things at once: accurately transfer complex body poses and joint angles from Image 1 while strictly preserving the exact face, identity, and clothing style from Image 2. Most models struggle with unnatural limb distortions, identity blending, incorrect fabric behavior on the new pose, and mismatched lighting, making it a strong test of advanced spatial reasoning and character consistency.

Blind Vote This Challenge

#1 — GPT Image 1 Mini

Edit instruction

“Use Image 1 as the exact pose reference and Image 2 as the character reference. Recreate the person/character from Image 2 in the exact dynamic pose and body position from Image 1. Keep the exact face, hair, clothing style/details, and expression from Image 2. Match the lighting and environment of Image 1. The final image must show the character from Image 2 performing the precise action/pose from Image 1 with perfect anatomy and natural integration.”

Voters were asked to judge by Pose Accuracy Character Consistency Correct Anatomy Lighting Match

Challenge Rankings

5 models

#	Model	Price	¢/img	Elo
1	GPT Image 1 Mini OpenAI	$	0.5¢	1179
2	FLUX.2 [klein] 9B Black Forest Labs	$	0.6¢	1134
3	Seedream 5.0 Lite ByteDance	$$	3.5¢	1100
4	Wan 2.7 Pro Alibaba	$$$	7.5¢	1079
5	FLUX.2 [dev] Turbo fal	$	0.8¢	1051

GPT Image 1 Mini leads the leaderboard with an 1179 Elo and a 66.7% win rate, outperforming the high-tier Wan 2.7 Pro by exactly 100 Elo points despite costing 15x less. Budget-friendly models dominate the top of this challenge, with FLUX.2 [klein] 9B (1134 Elo) providing the fastest competitive results at nearly double the speed of the trailing Seedream 5.0 Lite.