Three Image Models Tested Across Five Styles and Three Edits
TMTPost compared MiniMax H3-Image, ByteDance Seedream 5.5, and Alibaba Qwen-Image-3.0-Pro using one scene across five visual styles, followed by three sequential edits. H3-Image and Seedream 5.5 completed the edits while preserving much of the original scene; Qwen-Image-3.0-Pro lost the scene during the first two edits and the test was stopped. H3-Image produced 4K output throughout, while reported generation times were generally fastest for Doubao, followed by Qwen and H3-Image in some tests.
The hands-on comparison offers concrete observations about image-generation consistency, resolution, watermarking, and speed, but the outlet says its single-prompt, three-edit test does not represent overall product capability.
(本文作者为 象先志,钛媒体经授权发布)
文 | 象先志
最近,AI 画图模型又热闹起来了。
11 月 9 日,字节发布 Seedream 5.5;11 月 10 日,MiniMax 发布 H3-Image,还同步开源了模型权重。再加上阿里此前推出的 Qwen-Image-3.0-Pro,几家热门模型刚好凑到了一起。
AI 画图有个比较现实的问题:图生成得好不好,指定的文字有没有错误、风格对不对;为了局部修改和抽卡,结果模型却把整张图重画了,那其实也挺让人头疼的。
所以这次的测试,我们把三家放在一起,分别测试五种绘画风格,然后再拿生成的图片做三轮连续修改,重点观察几家的局部调整能力。
#MiniMax H3-Image#ByteDance Seedream 5.5#Alibaba Qwen-Image-3.0-Pro#MiniMax#ByteDance#Alibaba#MiniMax Code#Doubao#Qwen AI platform