News for hackers. Stuff that matters.

A comment on Qwen-Image: Crafting with Native Text Rendering. Parent comment

Re:Qwen-Image: Crafting with Native Text Rendering by vunderba on Monday August 04, 2025 @07:19PM (#44792465)
I've been playing around with it for the past hour. It's really good but from my preliminary testing it definitely falls short of gpt-image-1 (or even Imagen 3/4) where reasonably complex strict prompt adherence is concerned. Scored around ~50% where gpt-image-1 scored ~75%. Couldn't handle the maze, Schrödinger's equation, etc.

https://genai-showdown.specr.net

[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by supermatt on Monday August 04, 2025 @11:19PM (#44793983)
Is it fair to call the OpenAI octopus “real”?
[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by vunderba on Monday August 04, 2025 @11:55PM (#44794184)
Those are the wrong images - the CDN was caching older media - I've since purged it so the right ones should show up now. Thanks for the call out!
[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by supermatt on Tuesday August 05, 2025 @08:45AM (#44797329)
No problem. I am seeing the correct image now.
[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by bavell on Tuesday August 05, 2025 @12:06AM (#44794238)
Fantastic comparisons! Great to see the limitations of the latest models.
[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by xarope on Tuesday August 05, 2025 @01:12AM (#44794534)
interesting to see how many still can't handle the nine pointed star correctly
[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by nilsherzig on Tuesday August 05, 2025 @02:49AM (#44795013)
Great work, thanks.

Midjourneys images are the only ones which don’t make me uncomfortable (most of the time), hopefully they can fix their prompt adherence.

[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by jgalt212 on Tuesday August 05, 2025 @07:48AM (#44796905)
I too have have been underwhelmed by these pelican on a bicycle creators. We used to use them for company blog art, by lately we've switched to using attributed images from Wikimedia commons.
[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by cubefox on Tuesday August 05, 2025 @03:14AM (#44795157)
Prompt idea: "A person holding a wooden Penrose triangle." Only GPT-4o image generation is able to make Penrose triangles, as far as I can tell.
[ Parent | on HN ]
Re:Qwen-Image: Crafting with Native Text Rendering by imcritic on Tuesday August 05, 2025 @08:30AM (#44797206)
Thanks for that comparison/overview/benchmark!

However, you have mistakenly marked some answers as correct ones in the octopus prompt: only 1 generated image has octopus have sock puppets on all of its tentacles. And you marked that one image as an incorrect one due to sock looking more like gloves.

[ Parent | on HN ]