Has Image Generation Been Changed Recently?

0
1
Asked By CreativeSpark42 On

I've been using the image generation feature for a while now, and I've started to feel like it's not performing as well lately, especially when it comes to creating images based on existing references. For instance, I enjoy making comics or cartoons where I direct ChatGPT to depict a character X doing Y in a specific environment Z while matching the style of a reference image. Previously, this worked smoothly, but recently - perhaps since last week - it seems like the generated images come out with a totally different art style, or the characters end up with incorrect proportions or features. Sometimes the images turn out fine at first but then get progressively worse. I've been wondering if there were any updates or changes made that could explain this inconsistency, especially after those adjustments regarding the Ghibli situation from a few weeks back.

5 Answers

Answered By UserFriendly101 On

I don’t think anything has changed drastically. I tried generating some of my previous images, and they turned out exactly how I wanted. Maybe it's more about how you're prompting it lately.

Answered By PixelPonderer38 On

Yeah, I’m noticing issues too, especially on Firefox; it's acting up a lot for me. It seems like the images just don't come out right anymore.

Answered By CuriousCoder77 On

Honestly, it feels like they are draining their compute for something else, maybe for new projects? It's a bit annoying!

Answered By ArtisticFlare99 On

I've had a similar experience while trying to create social event images. They all appear so dull, like mustard instead of vibrant yellow—just not what I expected! I tried giving a ton of inspiration images, but it only brightened things up slightly. Eventually, I had to specify 'outdoor images' to get something that looked even remotely normal and colorful.

Answered By SemanticSkeptic22 On

Honestly, I think it might have more to do with the prompts you're using than any strict new rules. I've been using some similar prompts and variations, but something feels off this week with how it's interpreting them.

Related Questions

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.