I didn’t expect it to take this long for me to get the new text model under control. It’s much more intelligent in general, but the repetition and random fixations can get surprisingly bad.

Sorry about that. This is just a heads up that I’m still working on it, and it’ll likely remain a bumpy ride for the next couple of months.

Thanks to those who have complained here and on reddit, especially the posts that give concrete examples of issues, and attempted workarounds, etc.

Please do keep posting any issues/discoveries, ideally with share links to give me specific examples that I can test.

Other news:

  • I’m still working on the next image gen update, and am hoping it will improve fine details, text rendering, and overall image quality. It should also be about 2x faster. If all goes well (which, from experience, it won’t) I should be able to get that done in the next month or so.

  • For a long time I’ve been very close to shipping a multiplayer plugin and database plugin. I’ve actually made multiple ‘ready to ship’ draft plugins of both, but have been disatisfied with the design and ended up not publishing each time. Using these plugins won’t trigger ads on the generator, so I need to be really careful about the overall architecture to ensure the the servers for them are dirt-cheap to run. I also want to make them fun/simple to use of course. Once I get text and images stable, I’m going to focus on these two plugins and finally get them done.

  • hamsterneo@lemmy.worldEnglish
    25·
    3 months ago

    to EVERYONE: perchance isn’t lying.now image quality fixed very well. but something habit of prompt changed. if you tired of CARTOONS, try copy these to your image prompt and put at the front :

    “photorealistic, realistic,realistic person,realistic style.real photo,realistic photo,real person, photorealistic，hyperrealistic，real human portrait， candid photography，ultra-detailed skin texture，documentary style，natural body proportions， natural lighting.

    then NO MORE CARTOONS. you are welcome.

    • agracier@lemmy.worldEnglish
      1·
      3 months ago

      Tnx for the prompt. I’ve tried a few dozen renders and your ‘prefix’ does help a bit. I’ll try many more times with some permutations and such and see how it works. But still, it’s tricky and not really the same with some cartoonish elements showing through, like too much contrast on skin tones accentuating the muscles and such. And I used perchance for more academic artistic painting styles, not really for photos. Maybe I’ll get those back with some experimenting, but I bet not entirely if the underlying Stable Diffusion is no longer used.

      But in any case, tnx for the tip. I appreciate it.