The new version of the image generation is so incredibly garbage it's genuinely unusable. Instances of AI artifacting have multiplied exponentially, the system barely follows prompts, characters are jumbled messes of blurry features, it's awful. If it's an issue of expense, I would be willing to donate along with countless others if you took a collection. Please as a community we beg of you to change it back. I would be happy to pay for access to the better model, or a subscription for improved functionality, just anything but this.

you are viewing a single comment's thread
view the rest of the comments
[–] 5 points 8 months ago* (5 children)

I'm really sorry, but the generator is degraded. Images tends towards the cartoonish, anime or manga, really junky imaging that can done on any run-of-the-mill AI generator. Used to be before, you could make wonderful artistic images in an academic, Art Deco, vintage, beaux-Arts, representational art style, like Norman Rockwell covers or pulp fiction covers. Now it is virtually impossible to get that done. The default always pushes back towards cartoonish, childish stuff, big heads, large eyes, narrow waists on people.

This is so sad. Especially with all the months and months of work put into experimenting with and trying out prompts. Now you can get a good working prompt and two days later it produces lousy images. Can't be counted to work again. And all the work, time and effort put into making hundreds of good working prompts is all down the drain. And I surmise that when** yet another upgrade** is introduced, any prompts that work well now, will also end up being useless. Constant unasked for change ruins good things.

Yes, I too would be willing to pay IF the older model were reintroduced, the model prior to Jan 2026.

  • source
  • parent
  • hideshow 5 child comments
  • [+] -2 points 8 months ago (4 children)
  • [–] 3 points 8 months ago (2 children)

    When you get totally different results with the same prompt a day or two later, it is NOT a skill issue. It is an issue with consistency and never ending changes and updates. Enough people have commented on how results always tend towards cartoonish looking images, no matter what the prompt. The generator was magnificent from Aug 2025 to jan 2026. Then the update which was a downgrade and it is like wading through molasses to get anything like the previous results. If you want cartoonish, anime, manga or simplistic results, the new version probably doesn't make much difference. But serious artistic imaging, realistic, academic, beaux arts, art deco, bauhaus, representational art is now out of the picture. A real shame.

  • source
  • parent
  • hideshow 2 child comments
  • [+] -3 points 8 months ago (1 child)
  • [–] 1 point 8 months ago

    I've made some 10s of thousands of non-cartoonish renders since June of 2025, in the traditional art styles that I prefer and that you think of as 'slop'. That was something no other AI platform could do that I know of. The results were wonderful. No cartoons, no anime, no manga if that was not what you were trying to do. And I wasn't nor ever will.

    Cartoon render platforms are everywhere. In the galleries there isn't one image that I have found that has appeal (I stopped looking after a minute or two. They are all the same).

    The question remains the same: used to be that the platform allowed rendering of images that were quality-wise miles above other platforms and in truly serious historically significant artistic styles. Why downgrade the quality? Perchance was unique in that ability.

    I am not alone in this. That should be apparent from other comments.

    And as others have written, if it is a question of finance (as so much is) to get that quality back to how it was, I would be willing to pay. Because it was so unique before this latest version.

  • source
  • parent
  • load more comments (-1 replies)
  • load more comments (-1 replies)