Image Generation Is Ruined – Why Did You Do This?

What happened to image generation? This new update is ridiculous. The results are worse, slower, and full of arbitrary limitations that make zero sense for people who create content professionally. The images are bland, less creative, and often miss the mark entirely – and don’t even get me started on the lack of control or consistency.

Wasn’t this supposed to be a pro tool? Because right now it feels like we’re being forced to work with a broken toy instead of a creative engine. For those of us who rely on high-quality, customizable visuals, this change isn’t just annoying – it’s a real problem. We’re wasting time, losing quality, and wondering: why fix what wasn’t broken?

Did anyone actually ask for this? Or is this just another case of “we know better” decisions that ignore real user needs? It’s hard to believe this was meant to improve anything. Please, give us back the flexibility and quality we had before – or at least explain why this happened.

5 Likes

I fully agree! I hit a rate limitation after 3 images! With the plus subscription! Then it told me to wait 21 minutes…

And each image takes about 1-2 minutes to be processed…

And the results are a disaster compared to the previous version.

1 Like

Same here, gpt dalle 3 was much more creative, much better results, much faster, much more intricate and “inteligent”. Now it is just this filter applier that everyone is using and will last for few weeks and nobody will use anymore… and the reall good thing it should be doing was lost

1 Like

Its terrible… how can I go back to what it was before?

It’s still possible to use Dall-E 3 using the Dall-E custom GPT.

I hope this works for you.
If you don’t need the editing features, it’s also possible to generate images via API.

while I find the new image generation better for general use, it does kinda restrict and feel like it’s limiting what the AI can do, as for editing, it kinda never was all that great, but I think it listens more than before, as for quality, well, the quality went up, but the creativity went down as OpenAI made filters and content moderation tighter which kinda sucks. Otherwise, for general use it’s good, for creative usage, it would have been good without the filtering restricting what you can do with the AI, I wish they added a function where you can turn off the “safe mode”, cause in itself, the quality of image creation did upgrade, with much less artifacts appearing in frequency.

agreed. i’ve been trying to create images similar to before but no amount of prompting is helping. the new images are stilted, restrainted, and overall disappointing in composition and color. Whereas before I canceled my midjourney subscription for ChatGPT, I may have to do the opposite now.

You can still use the Dall-E 3 image generation here:

Hope this helps!