… to see if these current generators outputs can hold the Dall•e style?
This is not possible because DALL-E 3 is a diffusion model - not the same tech. The image below could never be made with 1.5:
Well then I got my answer to my thinking…and that’s a pity that, it wouldn’t work.
@jeffvpace can I “borrow” that image and test something?
Yes, I tried it, but I was not that happy with the result. I also thought about fine-tuning on an open system, but I lack the resources for that. Maybe i try to make a LoRA (Low-rank adaptation) once with some of my images.
It would be cool if the old systems were made open source (I think the company is called OpenAI?? 
, but my hardware is anyway to week for DallE3). But things are moving forward, I see progress everywhere, they are slowly learning to improve the models, and there are now even small ones that deliver better results than DALL·E 2. They have not reached DALL·E 3 yet, not yet…
The new architecture is different, “natively multimodal language models” (autoregressive + diffusion). It is therefore much less creative, but more precise. But I mainly make fantasy images, so the change has not helped me at all.
Sure, you can use any of my images for testing - no need to ask.
Oh thank god you have tried that..so that my idea wasn’t totally off. And yeah if your style is more fantasy, then i totally get you! Let’s seriously hope that this new one would work for you better
You could if you’re interested in check out this: GitHub - PRITHIVSAKTHIUR/Qwen-Image-Edit-2511-LoRAs-Fast-Lazy-Load: Demonstration for the Qwen-Image-Edit-2511 model with lazy-loaded LoRA adapters for advanced single- and multi-image editing. Supports 7+ specialized LoRAs including photo-to-anime, multi-angle camera control, pose transfer (Any-Pose), upscaling, style transfer, light migration, and manga tone. Features fast inference (4 steps default). · GitHub
@jeffvpace so kind. I just always want to ask and not take anyone’s creations.
@Daller i don’t know why that link looks like that
@LarisaHaster Links beauty Don’t matter.
Maybe the wrong place to make commercials for this, but, they not fix the bugs, so, …
You find pages with many LoRA’s. And this is something you can do as a private on a normal GPU.
There will be some when a fan boy witch but one for DallE3. (But it is not the same like a full model, LoRA’s are limited.)
Forgot to mention: You can edit DALL-E images with 1.5 to make changes and still maintain the DALL-E vibe.
@Daller haha i guess links beauty doesn’t matter, but it just disturbed me
@jeffvpace Yes…that was exactly what I was wondering
…maybe I wasn’t clear enough…being bilingual and probably undiagnosed dyslexic
OK the new one is now out.
(just reading now…)
(me diagnosed…)
Hah yeah like I said that we didn’t need to wait too long