Spaces:
Running
Newbie Here With Silly Question :)
Hi! Just wondering if its normal for the generated images to be so small and highly compressed jpgs, or is there any way we can access the full resolution uncompressed png which I believe is the format dalle-mini works in natively?
Hi, so this model can only do 256x256 images but some other people have tricks like using SwinIR or some diffusion models to make it larger
Hey Boris! Oh, I get it now! It's meant to be truly 'mini' hehe. Thanks. This is the most awesome AI art generator I've been able to use and the results truly mindblow me. Like, 7 out of 10 prompt results are incredible. Kudos to the creators!
You can use this AI to expand the images:
Thanks for the hint! Normally I use Waifu2x ( http://waifu2x.udp.jp/ ) which works great too - but the major problem is that, due to the compression artifacts of the jpg in high compression settings, these same artifacts are enlarged as well, as if they were part of the image as originally intended. It'd be great if it generated clean png's instead.
Thanks for the hint! Normally I use Waifu2x ( http://waifu2x.udp.jp/ ) which works great too - but the major problem is that, due to the compression artifacts of the jpg in high compression settings, these same artifacts are enlarged as well, as if they were part of the image as originally intended. It'd be great if it generated clean png's instead.
Recommend VanceAI Image Upscaler (https://vanceai.com/image-enlarger/) as well as the image enhancer (https://vanceai.com/image-enhancer/). I love the results! Register an account and they will give you 3 free credits every month. You won鈥檛 lose anything by giving it a try lol