4 ms·
I think Adobe is also very careful with copyrighted content not being a part of their models, which inherently makes them of lower quality.
by dvt 1y ago
I think Adobe is also very careful with copyrighted content not being a part of their models, which inherently makes them of lower quality.
- herval 1y agoThey have a much better and cleaner dataset than Stable Diffusion & others, so I’d expect it to be better with some kinds of images (photos in particular)
- doctorpangloss 1y agoas long as you don't consider the part of the model which understands text as part of the model, and as long as you don't consider copyrighted text content copyrighted :)