There's no way you tested this thing properly and determined it was objectively better than F2 Klein 9B
It's just not. It doesn't feel like the next generation model in any way. Why 5.6B? Why a much smaller text encoder than used by F2 Klein 9B? How skewed is your perception of what people expect that you thought this thing would be well-received?
Here's one of many examples I could give as to how this model is like at best a sidegrade from F2K 4B, but with obvious signs of being trained on way too many ChatGPT and Nano Banana gens that none of the versions of F2K really had.
I wrote a long post to this company here 2 months ago, wrote what people would love the next gen model to be. There was no reply.
So they released a TOTAL JOKE model with 5.6 b parameters. It is a joke. I told them 12b+ model is the minimum need. This is model is a total CRAP.
It is dumb, makes anatomy horror. Totally useless junk. This model is like anti-ad. It is shocking who would do this to his own company..
i really like F2K where contact
Here's one of many examples I could give as to how this model is like at best a sidegrade from F2K 4B, but with obvious signs of being trained on way too many ChatGPT and Nano Banana gens that none of the versions of F2K really had.
Could you explain what you expected to be better in your 1-girl?
bro they just look immature and creepy lol while all K2 are proper on the other hand... Even the pose on K3 is not natural
@slightlyoutofphase @MavenDE we've made some changes in our comfy workflows, please try those and let us know if they're closer to what you expected. We're always listening to feedback and want to do best by you, the community :)
- Maanav from BFL
bro they just look immature and creepy lol while all K2 are proper on the other hand... Even the pose on K3 is not natural
Did you specify 'girl' or 'woman' for your prompt?
@BingoBird I was commenting the shared results from @slightlyoutofphase that said your question to them could still be relevant... Anyway, personally I don't even know what I'm doing here I almost never use video or image generative AI... I don't like that. Text generation to me has risks but benefits/risks seem to me acceptable (so far), image an video generation on the other hand, I think the outcome is almost entirely negative... I was just curious when I saw this model pop and wanted to see what the drama was about :D But now, I'm out!