v14 is VERY good.
Hello,
Just wanted to say V14 is very good compared to everything else since v5.
Good identity retention. No extreme color shifting.
Good iob.
一致性不行
NSFW model-generated images still exhibit grid textures. :(
NSFW model-generated images still exhibit grid textures. :(
Some grid textures are expected and accepted as a compromise. They should be extremely minor and often imperceptible at recommended settings (especially with a few more steps).
Hey Phr00t. Appreciate all the work you've been putting in. It's made image generation a breeze. Thanks for all the work and time you've put into this and the Wan AIO. Cheers.
I AM SO MUCH AGREE!!!! the similarity between the ref image and the output can be said as identical 99.99% good job and MUCH MUCH THANK YOU !!!!!
丁丁直接变洞洞的概率小了, 但是, 丁丁头上少一个本该有的洞洞
NSFW model-generated images still exhibit grid textures. :(
Some grid textures are expected and accepted as a compromise. They should be extremely minor and often imperceptible at recommended settings (especially with a few more steps).
Looks like 14.1 is even better.
It's not going to blow in my 8G.
Hello,
Just wanted to say V14 is very good compared to everything else since v5.
Good identity retention. No extreme color shifting.Good iob.
Strangely this is the worst model for me yet for facial consistency (14.1 nsfw ver).
Using an input image of a blonde woman will give me a blonde woman output, but face is completely unrelated.
It really wants to make Asian women into Caucasians too.
Minor edits seem to maintain the face but placing the person into a new scene isn't possible.