search

tencent-ailab / IP-Adapter

The image prompt adapter is designed to enable a pretrained text-to-image diffusion model to generate images with image prompt.
Apache License 2.0
4.52k stars 298 forks source link

What are the differences between IP Adapter FaceID Plus V2 and IP Adapter FaceID Plus? #215

Open YZBPXX opened 6 months ago

YZBPXX commented 6 months ago

What are the differences between IP Adapter FaceID Plus V2 and IP Adapter FaceID Plus?

xiaohu2015 commented 6 months ago

V2 can adjust the weight of face structure condition

YZBPXX commented 6 months ago

okay. thank you

akk-123 commented 6 months ago

@xiaohu2015 IP Adapter update multi model after paper release，can you update IP-Adapter paper?

xiaohu2015 commented 6 months ago

@xiaohu2015 IP Adapter update multi model after paper release，can you update IP-Adapter paper?

We have no plans to update the paper. but if you has some questions, we can discusse here.

eezywu commented 6 months ago

@xiaohu2015 Hi, I have a question. When training IP-Adapter-FaceID-PlusV2, did you use ip-adapter-faceid_sd15.bin to initialize the MLPProjModel part? If so, do you fixed the MLPProjModel part and only trained the FacePerceiverResampler?

Weifeng-Chen commented 5 months ago

what's the difference when training v2 and v1? could you release the training details of both two models?