vamosc / caphuman Goto Github PK

Are the reference image and GT image during training from the same image of CelebA?

Hi, may I ask how to specify args.control_ckpt and where to download the ckpt?

Hi, it seems that we should also download all the files related to clip-vit-large-patch14 and place these files under /ckpts/clip-vit-large-patch14, but the preparation instruction doesn't mention it. Could you please revise the preparation instruction?

Hi, may I ask can caphuman generate full body (from head to feet) image with the help of an openpose controlnet?
Another question, it seems that the face of the generated human always looks at front, so can caphuman generate side-view faces? If caphuman can't, is there any way to achieve this requirement?

will you release code?

Hi, may I ask how to use openpose controlnet together with caphuman?

我使用的是你推荐的测试命令：
python inference.py --ckpt ckpts/caphuman.ckpt --vae_ckpt ckpts/vae-ft-mse-840000-ema-pruned.ckpt --model models/cldm_v15.yaml --sd_ckpt ckpts/Realistic_Vision_V3.0.ckpt --input_image examples/input_images/196251.png --pose_image examples/pose_images/pose1.png --prompt "a photo of a man wearing a suit in front of Space Needle"

我的结果：input_image | pose_image | out_img

看上去有两个问题：
1.Input_img的ID并没有保持
2.pose_img的姿态也没有得到体现。
是我哪里用错了吗？

Hi, this is an amazing and elegant work, and it inspired me a lot. May I ask how did you achieve the multi-human generation, that is, how to combine several head conditions and id conditions together then inject into the denoising process? Thank you!