cann

[theericma]

cann

[sunshineatnoon]

@theEricMa Hi, thanks for the update. Could you please share the pre-processed data through google drive or zip them in the Baidu Netdisk? It is hard to download a raw folder from Baidu Netdisk. Thanks!

[theEricMa]

Hi, I uploaded the missing file together with the dataset. Please see the updated Readme for detail. For any other questions, please feel free to update your question here.

[liangshuang1993]

@theEricMa still missing hdtf_cross_id.py

[sunshineatnoon]

By the way, it seems the Google drive link for processed data is not accessible.

[baihei-1122]

I have Download the dataset from Baidu Netdisk, but how to convert the mdb files to images on Ubuntu18.04, can you share the script to do that.

[theEricMa]

@liangshuang1993 Please check the latest update. I have uploaded the hdtf_cross_id.py.

[theEricMa]

@baihei-1122 If you really want to extract the image data from the .mdb file, my advice is to traverse the .mdb file by modifiying the dataset object

OTAvatar/data/dataset.py

Line 24 in 6e468be

class HDTFDataset(Dataset):

and save every image using cv2 or PIL.Image.

[baihei-1122]

@theEricMa much thanks!

[baihei-1122]

@theEricMa When I test the Face Animation, I get the error with no such checkpoint path 'pretrained/ffhqrebalanced512-64.pth', I just have ffhqrebalanced512-64.pkl file, so I change the pth to pkl, but also get the error RuntimeError: Invalid magic number; corrupt file?, so have any ways to make the inference success?

[theEricMa]

@baihei-1122 please check my updated readme on how to abtain the .pth file

[baihei-1122]

@theEricMa , thanks, I have found this script and convert to .pth file and I adjust the G_new.neural_rendering_resolution =64, through the output video it seems success. But I meet the new problem with miss this file pretrained/ws_std.npy which in trainers/decouple_by_invert.py

[87003697]

@baihei-1122 I uploaded it, you can pull the latest repo

[baihei-1122]

@theEricMa thanks bro! But I meet the new error, it seems like miss the checkpoint.

We train Generator
No checkpoint found at iteration 2000.

through the message I found it located at inference_refine_1D_cam.py - trainer.load_checkpoint

[amerssun]

@theEricMa thanks bro! But I meet the new error, it seems like miss the checkpoint.

We train Generator
No checkpoint found at iteration 2000.

through the message I found it located at inference_refine_1D_cam.py - trainer.load_checkpoint

hi, Have you trained this model？ I'm curious about how well this model works

[theEricMa]

@theEricMa Hi, thanks for the update. Could you please share the pre-processed data through google drive or zip them in the Baidu Netdisk? It is hard to download a raw folder from Baidu Netdisk. Thanks!

We have uploaded the zipped file in Google Drive, thanks for your advice!

[theEricMa]

@theEricMa thanks bro! But I meet the new error, it seems like miss the checkpoint.

We train Generator
No checkpoint found at iteration 2000.

through the message I found it located at inference_refine_1D_cam.py - trainer.load_checkpoint

hi, Have you trained this model？ I'm curious about how well this model works

The model is uploaded now. Please check the Readme, thanks!