andrescasado / pergamo Goto Github PK

Hello, After I successfully installed Expose、pifuHD、Self-Correction-Human-Parsing on ubuntu18.04(windows 10 before) and ran their own demo，I still have some issues. Could you please help me?
I get photos from my own video by ffmpeg and try to get smpl params by using Expose. The error is "Key Error: 'poses' is not a file in the archive". I find that the *.npz don't have the "pose" key(There are things like 'body_pose.npy' 'full_pose.npy' in the *.npz). So it seems that we need to use openpose with Expose and pifuHD?

Thanks a lot!

Sorry I am newbie.

I have finished Install instructions. But I have some problems about Running the project now.
I want to input my own video to drive the SMPL model. It seems that we need to convert the video(mp4) to amass scripts(motion sequences)， but I don't find the mothods or third-party(like alphapose) to do the thing. I find that there are many sequences in dataset, so Could you tell me how to convert the video(mp4) to the certain motion sequences in this project?

Thank you!

Hello, @AndresCasado
The method you guided me through in Blender didn't combine the clothes with the body model. What exactly is the role of _body.pc2? Currently, my visualization result only shows the clothes. I'm a bit lost on how to solve this issue, so I'd like to ask for your advice. Also, regarding the animation effect of the clothes, are they supposed to be colorless?

Looking forward to your response.

When I try to download the data, I am prompted with this problem, making it impossible to log in

Hello, the code is not complete, running can not run the project, can you send a detailed operation, thank you!

The links of datasets and model weights are expired.

when run python predict_amass_sequences.py
OSError: dlopen: cannot load any more object with static TLS

Hello @AndresCasado ：
I am trying to recreate the results using DatosDanCompressed.zip.
However, when I run the "run_recons.sh" file, it executes successfully but yields no results. I suspect that there might be an error with my path settings. Despite repeatedly modifying it, I still cannot get any results. I am very much looking forward to your reply.


I am replicating the code in a Windows environment and have installed Git for Windows.I am very much looking forward to your reply.

The environment setup as it's explained in the repo no longer works, because Chumpy is not up to date with numpy.

Temporal solution is to install an old version of numpy (with pip install numpy==1.21.2 for example, which I confirm works) and recompile/reinstall kaolin (go to its folder and rerun python setup.py develop)

I have downloaded DataDanBrown and ran the reconstruction_scipy.py. Then I got many events.out.tf.events.xxxxx.xxxx.xxx. Soory, But I don't know how to use it.

Which test did you use for the amass dataset, or is this-project predictable for all of amass?

Do you have the source code for the article "ULNeF: Untangled Layered Neural Fields for Mix-and-Match Virtual Try-On", I'm more interested in running it.

Hello, @AndresCasado
The command "python reconstruction_script.py --dir dataset\DatosDan\sequences" executes this Python script. However, no .pkl files are generated in the output, which are needed as input for the subsequent regressor.

Additionally, the code mentions the generation of rendering results, which are also not produced. How can this issue be resolved?

Looking forward to your response.

Hi, @AndresCasado
Which AMASS_.npz should be placed into the test_sequence folder?

Looking forward to your response.

Hello @AndresCasado :
I'm sorry for bothering you again with another question.
After several days of debugging, the code can now run preliminarily. However, I'm a bit unclear about which files correspond to the train and test processes. If I want to train, what sequence of files should I execute? If I want to test, what sequence of files should I execute? My current understanding is that during training, I should first run run_recons.sh, then train_regressor.py. During testing, I should first run run_recons.sh, then run_regression.sh. I'm not sure if my understanding is correct.
When I run the file train_regressor.py, it throws the following error:

Does the dataset originally included in the project, as shown in the image, still needed?

I'm eagerly awaiting your response. Thank you so much in advance.

Hello
I am trying to recreate the results using DataDanGreen folder

I got the following error when running run_recons.sh

RuntimeError:  Folder "smpl" does not exist for sequence "clips". Full path tested: "data/DataDanGreen/clips/clips_smpl".
Uncomment one of the commands and edit the path

Was hoping to get some help resolving this

Thank you