Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

generate wrong pictures #25

Open
sunkymepro opened this issue May 16, 2021 · 13 comments
Open

generate wrong pictures #25

sunkymepro opened this issue May 16, 2021 · 13 comments

Comments

@sunkymepro
Copy link

Dear author,thanks for your great work,it‘s very helpful for me
But when I run the generate_images.py,I get very strange result
019_01
I download the pretrain .pkl file,and set --model=’network-snapshot-020126.pkl’ and =network-snapshot-020126.pkl
The former result is added,the latter result is a blank picture.
I think the reason of this phenomenon is not loading the pkl files fail or loading wrong file.
Is there something needs to notice I miss?I really want to know the answer.

@YuDeng
Copy link
Contributor

YuDeng commented May 18, 2021

Hi, can you try not use --model argument and run the generate_images.py using the default setting? The code should download the pretrained model automatically.

@grinawiem
Copy link

Dear author, I'm studying your code about remaking the face with a new method. It's an amazing work ( DisentangledFaceGAN) in the field of information technology. Thank you for this wonderful and innovative effort. And here I am trying to study and understand your new strategy so that I can mention your work in my research that I am preparing, but when I tried to test it, it became clear to me that I had failed in creating the data correctly. Could you please help me and explain to me more about how to prepare it. Because when I try to teach the model, this result appears on the computer screen.
Please help me;
Screenshot from 2021-08-13 13-13-35
Screenshot from 2021-08-13 13-13-20

@YuDeng
Copy link
Contributor

YuDeng commented Aug 17, 2021

Dear author, I'm studying your code about remaking the face with a new method. It's an amazing work ( DisentangledFaceGAN) in the field of information technology. Thank you for this wonderful and innovative effort. And here I am trying to study and understand your new strategy so that I can mention your work in my research that I am preparing, but when I tried to test it, it became clear to me that I had failed in creating the data correctly. Could you please help me and explain to me more about how to prepare it. Because when I try to teach the model, this result appears on the computer screen.
Please help me;
Screenshot from 2021-08-13 13-13-35
Screenshot from 2021-08-13 13-13-20

Hi, it is strange to get NAN loss when training the VAE. What is your training data and what is the scale of your dataset? If there is too few data, it might fail to train a proper VAE.

@Yitian-Li
Copy link

Dear author, I'm studying your code about remaking the face with a new method. It's an amazing work ( DisentangledFaceGAN) in the field of information technology. Thank you for this wonderful and innovative effort. And here I am trying to study and understand your new strategy so that I can mention your work in my research that I am preparing, but when I tried to test it, it became clear to me that I had failed in creating the data correctly. Could you please help me and explain to me more about how to prepare it. Because when I try to teach the model, this result appears on the computer screen. Please help me; Screenshot from 2021-08-13 13-13-35 Screenshot from 2021-08-13 13-13-20

you should check your data path because you are using a dataset of size "0"

@chinmayjog13
Copy link

@YuDeng I am facing the same issue when I provide the model argument. If I don't, I face the issue referenced in #33 . I am using python 3.6 and the correct versions of packages given in the readme. Please suggest a solution

@Yitian-Li
Copy link

Yitian-Li commented Oct 16, 2023 via email

@fak111
Copy link

fak111 commented May 9, 2024

image
i get wrong picture too

@Yitian-Li
Copy link

Yitian-Li commented May 9, 2024 via email

@fak111
Copy link

fak111 commented May 10, 2024

i hava got the answer ,because of the environment. if u choose the another machine , do the again, maybe solove

@fak111
Copy link

fak111 commented May 10, 2024

i hava got the answer ,because of the environment. if u choose the another machine , do the again, maybe solove

or because the computer has enough size of gpu ,,,model can't do well in weak computer

@aurelianocyp
Copy link

我在3090跑的结果和你一样,不太对。换成了 2080 Ti就成功了

@jadechip
Copy link

jadechip commented Jun 9, 2024

Has anyone a solution to this? I working in an environment based off of the
tensorflow/tensorflow:1.13.2-gpu-py3-jupyter docker image, similar to the recommended specs and this is how my images look: 000_00

I have tried using multiple different GPUs.

Thank you.

@Yitian-Li
Copy link

Yitian-Li commented Jun 9, 2024 via email

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

8 participants