-
In the readme, it is mentioned that changes in FPS require significant code changes. I am trying to train on a custom dataset. Can someone point out what changes do I need to make in order to cater fo…
-
![25_1](https://user-images.githubusercontent.com/37431477/127949247-1c3e6959-e6f1-493b-893c-471273717a87.jpg)
when training wav2lip network, l1 recon loss is 0.02 and syncnet loss is 0.17 , but the …
-
I want to generate high-resolution images(96x96->384x384), but there are always checkboard artifacts on the generated images, and checkboard artifacts are concentrated in the mouth area. As shown in t…
-
I used avspeech data to train syncnet for 285 hours, and the training time dropped from 0.69 to 0.5. Now it has been consistently low to 0.5 for a long time. What is the reason for this. I use syncnet…
-
For a single image inference case (instead of providing mp4 video):
if I use an option (--static True) and provide mp4 video, then the result looks fine.
https://user-images.githubusercontent.co…
-
I am trying to run inference, getting this error :
AttributeError: module 'cv2' has no attribute 'gapi_wip_gst_GStreamerPipeline'
Running on Windows 11, Cuda : 10
-
When I drop the png and wav in the sadtalker tab in stable diffusion ui, and clicked generate, it processed for few seconds and raised an error:
Log:
> /Users/xiao9/Documents/AI/stable-diffusion-w…
-
Hi, If I change the fps to 25 as recommended by wav2lip model, the audio sync also changes.
How do I get the audio sync right again?
The following code is used for changing fps.
"ffmpeg -y -i {…
-
I am on a Mac, SD Web UI running perfectly, SadTalker extension seen and successfully loaded with all models. No matter if I select crop or full, my animation with generate and I’ll watch “Face Render…
-
Hello, I have a question
line 267 in mentioned file
interpolates = alpha * gt + ((1 - alpha) * fake_img)
What is fake_img? It's not defined anywhere, from what variable / func I should get it?