Anyone facing the problem that this tool is not utilizing the GPU (Graphics card) but running on CPU mode only 1. Open command prompt and navigate to folder where SadTalker is installed 2. type or copy/paste the following and press enter venv\scripts\activate 3. Paste the following command and press enter pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url download.pytorch.org/whl/cu113 Please refer the installation section github.com/OpenTalker/SadTalker
@mahmood392
4 ай бұрын
i did its still utilizing only CPU
@StableAIHub
4 ай бұрын
@@mahmood392 Please log an issue on Github repo github.com/OpenTalker/SadTalker/issues
@Axel-lp1tm
3 ай бұрын
@@StableAIHub is there any solution - yet for it?
@dizzysmit
Ай бұрын
thank you! everything works! finally full interface and possibilities!
@StableAIHub
Ай бұрын
Awesome 👍
@jonathan_henriques
4 күн бұрын
I had several problems. I played in the gpt chat and he told me what to fix. I changed the code, updated libs, changed images and audios that were missing, and in the end it worked!
@MarkWilder68
5 ай бұрын
Ok, still not sure yet but I use one of the default reference videos and it seems to be running. I did not know that it would try to use the audio from the reference video instead of the audio I uploaded. So maybe I don't do ALL, maybe just Blink & Pose since that's all I really want from the reference video. I want post anymore if it works great if not I'll figure something else out. Still thank, you have gotten me further that others.
@StableAIHub
5 ай бұрын
You can use your own audio, image and reference video. It worked for me.
@MarkWilder68
5 ай бұрын
Thank you for trying I do appreciate it...
@StableAIHub
5 ай бұрын
u r welcome.
@user-cr9rf7ln2o
5 ай бұрын
Wow. Works absolutely fine and in 1 go.
@DavidBrown-tv8fx
4 ай бұрын
I don't know the reason. Your version makes the footage blurry for more than ten seconds, while the original version does not have this phenomenon.
@StableAIHub
4 ай бұрын
It is original version only, I only compiled the repository to make it easy for everyone. You can try to clone the repository from Huggingface. huggingface.co/spaces/vinthony/SadTalker 1. Switch to Files tab 2. Click on three dots menu located at top-right. 3. Select Clone repository Any issues following above, let me know.
@MarkWilder68
5 ай бұрын
I need full size images for a project, I will try to see if it works the way you did it with the head shot but full size is what I need. Thanks
@StableAIHub
5 ай бұрын
I am not sure if it support full posture or only face. Will try and let you know.
@rluzentales
5 ай бұрын
Received "ERROR: No matching distribution found for torch==1.12.1+cu113". My torch and cuda are higher versions.
@StableAIHub
5 ай бұрын
Check this link and install accordingly developer.nvidia.com/cuda-downloads
@darbycarpenter3032
Ай бұрын
Just a general question. My SadTalker uses my CPU and it takes a long time. Is there a way to get SadTalker to use the GPU?
@StableAIHub
Ай бұрын
Several users faced this problem without solution. I posted another video related to SadTalker advanced features, try if that approach works. I would suggest post an issue on their GitHub page. You can also refer other tools for Lip Sync and see if that helps.
@MarkWilder68
5 ай бұрын
cv2.error: OpenCV(4.9.0) D:\a\opencv-python\opencv-python\opencv\modules\imgproc\src esize.cpp:4152: error: (-215:Assertion failed) !ssize.empty() in function 'cv::resize'
@StableAIHub
5 ай бұрын
if you can provide the input files you are using, I will try at my end,
@LoreCraftForge
9 күн бұрын
im getting this error ModuleNotFoundError: No module named 'basicsr'
@StableAIHub
9 күн бұрын
activate virtual environment and then pip install basicsr
@LoreCraftForge
9 күн бұрын
@@StableAIHub yeah I finally got it working but the installation is missing bunch of modules, and necessary files for audio, video, image. Got it working but it runs CPU not GPU so 1 video a minute long took 8 hours to render.
@StableAIHub
6 күн бұрын
@@LoreCraftForge Let me create a new tutorial as several users reported it is working on CPU. Please could you activate virtual environment and run the following command and see if torch is installed with CUDA pip list
@user-rt6nk9sc4y
4 ай бұрын
now this tool can make , two hand ,mouse , Natural movements and realistic performances
@StableAIHub
4 ай бұрын
Cool. I'm gonna look at it now.
@stevewarby12
Ай бұрын
Hi. Is it possible to batch process a few images in code ?
@StableAIHub
Ай бұрын
It should be possible. I am busy this week so if you remind me on Saturday I will try and see how it can be done. You can also post in SadTalker github issue section.
@stevewarby12
Ай бұрын
@@StableAIHub Thanks I have posted there as well
@AnjarMoslem
2 ай бұрын
it's the checkpoint on gdrive necessary? or I just need to copy from original sadtalker?
@StableAIHub
2 ай бұрын
You should download from original sadtalker
@AnjarMoslem
2 ай бұрын
@@StableAIHub btw can I use different reference video that not containing same audio?
@StableAIHub
2 ай бұрын
@@AnjarMoslem Yes you can.
@bengtovsky
Ай бұрын
Hey! First of all thank you! I am not an engineer or developer so pls don't judge me 😅 I installed everything like in the tutorial. When I run the webui , in cmd it says: Python not foundCannot find the path specified on the system. venv "venv\Scripts\Python.exe" It cannot find the path specified in the system. Initialization failed. Getting out. Press any key to continue. . . Can you help me pls I need this tool 😭
@StableAIHub
Ай бұрын
It seems you don't have python installed or the virtual environment was not activated. 1. Start > Command > type python and press enter Tell me what it says 2. Share a screenshot of SadTalker_withRefVideo folder. Just want to quickly check what's inside the folder 3. Share screenshot of the command prompt when you are trying to run WebUI
@bengtovsky
Ай бұрын
@@StableAIHub drive.google.com/drive/folders/1PNBKA9hHFHEGo3VaTSSGnV2ehxAw_UFI?usp=drive_link Hey hey thx for replying these are all the necessary screenshots I get an error :( when I run webui
@bengtovsky
Ай бұрын
@@StableAIHub Hi I am trying to reply to your comment to share the screenshots but I think you might've been blocked the comments with links so my previous replies must be in review section. I downloaded python 3.12.4 but I receive an error when I run webui it says: ERROR: Could not find a version that satisfies the requirement torch==1.12.1+cu113 (from versions: 2.2.0, 2.2.1, 2.2.2, 2.3.0, 2.3.1) ERROR: No matching distribution found for torch==1.12.1+cu113
@StableAIHub
Ай бұрын
@@bengtovsky Please uninstall python and install Python 3.10.6. All the tools posted on my channel works well with this version www.python.org/downloads/release/python-3106/ Give me some time. Let me prepare standalone package for you.
@StableAIHub
Ай бұрын
Ok. I have compiled the standalone package (includes everything) and it's working at my end. 1. Uninstall Python and install Python 3.10.6 (check by typing python --version in command prompt) 2. Download the following drive.google.com/file/d/1r3V3PGNYOk9X7WLInRGrS1xZCEf2btrm/view?usp=sharing 3. Scan the downloaded file with Antivirus (should always follow for anything downloaded from internet) 4. Unzip the file 5. Double Click on webui.bat I hope this works. Please let me know.
@user-rt6nk9sc4y
4 ай бұрын
How long time we can make out put video 5 nminutes or 10 or 30 minutes if we want
@StableAIHub
4 ай бұрын
Unfortunately I don't have high end graphics card with lots of VRAM. So can't benchmark the high duration videos.
@user-rt6nk9sc4y
4 ай бұрын
and any tool can make voice clone and train voice , thanks
@StableAIHub
3 ай бұрын
Yeah, working on that. Will post in coming week.
@user-rt6nk9sc4y
3 ай бұрын
@@StableAIHub ok
@DavidBrown-tv8fx
5 ай бұрын
but it's on cpu.....
@StableAIHub
5 ай бұрын
Nope. It's on GPU. It is CUDA enabled.
@Axel-lp1tm
5 ай бұрын
seems not to work with actual forge/ automatic1111 @@StableAIHub
@RaygonSmith
5 ай бұрын
@@StableAIHubMine is also running on CPU. How do I switch to GPU?
@StableAIHub
4 ай бұрын
@@RaygonSmith @DavidBrown-tv8fx try this Open Command prompt and switch to folder where you have installed SadTalker 1. Activate virtual command using venv\scripts\activate 2. pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url download.pytorch.org/whl/cu113
@StableAIHub
4 ай бұрын
try this Open Command prompt and switch to folder where you have installed SadTalker 1. Activate virtual command using venv\scripts\activate 2. pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url download.pytorch.org/whl/cu113
@thays182
Ай бұрын
I'm consistently getting this error.... Any way to fix this? File "A:\SadTalker\SadTalker_withRefVideo\venv\lib\site-packages\gradio\components\base.py", line 190, in hash_file with open(file_path, "rb") as f: FileNotFoundError: [Errno 2] No such file or directory: 'examples/driven_audio/japanese.wav' Launch unsuccessful. Exiting.
@StableAIHub
Ай бұрын
Please check if the file exists by pasting the following in explorer window address bar and press Enter A:/SadTalker/SadTalker_withRefVideo/examples/driven_audio/japanese.wav
@thays182
Ай бұрын
@@StableAIHubthere are a series of missing files. I keep running and getting new missing files. I’m just renaming other files to what it’s expecting but stoped trying after 5-6 missing files.
@StableAIHub
Ай бұрын
@@thays182 If you look at the recent comments it worked. Don't worry the video and installation steps includes the complete package and I compiled it as the installation is complex. I am sure the issue has to do with input files only. If you can provide the image and audio file I can try it for you.
@user-ui9sg2kh9r
3 ай бұрын
Thank for sharing, i got a question, when i try to use reference video, Error arise, how can i solve it? RuntimeError: Unexpected ASGI message 'websocket.close', after sending 'websocket.close'.
@StableAIHub
3 ай бұрын
Are you getting this during the first time use? I have noticed that it works fine the first time, but on subsequent use it comes up with the error you mentioned. If this is the case shutdown the server after each use and start again. Please also post the issue here with the log github.com/OpenTalker/SadTalker/issues
@user-ui9sg2kh9r
3 ай бұрын
thank you for your reply I encountered this problem on my first time use, it don't work for the first time. I suspect maybe the reference video I'm using has the wrong duration/size/resolution? Can you sharing with me the parameters of the reference video you used? Thanks
@StableAIHub
3 ай бұрын
@@user-ui9sg2kh9r You can refer the example folder for all input files used in the demo github.com/newgenai79/SadTalker_withRefVideo/tree/main/examples
Пікірлер: 69