Fastest growing GitHub | deep fake live | 3d assets from images | check description | how to
Вставка
- Опубліковано 18 вер 2024
- Hi everyone just having fun with this one, subscribe and like to help me show case stuff to others please.
########################################
Deep fake live
SETUP LIKE THIS TO HELP PREVENT FAILURES
Use microsoft store to download python 3.10
create a folder with called deep fake live
inside the folder
python3.10 -m venv venv
windows .\venv\Scripts\activate
mac source venv/bin/activate
then
git clone github.com/hac...
cd Deep-Live-Cam
then
pip install -r requirements.txt
then
pip uninstall onnxruntime onnxruntime-gpu
then
pip install onnxruntime-gpu==1.16.3
then you are ready
python run.py --execution-provider cuda
########################################
Comfyui -- remember take a picture someone else created and plop it in your comfyui window.
remember to update comfyui with the folder first before trying and failing
READ THIS CAREFULLY
comfyanonymous...
Save one of the image or all them and check them out.
MUST REMEMBER TO UPDATE COMFYUI
########################################
Finally put your images in www.tripo3d.ai...
this will turn them into 3d objects.
YOU CAN REFINE THEM IN THE APP
I use MSPAINT TO REMOVE THE BACKGROUND
I would love feedback a subscribe or like.
Have a nice day
Thanks for a good demonstration of results.
Wow thank you very much, I try so hard on all these videos and I feel UA-cam does not share my videos. Let me know if there is anything else you are interested in.
what video card are you using? I saw another demo that was pretty laggy
I am using a 3080 ti. The basically turned off as many things as possible to make sure it ran alright
Does the image resolution influence the results you'll get?
It crops the image size down but yeah I found the quality of the image helps but it is also not able currently to vastly change your face. For example the Santa image did not work well for me.
How you improve the video quality a little?
I think that is a great question and something I am going to look into. There is a trade off between latency and speed for the amount of vram I have. I use a 3080 ti so only have 12gb of vram. I will be testing on an online gpu soon and see if I get a better quality with more ram. But for starters I would say making yourself look more like the person in the image would help, having a higher quality reference image would also help.
Are u making it live? Please make full tutorial video step by step
Yes it is live Ganesh, in the description is a line by line of how to do it and this video covers the fix, ua-cam.com/video/5jZJnNE1kA0/v-deo.htmlsi=3uDwsp21jxNJ4yLc
Thank you for watching the channel
4070 or 3090 which is better@@eointolster
I prefer 3090 for the more vram, it has more capabilities with llms
Do you know how to use a VastAI GPU for this? I have no experience whatsoever in renting GPUs haha
Hello Hugo S, sorry I don’t have any experience in VastAI. Since the software is meant to be run locally I think there would be much more of a delay. I think this runs on CPU aswell as GPU. If you follow the information in the description and try it I would like to see your results
@@eointolster Hey there, I did get the software up and running with your tutorial, but I haven't tried using comfyUI or the Virtual (venv?). Not sure if those might be crucial for better performance but I'm using a 3060 Ti at the moment which is not handling this program so well unsuprisingly. When I try generating a 10 second clip the program stops responding for a very long time, so I haven't managed to make anything yet... :/
I do believe I saw the creator of this program said he used a rented A10 for his videos showcasing it so I think it should be possible 🧐
@@hugos3947 I’m sorry you have had issues with setting it up. Thank you for letting me know it can be done through online gpu. I hadn’t even thought about it as I had it working with what I have. I try to stay local but I know it is not an option even for me most the time. If I get the time I’ll try and learn how to do it on the cloud. Just never really occurred to me. Thank you for your feedback and message, means a lot
@@eointolster Wow this means a ton man! I just don't want you to go too out of your way to help a stranger on UA-cam 😅 I subscribed to your channel as a thanks for your response!
@@hugos3947 I’m not sure when I’ll get around to it but it sounds interesting and I do want to get
More familiar with online Gpu usage. So it is me who should thank you for a good enough reason to try it. If I succeed I’ll post it. I’ll try over the weekend or early next week. I make my videos for me mostly as not many people have watched before, they are like my notes if I want to make something again in the future, so it’s a good idea. I tend to post how to do stuff in my comments. Thank you for subbing Hugo
can you explained fully tutorials how to get deep fake face like that?
Hi man, I put detailed instructions in my details. Line by line of what I have done. Was that not helpful?
Let me know if you get stuck
whats comfy function in deepfake?
I think you want the inswapper onnx but comfyui does not do it realtime I believe