Donate me at Kofi: ko-fi.com/tongtong Paypal: tongtong83@gmail.com --------------------------------------------------------------------------------------------------------------------------------------------------------------- Update 29 May 2024: I tested roop-unleashed on 3 browsers, I personally use Opera by default which has no problems. Google Chrome: cpu runs at 100% at idle, ❌ recommended Opera: run normally, ✔ recommended Microsoft Edge: run normally, ✔ recommended To ensure you are running directml 1. Go to settings tab on top 2. Choose dml in Provider 3. Click Apply Settings, then Restart Server Preview: ru-vid.comUgkxh2Y3WsTbj9jnVxbmLuNY5DRMtC_NROvU Update 24 May 2024: At 1:39 in anaconda Advanced Installation Options, there are 4 options need to check. The second option "Add Miniconda3 to my PATH environment variable" is the most important. If you did not do so, it will lead to a series of problems like 1. your command prompt won't recognize conda command 2. your bat file won't run since it cannot recognize conda command If you do not want to reinstall miniconda to check all the options, your only solution is (that I can think of): 1. use anaconda prompt 2. cd to your roop unleashed installed location 3. conda activate ru 4. python run.py And yes, you have to type this every single time you want to use it. --------------------------------------------------------------------------------------------------------------------------------------------------------------- Video: By cottonbro studio: www.pexels.com/video/hands-girl-standing-safe-3959704/ Photo: By Bernard Lee: www.pexels.com/photo/woman-holding-blue-petaled-flower-1073567/ Mona Lisa: en.wikipedia.org/wiki/File:Mona_Lisa,_by_Leonardo_da_Vinci,_from_C2RMF_retouched.jpg Download Visual Studio 2022 Community: visualstudio.microsoft.com/downloads/ Download FFMPEG: ffmpeg.org/download.html Add System Environment Variables (your path may be different depends on where you put it): D:\PATH\ffmpeg\bin Installing Git git-scm.com/download/win Installing miniconda: docs.anaconda.com/free/miniconda/ Setup Roop Unleashed: conda create -n ru python=3.9 -y conda activate ru git clone github.com/C0untFloyd/roop-unleashed && cd roop-unleashed pip install -r requirements.txt && pip install onnxruntime-directml==1.15.1 run.bat: call conda activate ru python run.py --------------------------------------------------------------------------------------------------------------------------------------------------------------- Chapters 0:00 - Intro 0:09 - Setup Visual Studio 2022 Community 0:20 - Setup FFMPEG 1:08 - Update Your GPU Driver 1:25 - Setup Git and Miniconda 1:48 - Setup roop-unleashed 2:40 - Face Swap Demo I'm going to show you how to install Roop Unleashed on Windows for AMD GPUs. Before you begin, you need to have a few software installed. Download the Visual Studio 2022 Community and make sure you select Desktop Development with C++ during the installation. Download the FFMPEG as shown and make sure the version you download is master latest win64 GPL. Extract the FFMPEG in a folder of your choice. I extract mine at D:\PATH\ffmpeg. Go to the bin folder and copy the location path from the address bar. Type "environment variable" in the search bar on your computer's taskbar. Then, click on "Edit the system environment variables." A window will pop up. Click on "Environment Variables" again. Look for "Path" under System variables and double-click it. Then, click "New" and add the path for FFMPEG. Always make sure you update your GPU driver to the latest version too. I have seen too many people having problems with unknown weird errors because they never update their gpu drivers or are using Microsoft auto-detect drivers that are really out of date. Finally, install git and miniconda as seen. At the Advanced Installation Options for miniconda, check all the options available and click install. Choose a folder you want to install your Roop Unleashed. At the address bar, type CMD and press enter. Paste the conda create code followed by conda activate code. Next, paste the git clone code. After that, paste pip install code to install all the necessary dependencies for Roop. You can close the command prompt after it's done. Open a notepad, paste the following code and save in roop-unleashed folder as run.bat file. Double-click the run.bat to run Roop Unleashed. It will download all the files needed. After it's done, you will be presented with its web UI. It will download buffalo l zip file for the first time when you upload an image to source. As you can see, I'm using Mona Lisa face as the source file and another picture as target file. This will change the face in target file to the one in source file. Subscribe or donate if you like my video.
bro you are creating good contents on AI but you aren't gain more views. I suggest you to add #AI #Stable Diffusion #Face swap. You should add these kind of tags to your video titles so your videos will be more suggested when searched for relevant tutorials.
Can roop unleashed & rope pearl be used on amd gpu? If yes then which amd gpu should i buy & what are the minimum requirements for amd gpu to run these ai applications? Waiting 4 ur reply..😊
Good question, because I make this tutorial to install roop-unleashed for amd gpu, nothing more. How roop-unleashed works on programming level is beyond this scope.
@@tongtongrizhi What is the software level for? I'm talking about the fact that this program works exclusively on nvidia technology and on the processor. Where and how does the AMD video card take part in the acceleration? I have a rx-580 that does not participate, this is easily seen by the speed of the program and in the Task Manager - only the CPU works.
@@jobanarusnya_mustdie The onnx****-directml seen flashing by in the video should make it work on any directml-capable GPU. The rx580 is very old, though, so may not be supported. Have you been successful in enabling directml for any other application? Maybe your drivers need updating?
roop ejecuta correctamente pero al momento del render solo la CPU sigue haciendo el trabajo y no asi la GPU, he monitoreado desde el administrador de tareas y la CPU trabaja al 100% y la GPU al 2%, en "settings" >> "providers" me asegure que estuviese como DML, ayuda. gracias.
In my testing, Google Chrome will use 100% cpu even at idle, not sure if this is a isolate case or common case of everyone. If this happen to you, I suggest you try it at different browser such as opera and microsoft edge. Since you already set the provider to dml in setting, this is how it works at current version in roop unleashed
the conda code is working on CMD but when I paste the git clone command it says "git is not recognized as in internal or external command". both conda and git are downloaded any idea what might be wrong? thank you
Try restart your pc and see if it fixes the problem. If it doesn't, type environment variable at the search bar, go to Environment Variables > System variables > Path and check if it has "C:\Program Files\Git\cmd" If your Path doesn't have this, add it yourself by clicking "New" and press OK
keep getting the error of Error OpenCV(4.9.0) D:\a\opencv-python\opencv-python\opencv\modules\core\src\arithm.cpp:674: error: (-5:Bad argument) When the input arrays in add/subtract/multiply/divide functions have different types, the output array type must be explicitly specified in function 'cv::arithm_op' how to solve it?
at my first installation I can make video output size is same with original video. now second installation my video output suddenly became very huge from original 14MB to output 600MB. what do i have to do?
I wish to say it is impossible to get the output video same as the source, each frame is re-render and enhanced so most of the time, the output video will always be bigger than the original video. If you want to keep the file small, then do not use any post-processing, the file will be small but the face will not be as clear. Btw, why you install it twice?
That hapend 'cause roop_unleashed use --crf 14 by default ("Constant Rate Factor". Means Bitrate. 23-24 is for "normal" quality, but that's even give bigger file size than your imput video file size)
Getting error when trying to swap faces in image swap. 'Error parsing message' And When I try to click on the 'Stop' button after the error, then another error message appears : 'Error not enough values to unpack (expected 2, got 0)' Any solution Please?
This stuff is super advanced for me. I got stuck after entering cmd and trie to copy the command in the description and it just says "unable to create process using...." PLEASE HELP!
I did that and downloaded it the way you showed it, now my problem is the source face photos doesnt process and nothing happens, before when I processed more than 1 picture my whole pc crashed and shutdown
My drivers are all updated and I didnt do anything different then your tutorial and even in another tutorial I reinstalled it 3x now, it worked before and just started crashing out of nowhere, my pc has quite good specs so I doubt its bc of them
When I run 'conda create -n ru python=3.9 -y', the following message appears 'conda' is not recognized as an internal command or external, an operable program or a batch file. Can you help me?
@@tongtongrizhiDoing everything within the anaconda prompt worked. Including to run the ''run.bat'' file. Thank you very much! It's just very slow, I believe it's because of my video card, RX 580.
when i try to swap videos Error in execution: Non-zero status code returned while running Pad node. Name:'/ecd1/ecd1.0/ecd1.0.0/Pad' Status Message: D:\a\_work\1\s\onnxruntime\core\framework\bfc_arena.cc:376 onnxruntime::BFCArena::AllocateRawInternal Failed to allocate memory for requested buffer of size 67108864 pls tell me what should I do?
@@tongtongrizhi it's around 15,20,25 seconds video , my laptop has 16GB ram with 1TB SSD, 3050ti,i7 11 gen, I did video swaps 4-5 months ago on kaggle notebook, bt their is also not working now🙁
@@ishaa-in5kv I don't think cpu matter, it's the 3050 ti gpu that makes the installation different. I don't have nvidia card so I can't test it. I only know the requirement file for nvidia is just pip install -r requirements.txt and you need to download cuda and cudnn. developer.nvidia.com/cuda-11-8-0-download-archive developer.nvidia.com/rdp/cudnn-archive
