MyBits - Wan2GP with low Vram

  In this second episode I will show you how to download and install Wan2GP. How we have moved the installation to our G: Drive to eliminate the C: drive filling up too quickly by using Wan2GP’s features.

  This video is for the budget minded hobbyist who doesn’t want to break the bank to start doing some AI generations at home on a lower Vram machine.

  Follow the free series to build and install the free open source options to get started with AI at home for free.

  In this video we use a DIY budget PC with only 8GB VRAM. We use it to Generate free lip syncing video avatars using Wan2GP.

Wan2GP -Installation Instructions

 

 

Step 1. You need Miniconda installed on our system to install it go to this link and download it to your windows 11 machine:

https://www.anaconda.com/download/

Step 2. You also need to have Git installed on your windows 11 machine:

https://git-scm.com/downloads

 

Step 3. Open an Anaconda prompt by searching in the windows search bar for anaconda prompt.
Hit enter to start a prompt you should see:

(base)C:\Users\Your_user_name>

In this video we are changing where we are downloading and installing Wan2GP to our G: drive, to do this change your base by typing G: after the prompt:

(base)C:\Users\Your_User_Name>G:

You should now see:

(base)G:\>

Step 4.  At the the new prompt copy and paste the line below to download the Wan2GP git file.

git clone https://github.com/deepbeepmeep/Wan2GP.git

 

Step 5.  After it downloads change to the new directory it made when installing
by typing the line below:

(base)G:\>cd Wan2GP

you should now see:(base)G:\Wan2GP>

 

Step 6.  Now Create a conda Virtual Environment named wan2gp type or copy and paste the line below after the prompt:

conda create -n wan2gp python=3.10.9 -y

 

Step 7. Now activate the environment by typing:

conda activate Wan2GP

now you should see:

(Wan2GP) G:\Wan2GP>

you are now in the virtual environment Wan2GP

 

Step 8. Now Install Pytorch+cuda type or copy and paste the line below after the prompt:

pip install torch==2.6.0 torchvision torchaudio –index-url https://download.pytorch.org/whl/test/cu124

Be patient it may take awhile depending on your download speed……

 

Step 9. Install Dependencies for Pytorch and cuda copy and paste or type the line below after the promp:

pip install -r requirements.txt

this takes a long time to load if your download speed is slow….

 

Optional steps to help low vram cards

Step 10. Before running wan2gp for the first time add these next two Optional things to add speed to your generations, FusionXloras and sage attention. You will use either sage attention 1 or 2 not both. 2 is what I used for my lowvram machine.

Installing FusionX Lora go to this location:
https://huggingface.co/vrgamedevgirl84/Wan14BT2VFusioniX/tree/main

Once at the location open the FusionXLora folder and find the file :

Wan2.1_I2V_14B_FusionX_Lora.safetensors

Download this file to your computer, then when it is finished locate it in your downloads and move it to the loras file location below.

To find the location on your computer go to Home folder on our computer and in the address bar type: %Homepath%

open the wan2gp folder and find the folder:
loras_ i2v
open it and move the downloaded file:
Wan2.1_I2V_14B_FusionX_Lora.safetensors
inside it.

 

Step 11. To install Sage attention you need to install triton-windows copy and paste or type the following line after the prompt:

pip install triton-windows

Sage Attention 1(I do not use this one I only included it as another option if sage attention 2 does not work for you.)

pip install sageattention==1.0.6

 

Step 12. Sage Attention 2

( I use this one) for low vram copy and paste or type this command at the prompt:

pip install https://github.com/woct0rdho/SageAttention/releases/download/v2.1.1-windows/sageattention-2.1.1+cu128torch2.7.0-cp310-cp310-win_amd64.whl

this may take awhile……….

 

Running Wan2GP for the first time.

To Run Wan2gp type this command at the prompt:

python wgp.py –i2v

after it loads be patient the first time it takes a long time to load

you will see:
Running on local URL: http://localhost:7860
copy and paste http://localhost:7860  into your browser window and if you have done everything it should open Wan2GP.

 

To run Wan2GP the next time start an anaconda prompt
should see: (base)C:\Users\Your_User_Name>
(type:G: if you installed it on your G: drive like in the video if not you will skip this step.

should see;

(base)G:\>

type:
cd Wan2GP

should see: (base)G:\wan2gp>
then type:
conda activate wan2gp

should see (wan2gp)G:\wan2gp>

now type:

python wgp.py –i2v

 

 

to exit the virtual environment type:
conda deactivate
that will bring you back to the base environment
(base)C:\Users\Your_user_name>

then you can just exit the terminal

 

Using Flash Attention

( I found personally that Sage attention 2 and the defualt attention worked better than this option. Be sure to download the correct version for your system from the download page match the pytorch version the cuda version and the python version. Download it and then type:

pip install (path to the downloaded file -you can drag it here from file explorer)

https://huggingface.co/lldacing/flash-attention-windows-wheel/tree/main

In the configuration tab find the attention type and then switch it to Flash Attention and save.

 

To update Wan2GP

Get in the directory where WanGP is installed on your system and copy and paste or type:
git pull