Tenscent's Hunyuan Video generation model is probably the best open source model available right now.

This is a beginner friendly tutorial that allows you to add face swap to the existing text to video and video to video workflows!

It works like a wonder, and I've been pranking my brother all day, I can't stop laughing.  But please do remember to use it responsibly and do not invade anyone's privacy!

Tips：

- keep resolution low!  （for example, i'm running 720 x 480 on my RTX 4090)

- sdpa should be available now, but feel free to install flash attention or sage attention for faster generation if you feel comfortable handling dependencies

- if you want to switch to vid2vid, simply right click on the text2vid group, select "bypass model", then go to the vid2vid group, and select "set group nodes to always"

Installation guide:

1) Install the Hunyuan ComfyUI node by Kijai:  https://github.com/kijai/ComfyUI-HunyuanVideoWrapper , and install dependencies

2) Download the Hunyuan model and VAE file here:  https://huggingface.co/Kijai/HunyuanVideo_comfy/tree/main , and place them inside ComfyUI/models/diffusion_models （and vae)

3) Once node is installed, clip and llama should be installed automatically.  If not, goto Kijai's github repo to manually download the files

4) If you encounter problems, I suggest reading this post  by Total-Resort-3120: https://www.reddit.com/r/StableDiffusion/comments/1h7hunp/how_to_run_hunyuanvideo_on_a_single_24gb_vram_card/

Orthe fun text 2 video workflows:

Add face swap to Hunyuan Video Gen： https://openart.ai/workflows/odam_ai/hunyuan-video-generation-face-swap/GyIn6sC02Pg0urjVDh7a

Basic workflow:  https://openart.ai/workflows/odam_ai/hunyuan-video-generation-best-text2vid-vid2vid-model---beginner-friendly/VK1rjBjxxzpp133Q3NR8