If you want to get good quality video, recommend using a model Q5 or higher.
If you want to use Q5 but lack VRAM, consider using MULTIGPU nodes.
Make sure Git is installed on your system.
Navigate to the comfyui/custom_nodes
folder.
Shift + Right Click → Open Command Prompt (CMD).
Run the following command:
git clone https://github.com/kijai/ComfyUI-TeaCache.git
If an older version of TeaCache exists, delete it.
✅ Done!
Go to the ComfyUI-TeaCache
directory.
Shift + Right Click → Open Command Prompt (CMD).
Run the following command:
git pull
✅ Done!
If it still doesn't work, go to the ComfyUI_windows_portable directory,
hold Shift + Right Click, and open cmd. Then, enter the following command:
> .\python_embeded\python.exe -s -m pip install einops diffusers
Applied native TeaCache
Added Florence2 auto-captioning
Integrated GIMM VFI → Enables 60fps video generation
Added LoRA Loader
Introduced shift parameter
Changed scheduler: simple
→ sgm_uniform
Optimized workflow
ComfyUI-Florence2 (kijai's version)
ComfyUI-GIMM-VFI
ComfyMath
ComfyUI-HunyuanVideoMultiLora
ComfyUI-TeaCache (kijai's version)
ComfyUI-GGUF
ComfyUI-KJNodes
ComfyUI-Impact-Pack
ComfyUI-VideoHelperSuite
ComfyUI-Frame-Interpolation
ComfyUI-Custom-Scripts
rgthree-comfy
🔗 Download
📂 Save to: models/diffusion_models
🔗 Download
📂 Save to: models/diffusion_models
🔗 Download
📂 Save to: models/clip
🔗 Download
📂 Save to: models/clip_vision
🔗 Download
📂 Save to: models/vae
🛠 Recommended Model: 4x_foolhardy_Remacri
📥 Or download via ComfyUI Model Manager
📂 Save to: models/upscale_models
🔗 Download
🔗 Download
Recently, someone has stolen my workflow and falsely claimed it as their own.
Even though this is an open-source project, let’s not engage in such unethical behavior.
Keep in mind that creating a single workflow takes a significant amount of time and effort.
This workflow has been built after extensive testing and validation.
Go ahead and upload yours!
Your query returned no results – please try removing some filters or trying a different term.