AI Knowledge Central
patreon
π₯ Stop Struggling With OVI Setup β This Guide Has Every Link, Model & Step You Need
🕑 Added 2025-11-09 18:35:38 +0000 UTCHey Folks.
You're about to unlock OVI β a super cool AI video tool available right now. This guide will transform you from zero to hero in minutes.
New video just dropped! Watch it first, then come back here to build your setup like a pro.
β‘ THE FAST LANE (Patreon Exclusive)
π 1-CLICK INSTALLER β SKIP EVERYTHING BELOW
Why suffer through terminal commands when you don't have to?
π GET THE MAGIC BUTTON HERE
β¨ No terminal. No headaches. No missing files.
Just pure, automated perfection.
β Perfect for beginners or anyone who values their time
For the SAGE Attention Install guide look here: ποΈ
https://www.patreon.com/posts/speed-up-comfyui-136348957
π οΈ THE MANUAL WAY (For the Brave)
Want full control? Let's build this thing from scratch.
π PHASE 1: Foundation Setup
Step 1: Core Software Stack
WhatWhy You Need ItLinkGitDownloads and manages code repositoriesDownload GitComfyUIYour AI workspace/command center
Step 2: Install ComfyUI Manager
Your dashboard for managing all future extensions.
Bash
# Open CMD/Terminal in: ComfyUI\custom_nodes git clone https://github.com/Comfy-Org/ComfyUI-Manager.git
β Pro Tip: This is your new best friend. Most other nodes can be installed through this manager with one click.
𧩠PHASE 2: Essential Plugins
Install these custom nodes via ComfyUI Manager OR git clone:
πΈ WanVideo Wrapper β Core OVI functionality
https://github.com/kijai/ComfyUI-WanVideoWrapper
πΈ VideoHelper Suite β Video processing tools
https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite
πΈ Essentials β Quality-of-life improvements
https://github.com/cubiq/ComfyUI_essentials
πΈ Use Everywhere β Workflow organization
https://github.com/chrisgoringe/cg-use-everywhere
πΈ KJNodes β Advanced node collection
https://github.com/kijai/ComfyUI-KJNodes
πΈ rgthree β Power user tools
https://github.com/rgthree/rgthree-comfy
π§ PHASE 3: The Brain (AI Models)
Now for the good stuff β the actual AI models.
π CHOOSE YOUR PATH (Based on GPU VRAM)
π₯ 12-16 GB VRAM (Most RTX 3080/3090/4070Ti/4080 users)
Use the FP8 scaled versions β optimized for speed & efficiency
π Place in: models\diffusion_models
Video Model:
text
Audio Model:
text
πͺ 16+ GB VRAM (RTX 4090/5090/Professional Cards)
Use the BF16 versions β maximum quality, no compromises
π Place in: models\diffusion_models
Video Model:
text
Audio Model:
text
π Full Model Library: Browse here
π΅ PHASE 4: Audio Processing (EVERYONE NEEDS BOTH)
π Place in: models\vae
Audio VAE:
text
https://huggingface.co/Kijai/WanVideo_comfy/resolve/main/Ovi/mmaudio_vae_16k_fp32.safetensors
Audio BIG VAE:
text
β οΈ Critical: You must download BOTH audio files or sound generation won't work!
𧬠PHASE 5: Text Encoder (The Prompt Interpreter)
This is what translates your creative descriptions into AI instructions.
π Place in: models\text_encoders
Download:
text
https://huggingface.co/Kijai/WanVideo_comfy/resolve/main/umt5-xxl-enc-bf16.safetensors
πΌοΈ PHASE 6: VAE (Variational AutoEncoder)
The final piece β converts AI data into beautiful visuals.
π Place in: models\vae
Download:
text
https://huggingface.co/Kijai/WanVideo_comfy/resolve/main/Wan2_2_VAE_bf16.safetensors
β FINAL CHECKLIST
Before you launch ComfyUI, verify:
Git installed
β¬οΈWorkflows downloadedβ¬οΈ
ComfyUI downloaded & extracted
ComfyUI Manager installed
All 6 custom nodes installed
Video + Audio models (matched to your VRAM)
Both audio VAE files
Text encoder
Visual VAE
π¬ YOU'RE READY TO CREATE!
Everything is locked and loaded. Fire up ComfyUI and let's make some magic.
