Install the current ComfyUI release
Install ComfyUI for Windows, update it, then confirm that it sees your NVIDIA GPU before downloading the large model files.
Official ComfyUI installationVideo / WORKFLOW
Animate a character with Wan Animate 2
Transfer motion from a driving video onto a character image using the official Wan Animate 2 ComfyUI workflow and its complete five-file model set.
EstimatedSource checked 9/29/2026
LOCALRENTED GPU
Before you begin
Hardware
These rows describe the complete Wan Animate 2 INT8 ComfyUI pipeline for one 482×854, 81-frame clip. They are capacity estimates, not render-time benchmarks.
| Machine | Wan Animate 2 · INT8 |
|---|---|
| NVIDIA GPU · 24 GB | May need offload for the complete 81-frame graph; use 32 GB for a simpler first run. |
| NVIDIA GPU · 32 GB | Estimated starting point for the exact 482×854, 81-frame template; verify the full graph before increasing duration. |
| Rented NVIDIA GPU · 48 GB | More headroom for the full graph and video frames; confirm runtime support and price before renting. |
| AMD Radeon · 32 GB | No documented ComfyUI INT8 ConvRot route for this exact template on AMD; use a verified NVIDIA setup or rent one. |
| Intel GPU · 32 GB | No documented ComfyUI INT8 ConvRot route for this exact template on Intel; use a verified NVIDIA setup or rent one. |
| Apple Silicon Mac · 64 GB unified | FP8 has a reported MPS limitation; use a remote NVIDIA ComfyUI host until this exact Mac path is verified. |
| CPU-only · 64 GB RAM | No practical documented CPU-only route for this 14B video graph; use a compatible NVIDIA GPU or rented host. |
Mac, AMD, Intel and CPU rows explain route availability for this exact template, not a claim that these platforms can never run any Wan model. The Mac can still control a rented NVIDIA ComfyUI host through the browser.
VIDEO / CHARACTER ANIMATION / COMFYUISOURCE-LINKED SETUP
Animate a character from a reference video.
Use the first-party Wan Animate 2 ComfyUI template to transfer motion from a short video onto a character image. Download the five exact files, open the visual graph, provide both inputs, and render a first result.
Choose a package Load the visual workflow Check the result
01
BEFORE YOU BEGIN
The documented package
This setup uses the package documented for this task. Review its source and supported platforms before starting.
Save your machine in My Hardware to get an automatic starting choice. You can always choose any package yourself.
CURRENT SETUP
View weight source Wan Animate 2 · INT8 · official
The selected diffusion file is 16.7 GB and the text encoder is about 6.7 GB before the VAE, vision encoder, LoRA, activations, and video frames. A 32 GB NVIDIA GPU is a cautious 480p starting estimate, not a measured peak or a speed guarantee. No smaller-GPU rating is published here until the complete graph is measured.
02
THE WORKFLOW
Set up Wan Animate 2 · INT8 on your machine
Choose your computer first, then its GPU path. Follow the visual setup and import the linked workflow.
CHOOSE YOUR SETUPWindows
Use current ComfyUI and a working NVIDIA driver for the INT8 convrot checkpoint.
Other operating systems
Mac: The chosen INT8 ConvRot diffusion and FP8 text-encoder package has no verified native MPS render path; ComfyUI documents float8 conversion failure on a related Wan graph. Use a NVIDIA host remotely from Mac, or wait for an exact source-backed Mac package. Compatibility source
Put each model file in its exact folder
Download the five files below into the matching ComfyUI/models folders. The diffusion checkpoint is only one part of the pipeline. Restart ComfyUI after the files are in place.
Put these files in ComfyUI/models/5 required files
Open the official motion-transfer canvas
Download and drag the JSON onto ComfyUI, or open the Wan Animate 2 Motion Transfer workflow in its Video template library. Check that every model loader finds the file selected in the template. In the Motion Transfer node, select wan_animate_2_int8_convrot.safetensors and keep the publisher template’s first-run width 482, height 854, and 81 frames; use the listed LoRA, FP8 encoder, vision encoder, and VAE without substitutions.
Download the official Wan Animate 2 workflow See the workflow sourceProvide a character and a motion clip
Load a clear reference image for who the character is, then a short driving video for the motion to copy. Similar full-body or close-up framing in both inputs makes the first test easier to judge.
Example character image Example driving videoWrite appearance and motion prompts separately
In the Motion Transfer subgraph, describe the character and background in the main prompt without motion words. Describe the driving clip’s movement in the pose prompt. For the first run keep the exact Wan Animate 2 INT8 package, 482×854 resolution, 81 frames, and the template’s other default settings. The official JSON supplies these values; do not raise duration or resolution until one short clip renders.
Queue the graph and inspect the result
Click Queue Prompt. When SaveVideo finishes, compare the output with your character image and driving clip. Check identity, movement, and background before attempting longer clips or changing resolution.
03
SUCCESS CHECK · Wan Animate 2 · INT8
Your character moves like the driving clip
ComfyUI writes a playable video. Compare it with the reference image and driving video; verify the character identity and motion transfer rather than relying only on a completed queue.
The public template uses INT8 ConvRot and an FP8 text encoder. The selected graph is documented for NVIDIA; the linked ComfyUI issue shows the FP8 MPS dtype failure on a related Wan graph. That is a route-specific limitation, not proof that all future Wan Animate 2 Mac conversions are impossible. GPU ratings are estimates until the full graph is measured.
When it doesn’t go to plan
A model loader is red or says the file is missing.
Confirm every file is in the folder shown in the model tree with its exact filename. Restart ComfyUI and reselect it in the loader; a download saved as HTML is not a weight file.
The first render runs out of GPU memory.
Keep the official 482×854, 81-frame first run, close other GPU work and try ComfyUI low-memory mode. If 24 GB still fails, use a 32 GB NVIDIA GPU or rent a 48 GB GPU.
The output character does not follow the reference.
Use a clear full-body reference and short driving clip with similar framing. Check that both Load Image and Load Video nodes contain the intended files before adjusting prompts.
A Mac reports MPS float8 dtype conversion errors.
This published INT8/FP8 graph has no verified native Mac route. Use a supported NVIDIA machine or rent one and access its ComfyUI canvas from the Mac browser; do not silently substitute a different model.
REFERENCE LIBRARY
4 SOURCESSources and files
Original instructions, model files, and compatibility notes behind this setup.
The model behind this workflow
Wan-Animate-2 · 14BKeep exploring
Text
SETUP GUIDEQwen3.5 · 9B private chat setup
Source-backed package and platform-route plan for private local chat with Qwen3.5 · 9B. Preserves the official Ollama 9B package, a smaller 4B family fallback, and a clearly labeled community uncensored GGUF option.
Freshness 41 out of 100. Release age: 230 days old, Feb 27, 2026. Measures age, not quality.
Popularity 93 out of 100. 8,591,266 downloads in 30 days; 2,171 lifetime likes. Checked Oct 15, 2026. Measures adoption, not quality.
SOFTWAREOllama
Coding
SETUP GUIDEBuild a private coding assistant with Qwen3.8 27B
Run Qwen3.8 27B with Ollama, connect it to Cline in VS Code, and make one reviewable code change in a tiny local project.
Freshness 76 out of 100. Repository age: 71 days old, Aug 5, 2026. Measures age, not quality.
Popularity 95 out of 100. 7,363,968 downloads in 30 days; 17,749 lifetime likes. Checked Oct 15, 2026. Measures adoption, not quality.
Arena Open Source Web coding: rank 10 of 55. Ranked model: qwen3.8-27b. Original model ranking, not a benchmark of this package. Export dated Oct 13, 2026. Source checked Oct 15, 2026. Human preference ranking, not hardware suitability.+3
SOFTWAREOllama + Visual Studio Code + Cline + Python
Images
SETUP GUIDECreate images locally with FLUX.2 Klein
Generate images from text prompts in ComfyUI with the fast distilled or more adjustable base 4B model. Save your results and reuse the visual workflow.
Freshness 35 out of 100. Release age: 273 days old, Jan 15, 2026. Measures age, not quality.
Popularity 76 out of 100. 423,736 downloads in 30 days; 1,043 lifetime likes. Checked Oct 15, 2026. Measures adoption, not quality.
SOFTWAREComfyUI