Learning Timeline
Key Insights
Hardware Requirements
Running the original base model for HunyuanVideo 1.5 requires a CUDA GPU with a minimum of 14 GB of VRAM.
Storage Warning
The diffusion models are significant in size. The standard model and the 1080p upscaler are approximately 15.5 GB each. Ensure sufficient disk space before downloading.
Workflow Template Availability
While ComfyUI has a native 'Templates' browser, HunyuanVideo 1.5 may not appear there immediately. If searching 'Hunyuan' in the Templates menu yields no results, you must manually download and drag-and-drop the JSON workflows as described in the steps.
Step by Step
Updating ComfyUI & Importing Workflows
- Run the ComfyUI updater (e.g., 'update_comfyui.bat') or update via the Manager to ensure the software is on the latest version.
- Launch ComfyUI.
- Navigate to the official HunyuanVideo 1.5 repository or documentation page containing the workflow files.
- Locate the desired workflow JSON file (e.g., 'Text-to-Video' or 'Image-to-Video').
- Download the raw JSON file to your local drive (do not just view the code code; save the file).
- Drag and drop the downloaded JSON file directly onto the open ComfyUI interface to load the node graph.
Installing Required Models
- Download the 'Qwen 2.5VL' text encoder file.
- Move the downloaded Qwen file into the `ComfyUI/models/text_encoders` directory.
- Download the main Hunyuan Video diffusion model (standard 720p version).
- Move the diffusion model file into the `ComfyUI/models/diffusion_models` directory.
- Optionally, download the '1080p SR' (Super Resolution) file if upscaling is required.
- Move the 1080p SR file into the `ComfyUI/models/diffusion_models` directory.
- Refresh ComfyUI to ensure the new models are detected.
