---
title: "HunyuanVideo 1.5 in ComfyUI: T2V and I2V on Consumer GPUs"
description: "Run HunyuanVideo 1.5 in ComfyUI on a 24GB GPU: 8.3B text-to-video and image-to-video at 720p, plus a super-resolution upscaler for 1080p."
sidebarTitle: "HunyuanVideo 1.5"
---
import UpdateReminder from "/snippets/tutorials/update-reminder.mdx";
[HunyuanVideo 1.5](https://github.com/Tencent/HunyuanVideo) is a lightweight 8.3B parameter model developed by Tencent's Hunyuan team. It delivers flagship-quality video generation on consumer GPUs (24GB VRAM), dramatically lowering the barrier to entry without compromising on quality.
## Model highlights
- **Compact powerhouse**: Delivers SOTA performance comparable to larger models while running on consumer hardware.
- **Versatile generation**: Supports high-quality Text-to-Video and Image-to-Video (5-10s) with exceptional consistency.
- **Precise control**: Strong instruction following for camera movements, physics, and emotional expressions.
- **Cinematic quality**: Native 720p output (upscalable to 1080p) with professional aesthetics.
- **Rich features**: Supports diverse styles (realistic, anime, 3D) and in-video text rendering (Chinese/English).
## Common models for all workflows
The following models are used in both Text-to-Video and Image-to-Video workflows. Download and save them to the specified directories:
qwen_2.5_vl_7b_fp8_scaled.safetensors — save to ComfyUI/models/text_encoders/
byt5_small_glyphxl_fp16.safetensors — save to ComfyUI/models/text_encoders/
hunyuanvideo15_vae_fp16.safetensors — save to ComfyUI/models/vae/
#### Storage
```
ComfyUI/
├── 📂 models/
│ ├── 📂 text_encoders/
│ │ ├── qwen_2.5_vl_7b_fp8_scaled.safetensors
│ │ └── byt5_small_glyphxl_fp16.safetensors
│ └── 📂 vae/
│ └── hunyuanvideo15_vae_fp16.safetensors
```
## Hunyuan Video 1.5 Text-to-Video Workflow
HunyuanVideo 1.5 Text-to-Video generates 5-10 second videos from natural language descriptions with enhanced quality and reduced VRAM requirements.
Open in Comfy Cloud
Download JSON or search "Hunyuan Video 1.5 T2V" in Template Library
### Model downloads
hunyuanvideo1.5_720p_t2v_fp16.safetensors — save to ComfyUI/models/diffusion_models/
#### Model storage
```
ComfyUI/
├── 📂 models/
│ ├── 📂 text_encoders/
│ │ ├── qwen_2.5_vl_7b_fp8_scaled.safetensors // Shared model
│ │ └── byt5_small_glyphxl_fp16.safetensors // Shared model
│ ├── 📂 vae/
│ │ └── hunyuanvideo15_vae_fp16.safetensors // Shared model
│ └── 📂 diffusion_models/
│ └── hunyuanvideo1.5_720p_t2v_fp16.safetensors // T2V model
```
### Steps to run the workflow
1. Ensure the `DualCLIPLoader` node has loaded these models:
- clip_name1: `qwen_2.5_vl_7b_fp8_scaled.safetensors`
- clip_name2: `byt5_small_glyphxl_fp16.safetensors`
2. Ensure the `Load Diffusion Model` node has loaded `hunyuanvideo1.5_720p_t2v_fp16.safetensors`
3. Ensure the `Load VAE` node has loaded `hunyuanvideo15_vae_fp16.safetensors`
4. Click the `Queue` button or use the shortcut `Ctrl(Cmd) + Enter` to run the workflow
The workflow includes a super-resolution upscaler node. When enabled, it uses `hunyuanvideo1.5_1080p_sr_distilled_fp16.safetensors` to upscale the output to 1080p.
## Hunyuan Video 1.5 Image-to-Video Workflow
HunyuanVideo 1.5 Image-to-Video transforms static images into smooth, high-quality videos with improved consistency and motion dynamics.
Open in Comfy Cloud
Download JSON or search "Hunyuan Video 1.5 I2V" in Template Library
**Input materials**
Download the default input image, or use your own image as the starting frame.
### Model downloads
sigclip_vision_patch14_384.safetensors — save to ComfyUI/models/clip_vision/
hunyuanvideo1.5_720p_i2v_fp16.safetensors — save to ComfyUI/models/diffusion_models/
#### Model storage
```
ComfyUI/
├── 📂 models/
│ ├── 📂 clip_vision/
│ │ └── sigclip_vision_patch14_384.safetensors // I2V vision encoder
│ ├── 📂 text_encoders/
│ │ ├── qwen_2.5_vl_7b_fp8_scaled.safetensors // Shared model
│ │ └── byt5_small_glyphxl_fp16.safetensors // Shared model
│ ├── 📂 vae/
│ │ └── hunyuanvideo15_vae_fp16.safetensors // Shared model
│ └── 📂 diffusion_models/
│ └── hunyuanvideo1.5_720p_i2v_fp16.safetensors // I2V model
```
### Steps to run the workflow
1. Ensure the `DualCLIPLoader` node has loaded these models:
- clip_name1: `qwen_2.5_vl_7b_fp8_scaled.safetensors`
- clip_name2: `byt5_small_glyphxl_fp16.safetensors`
2. Ensure the `CLIPVisionLoader` node has loaded `sigclip_vision_patch14_384.safetensors`
3. Ensure the `Load Diffusion Model` node has loaded `hunyuanvideo1.5_720p_i2v_fp16.safetensors`
4. Ensure the `Load VAE` node has loaded `hunyuanvideo15_vae_fp16.safetensors`
5. Click the `Queue` button or use the shortcut `Ctrl(Cmd) + Enter` to run the workflow
## Super-resolution upscaler
Both workflows include a super-resolution node that can upscale the output video from 720p to 1080p. This optional upscaler uses a distilled model for efficient high-resolution output.
hunyuanvideo1.5_1080p_sr_distilled_fp16.safetensors — save to ComfyUI/models/diffusion_models/
All shared and workflow-specific models combined:
```
ComfyUI/
├── 📂 models/
│ ├── 📂 clip_vision/
│ │ └── sigclip_vision_patch14_384.safetensors
│ ├── 📂 text_encoders/
│ │ ├── qwen_2.5_vl_7b_fp8_scaled.safetensors
│ │ └── byt5_small_glyphxl_fp16.safetensors
│ ├── 📂 vae/
│ │ └── hunyuanvideo15_vae_fp16.safetensors
│ └── 📂 diffusion_models/
│ ├── hunyuanvideo1.5_720p_t2v_fp16.safetensors // T2V model
│ ├── hunyuanvideo1.5_720p_i2v_fp16.safetensors // I2V model
│ └── hunyuanvideo1.5_1080p_sr_distilled_fp16.safetensors // Upscaler
```