Video from text: HunyuanVideo 1.5

<!-- lesson: video-hunyuan -->

Lesson "video-hunyuan", in the group Video of the lesson list. Steps: 14.

#What you will be able to do

You build the HunyuanVideo 1.5 graph with a custom sampler and get a five-second 720p clip.

#What you need

• It goes on from Video on a modest graphics card: Wan 2.2 5B. The card says so; it is a hint, not a lock.
• Model files the lesson checks:
  • for unet_name of Load Diffusion Model: a file whose name has hunyuanvideo1.5 and t2v;
  • for clip_name1 of DualCLIPLoader: a file whose name has qwen_2.5_vl;
  • for clip_name2 of DualCLIPLoader: a file whose name has byt5;
  • for vae_name of Load VAE: a file whose name has hunyuanvideo15_vae.
• Nodes: Load Diffusion Model, DualCLIPLoader, Load VAE, CLIP Text Encode (Prompt), ModelSamplingSD3, Empty Hunyuan Video 1.5 Latent, RandomNoise, KSamplerSelect, BasicScheduler, CFG Guider, SamplerCustomAdvanced, VAE Decode / VAE Decode (Tiled), Create Video, Save Video.
• The app downloads and installs nothing. The step texts below say which files and folders, and which ComfyUI template names them.

#The steps

The words of each step, as the lesson shows them. Node names are in bold.

1. What it costs, first (reading)

   HunyuanVideo 1.5 by Tencent is an 8B video model, between the small and the large Wan in size. Its graph uses the custom sampler: noise, a guider, a sampler and a list of steps as separate nodes.

   What you need: the 720p_t2v model, about 17 GB in fp16, in models/diffusion_models; two text encoders (qwen_2.5_vl_7b in fp8 and the small byt5 for lettering) in models/text_encoders; hunyuanvideo15_vae in models/vae. A card with 16 GB or more is comfortable; 121 frames at 1280 × 720 take a long time, a smaller size much less.

   If the canvas already holds another graph, start from an empty one: "File ▸ New project".

2. Load the model

   Add a Load Diffusion Model node and choose the hunyuanvideo1.5_720p_t2v file in unet_name.

   If the file you need is not in the list, your ComfyUI does not have it. The app downloads nothing. The easiest way is to open, in ComfyUI itself, the "Hunyuan Video 1.5 Text to Video" template: it shows which files are missing and where to put them. Then choose "Sync ▸ Sync now" in this app's menu, and the files appear in the lists.

3. Load the two text encoders

   Add a DualCLIPLoader node. In clip_name1 choose qwen_2.5_vl_7b, in clip_name2 choose byt5_small_glyphxl, and set type to hunyuan_video_15.

   The first reads the prompt, the second helps with text written in the picture.

4. Load the VAE

   Add a Load VAE node and choose hunyuanvideo15_vae in vae_name. The VAE of HunyuanVideo 1.0 does not fit version 1.5.

5. Write the prompt

   Add two CLIP Text Encode (Prompt) nodes and connect the CLIP output of DualCLIPLoader to the clip input of each. In the text field of one, describe the clip: who or what is in the frame, what happens, how the camera moves. The other is the negative prompt: what must not be there.

   For video the action matters most. "A cat jumps onto a windowsill, the camera follows" works better than "a cat on a windowsill".

6. Set the shift

   Add a ModelSamplingSD3 node and connect the MODEL output of the loader to its model input. Set shift to 7.

7. Size and length

   Add a Empty Hunyuan Video 1.5 Latent node. Set width to 1280, height to 720 and length to 121.

   121 frames at 24 fps make five seconds; the number of frames is of the form 4n + 1. For a lighter run take the 480p model and 848 × 480.

8. Noise, sampler and steps

   Add RandomNoise, KSamplerSelect and BasicScheduler.

   • In sampler_name of KSamplerSelect choose euler.
   • Connect the model to the model input of BasicScheduler, set scheduler to simple and steps to 20.

   The scheduler works out the list of noise levels for the twenty steps; the noise node holds the seed.

9. The guider

   Add a CFG Guider node. Its model input takes the output of ModelSamplingSD3; its positive and negative inputs take the two prompts. Set cfg to 6.

10. The sampler

Add a SamplerCustomAdvanced node and connect to it: noise from RandomNoise, guider from CFG Guider, sampler from KSamplerSelect, sigmas from BasicScheduler, latent_image from Empty Hunyuan Video 1.5 Latent.

11. Turn the result into frames

Add a VAE Decode node. Connect the output of SamplerCustomAdvanced to its samples input and the VAE to its vae input. With little memory take VAE Decode (Tiled) instead.

12. Make the clip and save it

Add Create Video and Save Video. The frames from VAE Decode go into the images input of Create Video, and its output into the video input of Save Video. Set fps to 24.

HunyuanVideo 1.5 draws 24 fps. With another number the clip runs faster or slower than it was meant to.

13. What else the template does (reading)

The official template adds two things on top of this graph:

• EasyCache between the loader and the shift: it skips work on steps that change little, faster with a small loss of detail;
• a second stage that enlarges the clip to 1080p with its own model and latent upscaler.

Both can be added later; the app accepts EasyCache on the way of the model.

14. The graph is ready (reading)

Save the graph with "File ▸ Save workflow for ComfyUI…", open the file in ComfyUI and run it there. This app runs nothing in ComfyUI itself.

#What the lesson checks

A reading step is done when you press Got it, next. A doing step ticks itself off when the canvas holds:

• Step 2, Load the model:
  • a Load Diffusion Model node.
  • unet_name of Load Diffusion Model holds a file whose name has hunyuanvideo1.5 and t2v (capitals, dots and dashes do not matter).
• Step 3, Load the two text encoders:
  • a DualCLIPLoader node.
  • clip_name1 of DualCLIPLoader holds a file whose name has qwen_2.5_vl (capitals, dots and dashes do not matter).
  • clip_name2 of DualCLIPLoader is not the same file as clip_name1 of DualCLIPLoader.
  • type of DualCLIPLoader is hunyuan_video_15.
  • clip_name2 of DualCLIPLoader holds a file whose name has byt5 (capitals, dots and dashes do not matter).
• Step 4, Load the VAE:
  • a Load VAE node.
  • vae_name of Load VAE holds a file whose name has hunyuanvideo15_vae (capitals, dots and dashes do not matter).
• Step 5, Write the prompt:
  • a CLIP Text Encode (Prompt) node.
  • a CLIP Text Encode (Prompt) node.
  • CLIP of DualCLIPLoader into clip of CLIP Text Encode (Prompt) (prompt).
  • CLIP of DualCLIPLoader into clip of CLIP Text Encode (Prompt) (negative prompt).
  • text of CLIP Text Encode (Prompt) (prompt) holds at least 3 characters.
• Step 6, Set the shift:
  • a ModelSamplingSD3 node.
  • MODEL of Load Diffusion Model into model of ModelSamplingSD3 (a LoraLoaderModelOnly or EasyCache in between is accepted).
  • shift of ModelSamplingSD3 is from 5 to 9; Do this for me sets 7.
• Step 7, Size and length:
  • a Empty Hunyuan Video 1.5 Latent node.
  • width of Empty Hunyuan Video 1.5 Latent is from 640 to 1280; Do this for me sets 1280.
  • height of Empty Hunyuan Video 1.5 Latent is from 368 to 720; Do this for me sets 720.
  • length of Empty Hunyuan Video 1.5 Latent is from 33 to 121; Do this for me sets 121.
• Step 8, Noise, sampler and steps:
  • a RandomNoise node.
  • a KSamplerSelect node.
  • a BasicScheduler node.
  • MODEL of Load Diffusion Model into model of BasicScheduler (a LoraLoaderModelOnly or EasyCache in between is accepted).
  • sampler_name of KSamplerSelect is euler.
  • scheduler of BasicScheduler is simple.
  • steps of BasicScheduler is from 8 to 50; Do this for me sets 20.
• Step 9, The guider:
  • a CFG Guider node.
  • MODEL of ModelSamplingSD3 into model of CFG Guider (a LoraLoaderModelOnly or EasyCache in between is accepted).
  • CONDITIONING of CLIP Text Encode (Prompt) (prompt) into positive of CFG Guider.
  • CONDITIONING of CLIP Text Encode (Prompt) (negative prompt) into negative of CFG Guider.
  • cfg of CFG Guider is from 1 to 8; Do this for me sets 6.
• Step 10, The sampler:
  • a SamplerCustomAdvanced node.
  • NOISE of RandomNoise into noise of SamplerCustomAdvanced.
  • GUIDER of CFG Guider into guider of SamplerCustomAdvanced.
  • SAMPLER of KSamplerSelect into sampler of SamplerCustomAdvanced.
  • SIGMAS of BasicScheduler into sigmas of SamplerCustomAdvanced.
  • LATENT of Empty Hunyuan Video 1.5 Latent into latent_image of SamplerCustomAdvanced.
• Step 11, Turn the result into frames:
  • a VAE Decode or VAE Decode (Tiled) node.
  • output of SamplerCustomAdvanced into samples of VAE Decode or VAE Decode (Tiled).
  • VAE of Load VAE into vae of VAE Decode or VAE Decode (Tiled).
• Step 12, Make the clip and save it:
  • a Create Video node.
  • a Save Video node.
  • IMAGE of VAE Decode or VAE Decode (Tiled) into images of Create Video.
  • VIDEO of Create Video into video of Save Video.
  • fps of Create Video is 24.

#See also

• HunyuanVideo
• Video generation
• How lessons work