Wan 2.2 5B Text-to-Video (ComfyOrg Default)
Write a prompt and generate a short video with the compact Wan 2.2 5B hybrid model. This is the standard ComfyOrg template and a solid starting point.
- L40S
- Text to video
- Free template
How it works
This is a 13-node workflow built from core ComfyUI nodes. UNETLoader loads wan2.2_ti2v_5B_fp16.safetensors, CLIPLoader loads umt5_xxl_fp8_e4m3fn_scaled, and VAELoader loads wan2.2_vae.
The positive and negative prompts go through two CLIPTextEncode nodes. Wan22ImageToVideoLatent creates the video latent at 1280 by 704 with 121 frames. A LoadImage node is connected to it but set to bypass, so the saved workflow behaves as text-to-video.
ModelSamplingSD3 adjusts the model, then KSampler runs 20 steps at CFG 5 with uni_pc and the simple scheduler. VAEDecode converts the latent, and CreateVideo and SaveVideo write the MP4. There is no speed LoRA in this one, so it follows the standard sampling settings.
Tips
- Change the width, height and length in
Wan22ImageToVideoLatentto trade quality against time. - Keep the default negative prompt, which targets overexposure, blur and bad hands.
- Enable the Load Image node if you want to start from a picture.
- For fewer steps, look at the Wan 2.2 14B text-to-video workflow, which adds lightning LoRAs.
- Because it is the stock ComfyOrg template, results are a good reference point when you compare it against LoRA-accelerated Wan 2.2 workflows.
What you can use it for
- Short clips from a text prompt
- A baseline to test Wan 2.2 5B
- Prompt experiments before heavier models
- Starting point for custom video workflows
Run it in three steps
Open the template
Click Run on ComfyDeploy. The workflow, its custom nodes and models are set up for you on a cloud machine.
Add your inputs
Provide positive prompt and negative prompt in the playground or the ComfyUI editor.
Run, share or deploy
Run it on an L40S, share it with your team as a playground, or deploy it as an API endpoint.
Frequently asked questions
What GPU does this Wan 2.2 workflow need?
It is set to run on an L40S GPU on ComfyDeploy. It uses the 5B model, which is smaller than the 14B Wan 2.2 versions.
Can it do image-to-video too?
The model is wan2.2_ti2v_5B, a text and image model, and the graph has a Load Image node, but it is bypassed in this saved version. Enable it to start from an image.
How do I run it as an API?
Deploy it on ComfyDeploy and expose the prompt inputs. It only uses core ComfyUI nodes.
