Skip to main content
Tencent Hunyuan Image 3.5 Preview is Tencent’s preview image model, available in ComfyUI through Partner Nodes. It generates images from text and edits existing images from an instruction, with a focus on multilingual text rendering, photorealism, and consistency between a reference image and the edited result. Compared with earlier Hunyuan image models, HY Image 3.5 Preview strengthens three areas: multilingual text rendering and typesetting for advertising, infographics, and UI design; photorealism and stylistic range across cinematic, comic, commercial photography, and illustration; and editing consistency that preserves the key traits of people and products across background, outfit, and style changes. Two workflows are available. Tencent HY Image: Text to Image generates an image from a prompt, and Tencent HY Image: Edit edits or combines up to five reference images. Both run on Tencent’s servers through Comfy API, so no model downloads or local GPU are required.

​
What HY Image 3.5 Preview is good at

  • Multilingual text and typesetting: renders legible text for posters, infographics, and UI mockups instead of plain scene images
  • Photorealistic range: covers cinematic stills, comic art, commercial photography, and illustration from short prompts
  • Editing consistency: keeps the defining traits of people and products stable when you change the background, outfit, or style
  • Up to 5 reference images: combine several images in one edit and reference each one in the prompt
  • Output up to 4K: renders up to 2K natively and upscales 4K output from a 2K render
To use the Partner Nodes, you need to ensure that you are logged in properly and using a permitted network environment. Please refer to the Partner Nodes Overview section of the documentation to understand the specific requirements for using the Partner Nodes.
Make sure your ComfyUI is updated.Workflows in this guide can be found in the Workflow Templates. If you can’t find them in the template, your ComfyUI may be outdated.If nodes are missing when loading a workflow, possible reasons:
  1. You are not using the latest ComfyUI version (Nightly version)
  2. Some nodes failed to import at startup

​
Tencent HY Image 3.5 Preview text to image workflow

Generate one image from a text prompt. The model rewrites and expands the prompt before rendering, so short prompts work.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “HY Image 3.5” in Template Library
Poster reading HY Image 3.5 Preview in white capitals over a glossy liquid-metal shape with iridescent highlights on a dark background

​
Steps to run the workflow

  1. Edit the prompt in the Tencent HY Image: Text to Image node
  2. (Optional) Set aspect_ratio to control the output shape. With auto the model picks the aspect ratio from the prompt, and auto is not available at 4K
  3. (Optional) Change resolution from the default 2K. 1K is about 1024x1024 and 2K is about 2048x2048 pixels. 4K output is rendered at 2K and then upscaled by the model
  4. (Optional) Switch resolution to custom to set an exact width and height instead. Both must be multiples of 16 and the total area can be up to 4096x4096. Above roughly a 6:1 ratio the model starts repeating the subject
  5. Click Run or press Ctrl(cmd) + Enter to generate the image
  6. View the result in the Save Image node

​
Tencent HY Image 3.5 Preview image edit workflow

Edit or combine up to five reference images from a text instruction. Use it to change a background, swap an outfit, restyle a product shot, or merge several references into one image while keeping the subject’s defining traits. Tencent HY Image 3.5 Preview image edit example output

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “HY Image 3.5” in Template Library
Angular matte black concept motorcycle on an orange paddock stand against a deep red background

​
Steps to run the workflow

  1. Upload the sample image into the Load Image node that is already connected to the Tencent HY Image: Edit node. The downloaded workflow refers to the file by name (futuristic_brutalist_motorcycle.png), so that node stays empty until you upload it. To edit your own images instead, connect 1 to 5 images to the images input of the Tencent HY Image: Edit node
  2. Write the editing instruction in prompt. Reference the inputs in order as @Image1, @Image2, and so on. A batched input counts once per image
  3. (Optional) Set aspect_ratio and resolution. With auto, the output follows the aspect ratio of the first reference image. 4K output is rendered at 2K and then upscaled by the model
  4. (Optional) Set reference_detail to high to let the model see references at up to 2048x2048 pixels instead of 1024x1024. This preserves small text and fine detail better but takes longer
  5. Click Run or press Ctrl(cmd) + Enter to generate the image
  6. View the result in the Save Image node

​
Prompting tips

  • Describe the edit and the parts that must stay unchanged separately, for example @Image1 change the background to a night market, keep the subject and jacket unchanged
  • Number references in the order they are connected. The numbering in the prompt has to match the input order
  • watermark sits in the node’s advanced options and is off by default. Turn it on to add an AI-generated watermark to the output

​
Get started

  1. Update ComfyUI to the latest version, or use Comfy Cloud
  2. Go to Template Library and search for HY Image 3.5
  3. Open the text to image or image edit workflow and run it