Natural detail
Shape light, reflections, and material texture with output options up to 4K.
Create with MiniMax H3 Image
MiniMax H3 Image
Create detailed images, direct the composition, and bring up to 16 references into a single creative workflow.
From a clear brief to a detailed image
Directed composition
Give subjects, positions, and relationships a clear place in your prompt.
Up to 16 references
Bring product, setting, and style references into one image.
Focused image edits
Describe what should change and what you want to preserve.
Key features
Make the details serve your idea
Explore product photography, carefully arranged scenes, and reference-led image editing.
Natural detail · up to 4K
Give every material its own character
Build a product image around the way its surfaces catch the light. H3 Image supports detailed image generation, with 1K, 2K, and 4K options in the Nanobana workflow.
- Describe glaze, bamboo, stone, and other materials separately.
- Set the direction and softness of your light.
- Choose the output size for your next use and inspect the details.
Prompt adherence
Put your composition into words
Specify the number of subjects, their positions, and the space between them. Give the model a scene to arrange, from a museum display to an editorial still life.
- State how many objects belong in the scene.
- Use left, center, and right to describe their positions.
- Review object counts, proportions, and spatial relationships.
Multiple image references
Give each reference a role
Combine a product reference with a setting or visual direction. H3 Image accepts up to 16 reference images, so you can explain which details each image should contribute.
- Add JPEG, PNG, or WebP references, up to 20 MB each.
- Name the subject, setting, or layout to take from each image.
- Check identity, scale, and lighting in the combined result.
Reference images
Combined scene
Image editing
Change the glaze, keep the scene
Use an image as your starting point and describe one focused change. Name the surrounding shapes, materials, and lighting you want to keep, then compare the result with your reference.
- Describe the specific object or surface to change.
- State which details and framing should stay consistent.
- Review the result before using it for your next edit.
Before
After
H3 Image demos and community walkthroughs
Explore model-specific demonstrations from SECourses and CongZ AI. These videos cover local ComfyUI workflows; Nanobana provides its own web workspace.
Creator demonstrations
MiniMax H3 Image Is Here: 16 MP Images and Edits in 6 Steps, Running Locally in ComfyUI
SECourses
MiniMax H3 Image Is Here: 16 MP Images and Edits in 6 Steps, Running Locally in ComfyUI
View original · YouTubeMiniMax H3 Image Tutorial: Create and Edit Images in ComfyUI
CongZ AI
MiniMax H3 Image Tutorial: Create and Edit Images in ComfyUI
View original · YouTubeHow it works
Create with H3 Image in three steps
Use the same image workspace as the Nanobana homepage.
01
Describe your image
Write the subject, arrangement, materials, and lighting. Select an aspect ratio and output size.
02
Add references when needed
Upload images to edit or combine. Tell the model what to take from each reference and what to preserve.
03
Generate and refine
Sign in, review the displayed credit estimate, and generate. Inspect and download the result or continue working in Canvas.
MiniMax H3 Image questions
What is MiniMax H3 Image?+
MiniMax H3 Image is an image generation and editing model developed by MiniMax. It supports text-to-image, reference-based generation, and image editing. It is a separate image model from the MiniMax H3 video option.
How does MiniMax H3 Image relate to Nanobana?+
Nanobana is an independent creative platform that provides a web workflow for third-party models. MiniMax develops H3 Image; Nanobana does not develop or own it, and this page does not imply a partnership or endorsement.
Do I need a separate MiniMax account?+
You do not need to enter a MiniMax API key in this generator. Sign in to Nanobana and use its supported generation workflow and account credits.
Which modes are available here?+
Nanobana supports text-to-image and image-to-image with MiniMax H3 Image. Start from a prompt or add reference images for an edit. This image workflow does not generate video or offer mask-based editing.
How many reference images can I use?+
You can add up to 16 JPEG, PNG, or WebP reference images, each up to 20 MB in this workflow. Explain the role of each reference, especially when combining subjects and settings.
Which sizes and formats can I select?+
The generator offers 1K, 2K, and 4K output levels, automatic or fixed aspect ratios, and PNG or JPEG output. Pixel dimensions depend on the selected aspect ratio and output level. This workflow does not offer transparent backgrounds.
How are credits calculated?+
The estimate depends on output size, image quantity, and the number of reference images. Each reference contributes to the per-image cost. Check the current estimate in the generator before submitting.
Will my references and edited details stay exactly the same?+
Reference images guide generation, but they do not guarantee an exact match. Describe the details to keep and the changes to make. Review text, object counts, identity, geometry, and fine details before using the output.
Start with the image you have in mind
Describe a new scene or bring a reference into the H3 Image workspace.
Create with MiniMax H3 Image