Designers and Illustrators
Develop visual directions and compare styles before refining a composition.
MiniMax H3 Image generates and edits detailed images from text and visual references. Its native output is around 16 megapixels, with support for up to 16 reference images to guide subjects, style and composition.
Explore official examples, then use an editable starting prompt to develop your own image. Results vary with the selected model and settings.

MiniMax H3 Image
Western Key Art

MiniMax H3 Image
Vehicle Design Study

MiniMax H3 Image
Classical Table Tennis

MiniMax H3 Image
Framed Horizon

MiniMax H3 Image
Environment Concepts

MiniMax H3 Image
Furnished Living Room
An image generation and editing model from MiniMax.
MiniMax H3 Image brings text-driven creation and reference-guided editing into one model. Its community license records a November 10, 2026 release date; the official launch article is dated November 11.
The model produces detailed images directly, without an upscaling stage. Its published examples span photography, illustration, characters, environments and precise visual edits.
Build an image in three ways:
Guide detail, composition and edits with a clear visual brief.
Generate around 16 megapixels directly. Published examples include square, portrait and panoramic dimensions.
Combine visual guidance from several images in one editing request. Identify the role of each reference clearly.
Change selected subjects or actions while describing what should stay consistent. Review the result for unintended changes.
Explore natural textures, lighting and small details in portraits, objects and environments.
Describe subject relationships, framing and atmosphere together to guide a complex composition.
Develop photographic scenes, illustrations, character studies and game environments from the same creative workflow.
Explore ideas across photography, design and visual storytelling.
Develop visual directions and compare styles before refining a composition.
Explore characters, props and environments with a consistent visual brief.
Use references to plan product scenes and campaign concepts.
Try changes to subjects, poses and atmosphere while reviewing fine details.
Start with a brief, review the settings and refine your result.
Write the subject, style and composition. Add reference images for an edit or start with a gallery example.
Review the selected model, available dimensions and credit cost in the generator.
Submit when ready. Inspect details and text, then adjust the prompt for your next version.
Published model capabilities; the generator shows the settings for your selected model.
| Developer | MiniMax |
| Modes | Text to image; reference-guided generation; image editing |
| Reference images | Up to 16 |
| Native output | Around 16 megapixels, without an upscaler |
| Square example | 4096 × 4096 (1:1) |
| Landscape example | 5472 × 3072 (16:9) |
| Panoramic example | 6272 × 2688 (21:9) |
| Published working range | About 4–20 megapixels; best results around 16 megapixels |
Last updated: November 12, 2026
MiniMax published capabilities compared with Nano Banana 2 settings offered on Shotyard. Output labels describe different sizing systems.
| Capability | MiniMax H3 Image | Nano Banana 2 |
|---|---|---|
| Text-driven creation | Yes | Yes |
| Reference-based editing | Yes | Yes |
| Reference image limit | Up to 16 | Up to 10 on Shotyard |
| Output sizing | Around 16 megapixels natively | 1K, 2K and 4K settings on Shotyard |
MiniMax H3 Image Advantages: Native high-resolution images and a larger published reference-image limit.
Nano Banana 2 Advantages: Flexible output settings for iterative image creation and editing.
Explore MiniMax H3 Image for detailed, reference-rich compositions. Choose Nano Banana 2 for image creation with adjustable output settings.
Choose a Shotyard plan for image creation. The generator shows the credit cost for the selected model and settings before you submit.
Learn about references, resolution and the creative workflow.
It is MiniMax’s image generation and editing model, built to interpret text and visual references in a unified workflow.
MiniMax documents support for up to 16 reference images. Name the role of each image and check the input limit of your selected tool.
Yes. MiniMax documents native output around 16 megapixels, including 4096 × 4096 square images, without an upscaler.
Yes. Provide a reference and describe the changes. Specify what should remain consistent, then review faces, hands, text and object relationships.
The gallery pairs official examples with editable starting prompts. These prompts describe related scenes, rather than reproduce unpublished generation settings. The room example includes a source-and-result board; its button starts a new room brief without uploading that board.
Image generation on Shotyard uses credits. Review the cost shown for the selected model and settings before submitting. Applying an example only fills the form.
Explore the MiniMax H3 Image AI Image Generator on Shotyard. Choose an example or enter your own visual brief.