MiniMax H3 Image

MiniMax H3 Image Generator & Editor

Text to image. Image to image. One MiniMax H3 Image workspace for generating and editing.

About this tool

Create images from a prompt, or upload up to 16 reference images and direct the edit in plain language—swap objects, restyle scenes, or compose people, products, and backgrounds from multiple photos. Pick any of 15 aspect ratios, export PNG, JPG, or WebP up to 4K, and pay only for what you generate—no API key, no setup.

0 / 16

Editing prices scale with the number of reference images—each uploaded image adds a small per-image fee. The price updates live below.

1188/4000
Current Credits: 016 credits / image

Examples

Original photo, example 1
MiniMax H3 Image result, example 1
Use the uploaded face photo as the only identity reference. Reproduce the same woman with recognizable facial structure, proportions and natural likeness, but reinterpret her entirely as a hand-painted fashion illustration — never keep the photographic texture. Paint a chest-up 4:5 portrait: light-wash denim jacket with visible stitching over a crisp white shirt, a wide woven straw hat, silver hoop earrings, and round mirror sunglasses. Her head tilts gently upward; the expression is serene and confident; lips carry a glossy red finish; loose strands escape a soft low bun. Work in transparent watercolor layers with wet-on-wet blending, pigment blooms, splashes and visible brush strokes; add fine pen-and-ink contours for the features. Inside the round lenses render a crisp reflection of tall palm trees against a bright blue sky. Warm sunlight falls from the upper left, casting soft shadows under the hat brim. Keep the background airy white paper with scattered muted-blue and beige splatters, generous negative space, and a luxury travel-editorial mood. Avoid: photorealism, 3D render, plastic skin, distorted reflections, extra fingers, heavy outlines, text or watermark.
12 credits
per image, from
16
reference images per edit
15
aspect ratios
4K
max resolution

MiniMax H3 Image Examples

Switch between Image to Image (before on the left, MiniMax H3 Image edit on the right) and Text to Image (prompt-only examples). Click any example to load its prompt into the generator above.

How the MiniMax H3 Image Generator & Editor Works

Pick a mode, feed it a prompt or your photos, and download the result.

Step 1

Pick a mode

Choose Text to Image to create from a prompt alone, or Image to Image to edit reference photos. H3 follows plain-language instructions precisely: transform only the object you name, keep faces, poses, and lighting intact, and leave everything else untouched.

Step 2

Add your prompt and images

Describe the image you want, or start from a built-in preset. For edits, upload 1 to 16 reference images—combine a person from one photo with a product or scene from others—then set the aspect ratio, output format, and resolution.

Step 3

Generate and download

Your image renders in about half a minute at 1K. Download it as PNG, JPG, or WebP, or step up to 2K at the same price and 4K for large-format prints. Results are saved to the Recent generations page for 30 days, so download a local copy before expiry.

Why This MiniMax H3 Image Generator

The full MiniMax H3 Image model surface, wrapped in a workspace you can actually use

Generator & Editor in One

Text-to-image and image-to-image share the same prompt box, aspect ratios, and formats—draft from text, then refine the keeper with references instead of starting over.

Up to 16 Reference Images

H3 image editing accepts up to 16 references per generation: merge subjects, products, and environments from separate photos and steer the composition with instructions.

Precise, Localized Edits

Name the object, region, or text you want changed and H3 edits exactly that—identity, pose, and surroundings stay recognizable, which makes it strong for product retouching and signage swaps.

15 Aspect Ratios, Up to 4K

From 9:21 story formats to 21:9 banners in 1K, 2K, or 4K—1K and 2K cost the same, so re-renders at 2K never surprise you at checkout. Output as PNG, JPG, or WebP.

What to Use MiniMax H3 Image For

Generation and editing use cases the MiniMax H3 Image model handles particularly well.

Product & e-commerce visuals

Drop in product photos and swap backgrounds, restage lighting, or place items in new scenes with instructions—no reshoots, no Photoshop masking. 1:1 renders drop straight into store listings.

Multi-image composition

Combine a model from one shot, a garment from another, and a location from a third—H3 blends up to 16 references into one coherent image while keeping each subject recognizable.

Marketing & social creatives

Generate campaign concepts, ad backgrounds, and post visuals from a prompt in 9:16, 16:9, or 21:9, then iterate on the winners by editing rather than re-prompting from scratch.

Portraits & identity-preserving edits

Restyle a portrait into watercolor illustration, editorial studio looks, or double-exposure art while the face stays clearly recognizable—references keep identity locked.

MiniMax H3 Image Generator & Editor FAQ

Answers about generating and editing images with MiniMax H3 Image

MiniMax H3 Image is the image generation and editing model family from MiniMax. It creates images from text prompts and edits reference images from natural-language instructions, with a particular strength in precise, localized edits and multi-image composition with up to 16 references. On this page both abilities run in one workspace: the same prompt box, aspect ratios, and output formats serve the Text to Image and Image to Image tabs.