Text-to-Image
Turn a text prompt into a detailed image.
AI image generator & editor
MiniMax H3 Image Generator
Create detailed images from text, transform existing visuals with natural-language edits, and bring multiple references together with MiniMax H3 Image.
Explore high-resolution image generation, rich visual details, and more control over composition, style, and spatial relationships.
- Text-to-Image
- Image Editing
- Up to 16 Reference Images
- 1K / 2K / 4K Output Options
Describe the subject, setting, composition, lighting, and style you want.
Try an example
Your images
Log in to generate and see your recent images.
Each run keeps its settings, so you can review the details and refine the prompt.Create images from text or reference photos
One Model for Image Creation and Editing
Start with a text prompt or upload reference images to guide your result. MiniMax H3 Image brings image generation and editing into one workflow, helping you turn creative ideas into detailed visuals.
Describe the subject, composition, lighting, materials, and style you have in mind. When editing, explain what should change and which parts of the original image should remain consistent.
Image Editing
Edit an existing image with natural-language instructions.
Create more with MiniMax H3 Image
What Makes MiniMax H3 Image Different?
Generate Images from Text
Turn a written idea into a complete visual. Describe your subject, setting, camera angle, lighting, and artistic direction to create images for personal projects, product concepts, and creative work.
Edit Images with Natural-Language Instructions
Start with an existing image and describe the changes you want. Replace an object, explore a new visual style, adjust the setting, or develop a new composition from your reference.
Combine Up to 16 Reference Images
Bring multiple visual references into one workflow. Use one image for the subject, another for the visual style, and others for the layout, colors, or environment. Explain the role of each reference to guide the final composition.
Create High-Resolution Images
Choose from 1K, 2K, and 4K output-area settings. Create visuals for different layouts, from portrait compositions to wide landscape scenes.
Follow Detailed Creative Instructions
Give your image a clear structure. Describe where objects belong, how they relate to one another, which perspective to use, and what details matter most. H3 Image is designed to follow complex visual instructions and spatial relationships.
Explore Rich Textures and Realistic Details
Create visuals with natural lighting, believable shadows, and detailed materials. Explore product surfaces, fabric, fur, skin, metal, environments, and other visual textures.
Use cases
What Can You Create with H3 Image?
Pick a direction and load a matching prompt into the generator.
Product Photography and Product Concepts
Create product-focused visuals, lifestyle scenes, material studies, and concept renders. Use reference images to guide the product's appearance and the environment around it.
Character Design and Game Art
Explore character concepts, creature designs, game key art, environment concepts, and visual development sheets. Use references to guide the subject, style, and composition.
Posters, Covers, and Social Graphics
Develop poster concepts, cover artwork, campaign visuals, and social media graphics. Keep important text short and review every label before publishing.
Infographics and Explanatory Visuals
Turn structured ideas into diagrams, labeled illustrations, and visual explanations. For dense or long-form copy, add final text in a design editor after image generation.
Interior and Scene Concepts
Combine room references, furniture images, materials, and lighting directions to explore new interior compositions and spatial ideas.
Prompt examples
MiniMax H3 Image Prompt Examples
Pick an example and load it into the generator. These prompts were written to show what the model can do, so adjust them to fit your own idea.
Product Photography
Create a premium product photograph of a matte black insulated water bottle on a light travertine pedestal. Use a clean studio background, soft directional daylight, subtle natural shadows, realistic brushed-metal details, and a refined modern composition. Show the entire bottle clearly. No extra products, no watermark, no decorative text.
Cinematic Game Key Art
Create cinematic fantasy game key art showing a lone explorer standing on a rocky ridge above an ancient city. The city fills the distant valley, with layered mountains fading into the atmosphere. Use a wide composition, dramatic but natural late-afternoon light, detailed stone and fabric textures, and a polished AAA game concept-art finish. Keep the explorer clearly readable against the environment.
Character Design Sheet
Create a clean character design sheet for the same original fantasy courier. Show front, side, and back views in a consistent neutral standing pose. Keep the character's face, hairstyle, clothing, accessories, body proportions, and color palette consistent across all views. Use a plain light background, even studio lighting, and clear spacing between views.
Multi-Reference Composition
Use image 1 as the room layout and camera perspective. Use images 2–4 as the furniture references. Place the referenced furniture naturally in the room while preserving each item's recognizable shape, color, and material. Match the lighting and perspective to image 1. Do not add unrelated furniture or change the room architecture.
Image Editing
Replace the small table in image 1 with the wooden side table shown in image 2. Keep the sofa, wall, floor, camera angle, room layout, and overall lighting unchanged. Match the new table's scale, perspective, contact shadow, and color temperature to the original scene.
Educational Graphic
Create a clean educational infographic explaining the water cycle. Show evaporation, condensation, precipitation, and collection as four clearly connected stages. Use a simple readable layout, distinct visual sections, concise English labels, clear arrows, and a friendly science-education illustration style. Keep each label short and avoid unnecessary decorative elements.
Wide Editorial Banner
Create a wide editorial banner for a modern travel magazine. Show a quiet coastal village built along steep green cliffs above a deep blue sea. Use a panoramic composition, soft morning haze, natural colors, realistic architecture, and detailed water reflections. Leave uncluttered negative space on the left for a headline to be added later. Do not render any text.
How it works
How to Use MiniMax H3 Image
- 01
Describe Your Idea
Enter a prompt describing the subject, setting, composition, style, and visual details you want.
- 02
Add Reference Images (Optional)
For image editing, upload one or more reference images. Tell the model what each image should contribute, such as the subject, color palette, layout, or background.
- 03
Choose Your Output Settings
Select an available resolution, aspect ratio, and number of images based on the options provided by this tool.
- 04
Generate and Review
Generate your image, review the details, and refine the prompt when needed. For edits, check that the important elements remain as intended.
- 05
Download Your Result
Save the generated image in one of the supported output formats.
Prompt guide
Tips for Writing Better H3 Image Prompts
A useful prompt explains more than the subject. Include the visual information that matters to the final result:
- Subject
- Who or what should appear?
- Setting
- Where does the scene take place?
- Composition
- What should be in the foreground, middle ground, and background?
- Camera and perspective
- Is the view close-up, wide-angle, eye-level, overhead, or first-person?
- Lighting
- Should the scene use soft daylight, studio lighting, sunset light, or dramatic shadows?
- Materials and texture
- Should surfaces look like brushed metal, natural fabric, glass, stone, or polished wood?
- Style
- Should the image look like product photography, editorial photography, concept art, illustration, or a 3D render?
- Constraints
- What must remain visible, and what should be avoided or kept unchanged?
For image editing, be specific about both the requested change and the elements to preserve. For example: “Replace the bag with the reference product. Keep the model, pose, camera angle, lighting, and background unchanged.” The instruction guides the edit, but the final image should still be reviewed.
Prompt templates
Fill in the brackets, then load the template into the generator.
Create [subject] in [environment or setting]. Composition: [framing, subject placement, foreground/background]. Camera: [viewpoint, angle, lens or perspective]. Lighting: [light source, direction, mood]. Materials and details: [textures, colors, important features]. Visual style: [photography, illustration, concept art, 3D render, etc.]. Requirements: [important details to include or avoid].
Use image 1 as the source image. Change: [describe the exact element or area to modify]. Reference: Use image 2 for [subject / product / style / color / texture]. Preserve: [list the subject, pose, background, camera, lighting, layout, or other elements to keep]. Integration: Match scale, perspective, lighting, shadows, and color temperature. Avoid: [unwanted changes or extra elements].
Image 1: Use as the main composition and camera reference. Image 2: Use as the subject reference. Image 3: Use as the color palette and material reference. Image 4: Use as the environment reference. Combine these references into one coherent image. Preserve the specified characteristics of each reference and follow the requested spatial arrangement.
- 1
Say what is in the image
Name the subject, the scene, and the key details instead of asking for something “beautiful”.
- 2
Say where things go
For complex compositions, describe the foreground, the background, left and right positions, the viewpoint, and how objects relate to one another.
- 3
Say what must not change
When editing, use clear instructions such as “Change only…” and “Keep… unchanged”, then review the result.
Specs
Technical Specifications
The settings available when you use MiniMax H3 Image in this tool.
- Model
- MiniMax-H3-ImageSelected by default. GPT Image 2.6 Flare and Sunburst are available in the model menu.
- Modes
- Text-to-image and image editingCreate a new image from a prompt, or edit with reference images.
- Reference images
- 1–16 per editExplain what each image contributes: subject, style, layout, colors, or environment.
- Reference files
- JPG, PNG or WEBP · up to 20 MB each128–8192 px on each side.
- Output settings
- 1K, 2K or 4KOutput-area settings. The final width and height depend on the aspect ratio you choose.
- Aspect ratio
- Auto, or 1:3 to 3:1Portrait, square, landscape, and wide frames.
- Images per run
- 1–4Credits are charged per image.
- Output formats
- JPEG, PNG or WebPJPEG is the default.
- Prompt length
- Up to 4,000 charactersPrompt enhancement adds detail to short ideas.
- Not supported
- Masked edits, transparent backgroundsDescribe edits in natural language. Exporting to PNG does not create transparency.
Models
MiniMax H3 Image vs GPT Image 2.6
All three models create and edit images in the same generator. Credits are charged per image and depend on the model and quality.
| Model | Choose it for | Reference images | Credits per image | Try |
|---|---|---|---|---|
| DefaultMiniMax H3 ImageMiniMax | Detailed images, multi-reference edits, 1K–4K output | Up to 16 | 2 · 1K, 2 · 2K, 4 · 4K | |
| GPT Image 2.6 FlareOpenAI | Fast drafts and quick edits | Up to 10 | 2 · Low, 4 · Medium, 5 · High, 10 · X-High, 20 · Max | |
| GPT Image 2.6 SunburstOpenAI | Dense text, precise layouts, careful edits | Up to 10 | 2 · Low, 4 · Medium, 5 · High, 10 · X-High, 20 · Max |
Credits
Simple credits for MiniMax H3 Image and GPT Image 2.6
Every $1 buys 20 credits. With MiniMax H3 Image, each image costs 2 credits at 1K or 2K and 4 at 4K; GPT Image 2.6 costs 2 to 20 credits depending on quality. The exact cost is shown on the Generate button, and failed runs are refunded automatically.
See plansFAQ
Frequently Asked Questions
01What is MiniMax H3 Image?
MiniMax H3 Image is an AI model for text-to-image generation and image editing. You can create a new image from a text prompt or provide reference images and describe how you want the result to change.
02Can MiniMax H3 Image generate images from text?
Yes. H3 Image supports text-to-image generation. Describe the subject, setting, composition, lighting, and style you want, then generate an image from your prompt.
03Can I edit an existing image with MiniMax H3 Image?
Yes. H3 Image supports image editing with one or more reference images. Describe the change you want and specify which parts of the original image should be preserved.
04How many reference images can I upload?
MiniMax H3 Image supports up to 16 reference images in one editing request. In this tool each reference image can be up to 20 MB, in JPG, PNG, or WEBP format. GPT Image 2.6 takes up to 10 here.
05Can I use multiple images to create one composition?
Yes. You can provide up to 16 reference images in an image-editing request. For better direction, explain what each reference contributes—for example, the subject, style, color palette, layout, or environment.
06What resolutions does MiniMax H3 Image support?
This tool offers the 1K, 2K, and 4K output-area settings with aspect ratios from 1:3 to 3:1. The final width and height depend on the aspect ratio you choose.
07What image formats can I download?
You can export MiniMax H3 Image results as JPEG, PNG, or WebP. Pick the format in the output settings before you generate.
08Does MiniMax H3 Image improve prompts automatically?
Yes. MiniMax enables prompt enhancement by default for text-to-image generation, and image-editing requests always use it. That is why you can start with a simple idea, and why the result may add details you did not write.
09Can MiniMax H3 Image generate text inside an image?
H3 Image can create titles, labels, and short annotations. For long paragraphs, dense small text, or content that must be letter-perfect, review the result carefully and consider adding the final copy in a separate design editor.
10Does MiniMax H3 Image support transparent backgrounds?
No. The current MiniMax image-generation API documentation lists transparent backgrounds as unsupported, so exporting to PNG does not create transparency.
11Does MiniMax H3 Image support masked or brush-based editing?
No. The current API documentation lists masked edits as unsupported. You can describe an edit in natural language, but this is not the same as a dedicated mask-based inpainting workflow.
12Is MiniMax H3 Image the same as the MiniMax H3 video model?
No. MiniMax H3 Image is designed for still-image generation and editing. It is related to the MiniMax H3 model family, but an image-generation request does not itself create an animated video.
13How many credits does an image cost?
MiniMax H3 Image costs 2 credits per image at 1K or 2K and 4 credits at 4K. GPT Image 2.6 costs 2 credits at low quality up to 20 at max. The exact cost is shown on the Generate button before you start.
14What happens if a generation fails?
The credits for that run are refunded to your balance automatically, including runs stopped by the content check.
15Can I use the images commercially?
Our Terms of Service explain how you may use what you make. In short, you need the rights to any images you upload, and your prompts and uploads must follow our content policy.
16Is this the official MiniMax site?
No. This is an independent site that gives you access to MiniMax H3 Image and GPT Image 2.6 through their official APIs. We are not affiliated with MiniMax or OpenAI.
MiniMax H3 Image
Bring Your Next Image Idea to Life
Start with a sentence, a sketch, a product photo, or a collection of visual references. Create a new image or explore a different direction with MiniMax H3 Image.
Independent site, not affiliated with or endorsed by MiniMax or OpenAI. Model names are trademarks of their owners and are used only to identify the models.