README.md
| 1 | --- |
| 2 | pipeline_tag: text-to-image |
| 3 | license: other |
| 4 | license_name: minimax-h3-image-community-license-agreement |
| 5 | license_link: LICENSE |
| 6 | library_name: minimax-h3 |
| 7 | language: |
| 8 | - en |
| 9 | tags: |
| 10 | - text-to-image |
| 11 | - image-to-image |
| 12 | - image-editing |
| 13 | - image-generation |
| 14 | - minimax-h3 |
| 15 | - diffusers |
| 16 | base_model: |
| 17 | - MiniMaxAI/MiniMax-H3 |
| 18 | --- |
| 19 | |
| 20 | <div align="center"> |
| 21 | <img width="100%" src="assets/minimax-h3-image.jpg" alt="MiniMax H3 image, Open Weights"> |
| 22 | </div> |
| 23 | |
| 24 | <p align="center"> |
| 25 | <a href="https://hailuoai.video" target="_blank"><img src="https://img.shields.io/badge/Hailuo%20AI-FF6C37?logo=minimax&logoColor=white" alt="Hailuo AI"></a> |
| 26 | <a href="https://platform.minimax.io" target="_blank"><img src="https://img.shields.io/badge/API-FF6C37?logo=minimax&logoColor=white" alt="API"></a> |
| 27 | <a href="https://www.minimax.io" target="_blank"><img src="https://img.shields.io/badge/MiniMax%20Website-FF6C37?logo=minimax&logoColor=white" alt="MiniMax Website"></a> |
| 28 | <a href="https://github.com/MiniMax-AI/MiniMax-H3-Image" target="_blank"><img src="https://img.shields.io/badge/GitHub-181717?logo=github&logoColor=white" alt="GitHub"></a> |
| 29 | <a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image" target="_blank"><img src="https://img.shields.io/badge/Hugging%20Face-FFD21E?logo=huggingface&logoColor=black" alt="Hugging Face"></a> |
| 30 | <br> |
| 31 | <a href="https://modelscope.cn/models/MiniMax/MiniMax-H3-Image" target="_blank" rel="noopener noreferrer"><img alt="ModelScope MiniMax AI" src="https://img.shields.io/badge/ModelScope-MiniMax%20AI-white?labelColor=%23EF3D5D"></a> |
| 32 | <a href="https://platform.minimax.cn/docs/faq/contact-us" target="_blank"><img src="https://img.shields.io/badge/WeChat-07C160?logo=wechat&logoColor=white" alt="WeChat"></a> |
| 33 | <a href="https://discord.com/invite/dbMxutw7tP" target="_blank"><img src="https://img.shields.io/badge/Discord-5865F2?logo=discord&logoColor=white" alt="Discord"></a> |
| 34 | <a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/LICENSE"><img src="https://img.shields.io/badge/LICENSE-4CAF50?logo=creativecommons&logoColor=white" alt="LICENSE"></a> |
| 35 | </p> |
| 36 | |
| 37 | # MiniMax-H3-Image |
| 38 | |
| 39 | ## Online API |
| 40 | |
| 41 | Use MiniMax-H3-Image directly via API. |
| 42 | |
| 43 | - Global: [platform.minimax.io](https://platform.minimax.io/docs/guides/image-generation) | CN: [platform.minimax.cn](https://platform.minimax.cn/docs/guides/image-generation) |
| 44 | |
| 45 | API docs: |
| 46 | - Text-to-image: [EN-docs](https://platform.minimax.io/docs/api-reference/image-generation-h3-generations), [CN-docs](https://platform.minimax.cn/docs/api-reference/image-generation-h3-generations) |
| 47 | - Image editing: [EN-docs](https://platform.minimax.io/docs/api-reference/image-generation-h3-edits), [CN-docs](https://platform.minimax.cn/docs/api-reference/image-generation-h3-edits) |
| 48 | - Prompt expansion: [EN-docs](https://platform.minimax.io/docs/api-reference/image-generation-h3-context-ir), [CN-docs](https://platform.minimax.cn/docs/api-reference/image-generation-h3-context-ir) |
| 49 | |
| 50 | ## Online App |
| 51 | |
| 52 | Use MiniMax-H3-Image directly via App. |
| 53 | |
| 54 | - WebApp Global: [hailuoai.video](https://hailuoai.video/) | CN: [hailuoai.com](https://hailuoai.com/) |
| 55 | - Desktop Global: [agent.minimax.io](https://agent.minimax.io/) | CN: [agent.minimax.cn](https://agent.minimax.cn) |
| 56 | |
| 57 | ## Introduction |
| 58 | |
| 59 | MiniMax-H3-Image is a state-of-the-art open-source model for image generation and editing from |
| 60 | the [MiniMax H3](https://huggingface.co/MiniMaxAI/MiniMax-H3) family. It combines intelligent |
| 61 | understanding of creative intent with the ability to build visual worlds: with precise |
| 62 | instruction following, it develops a creative concept into concrete characters, scenes and |
| 63 | atmosphere. Generation and editing are unified in one checkpoint, which generates images from |
| 64 | text (text-to-image) and edits one or more reference images (image editing), giving creators |
| 65 | and developers a flexible and controllable way to create images. |
| 66 | |
| 67 | It generates images of about 16 megapixels at every aspect ratio, from 4096x4096 (1:1) to |
| 68 | 6272x2688 (21:9), directly and without an upscaler (see [Sizes](#sizes)). The checkpoint is |
| 69 | distilled to 6 sampling steps (6 NFEs) with no classifier-free guidance. Its diffusion |
| 70 | transformer (DiT) has the H3 architecture without the audio layers, and it uses the |
| 71 | MiniMax-H3 text encoder unchanged. |
| 72 | |
| 73 |  |
| 74 | |
| 75 | **Highlights** |
| 76 | |
| 77 | MiniMax-H3-Image brings together three core strengths: speed, image quality, and instruction |
| 78 | following. |
| 79 | |
| 80 | 1. **16-megapixel images at a practical computational cost.** Compared with the 1–4 MP outputs |
| 81 | common today, 16 MP (4096 × 4096) images offer far more detail but typically require |
| 82 | substantially more computation. MiniMax-H3-Image keeps that computational cost manageable, |
| 83 | making 16 MP generation practical for everyday creation. |
| 84 | 2. **Rich detail. Remarkable realism.** MiniMax-H3-Image goes beyond the familiar “AI look” to |
| 85 | produce images with intricate detail and lifelike realism, opening up more possibilities for |
| 86 | creative and professional use. It brings generated imagery closer to the richness and nuance |
| 87 | of the real visual world. |
| 88 | 3. **Stronger instruction following.** Greater creative control. Great image generation goes |
| 89 | beyond beautiful pictures—it helps creators bring their ideas to life. MiniMax-H3-Image |
| 90 | delivers the strongest instruction following among open-source models and supports up to 16 |
| 91 | reference images, giving creators more control over the final result. |
| 92 | |
| 93 | ## Model Zoo |
| 94 | |
| 95 | | Model | Role | Precision | Size | Hugging Face | |
| 96 | | --- | --- | --- | --- | --- | |
| 97 | | **MiniMax-H3-Image** | Text-to-image and image editing | BF16 (DiT) | 68.65 GB | [MiniMaxAI/MiniMax-H3-Image](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image) | |
| 98 | | **MiniMax-H3-Image-PE** (optional) | Prompt expander for text-to-image | NVFP4 | 17.9 GB | [MiniMaxAI/MiniMax-H3-Image-PE](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image-PE) | |
| 99 | |
| 100 | ## Model Architecture |
| 101 | |
| 102 | <!-- TODO(release): architecture figure, e.g. assets/architecture.png, adapted from MiniMax-H3's assets/full-arch.png without the audio branch. --> |
| 103 | |
| 104 | MiniMax-H3-Image follows the H3-Base design with the audio branch removed. The text |
| 105 | prompt and the reference images are encoded by the H3 text encoder; the reference images are |
| 106 | also encoded by the image VAE. All tokens form one packed sequence for a single-stream |
| 107 | transformer, which predicts the latent of the target image. The VAE decodes that latent |
| 108 | into the image. |
| 109 | |
| 110 | - **Text encoder**: the full Qwen3-VL text encoder of MiniMax-H3, unchanged. The DiT reads |
| 111 | the hidden states of its 50th layer. |
| 112 | - **Image VAE**: a CNN encoder and a 2D ViT decoder (24 layers), with 16x spatial |
| 113 | downsampling and 24 latent channels. The DiT patchifies the latent with a 2x2 patch, so each |
| 114 | transformer token covers 32x32 pixels. |
| 115 | - **Diffusion transformer**: 33.1B parameters, a dense single-stream transformer with 50 |
| 116 | layers (hidden size 5376, 56 attention heads, FFN size 14336) and 3D multimodal RoPE, the same |
| 117 | as the H3-Omni-Transformer. The weights are BF16; the embedding, timestep and output layers |
| 118 | are FP32. |
| 119 | |
| 120 | ## Quick Start |
| 121 | |
| 122 | ### Download |
| 123 | |
| 124 | | Component | Content | Size | |
| 125 | | --- | --- | --- | |
| 126 | | `transformer/` | DiT: BF16, with the embedding, timestep and output layers in FP32. | 66.28 GB | |
| 127 | | `pdd_fused_heads.safetensors`, `pdd_config.json` | The per-step output heads that replace the transformer's output layer, and the sampling schedule. | 12 MB | |
| 128 | | `vae/` | Image VAE, FP32. | 2.34 GB | |
| 129 | | `tokenizer/`, `processor/` | Qwen3-VL tokenizer and processor, identical to MiniMax-H3. | 23 MB | |
| 130 | | Text encoder | Not in this repository. `model_index.json` loads it from [`MiniMaxAI/MiniMax-H3`](https://huggingface.co/MiniMaxAI/MiniMax-H3), folder `text_encoder/`. If you already have MiniMax-H3, it is not downloaded again. | 66.7 GB | |
| 131 | |
| 132 | ```text |
| 133 | MiniMax-H3-Image/ |
| 134 | ├── model_index.json |
| 135 | ├── pdd_config.json |
| 136 | ├── pdd_fused_heads.safetensors |
| 137 | ├── processor/ |
| 138 | ├── tokenizer/ |
| 139 | ├── transformer/ |
| 140 | └── vae/ |
| 141 | ``` |
| 142 | |
| 143 | Download this repository and the MiniMax-H3 text encoder into the Hugging Face cache: |
| 144 | |
| 145 | ```bash |
| 146 | hf download MiniMaxAI/MiniMax-H3-Image |
| 147 | hf download MiniMaxAI/MiniMax-H3 --include "text_encoder/*" |
| 148 | # optional prompt expander |
| 149 | hf download MiniMaxAI/MiniMax-H3-Image-PE |
| 150 | ``` |
| 151 | |
| 152 | ### Local Deployment |
| 153 | |
| 154 | We recommend the following inference frameworks to serve the model: |
| 155 | |
| 156 | - [SGLang](https://docs.sglang.io/) - see [cookbook](https://docs.sglang.io/cookbook/diffusion/MiniMax/MiniMax-H3-Image) |
| 157 | - [vLLM-Omni](https://github.com/vllm-project/vllm-omni) - see [vllm-omni recipe](https://recipes.vllm.ai/MiniMaxAI/MiniMax-H3-Image) |
| 158 | - [diffusers](https://github.com/huggingface/diffusers) - see [diffusers docs](https://huggingface.co/docs/diffusers/main/en/api/pipelines/minimax_h3_image) |
| 159 | - [LightX2V](https://github.com/ModelTC/LightX2V) - see [Usage guide](https://github.com/ModelTC/LightX2V/tree/main/scripts/minimax_h3_image) |
| 160 | - [DiffSynth-Studio](https://github.com/modelscope/DiffSynth-Studio) - see [DiffSynth docs](https://diffsynth-studio-doc.readthedocs.io/en/latest/Model_Details/MiniMax-H3-Image.html); try low-vram inference & training |
| 161 | - [ComfyUI](https://github.com/Comfy-Org/ComfyUI) - see [Comfy tutorial](https://docs.comfy.org/tutorials/image/minimax/minimax-h3-image); use [T2I template](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/image_minimax_h3_image_t2i.json) / [Edit template](https://github.com/Comfy-Org/workflow_templates/blob/main/templates/image_minimax_h3_image_edit.json) |
| 162 | |
| 163 | #### vLLM-Omni Deployment |
| 164 | |
| 165 | See the [vLLM-Omni recipe](https://recipes.vllm.ai/MiniMaxAI/MiniMax-H3-Image) for |
| 166 | per-hardware deployment configurations and measured latency. Both NVIDIA and AMD accelerators are |
| 167 | supported, including NVIDIA H200, RTX PRO 6000 Blackwell, AMD Instinct MI350X and more. |
| 168 | |
| 169 | Start the server on one GPU: |
| 170 | |
| 171 | ```bash |
| 172 | vllm serve MiniMaxAI/MiniMax-H3-Image --omni --host 0.0.0.0 --port 8000 |
| 173 | ``` |
| 174 | |
| 175 | For lower latency, split each request over several GPUs with sequence parallelism |
| 176 | (Ulysses), for example on 4 GPUs: |
| 177 | |
| 178 | ```bash |
| 179 | vllm serve MiniMaxAI/MiniMax-H3-Image --omni --host 0.0.0.0 --port 8000 \ |
| 180 | --num-gpus 4 --usp 4 --text-encoder-tp-size 4 |
| 181 | ``` |
| 182 | |
| 183 | #### SGLang Deployment |
| 184 | |
| 185 | Here we use SGLang as a deployment example. See the [MiniMax-H3-Image deployment guide](https://docs.sglang.io/cookbook/diffusion/MiniMax/MiniMax-H3-Image) for |
| 186 | additional deployment configurations. |
| 187 | |
| 188 | Start the server on one GPU: |
| 189 | |
| 190 | ```bash |
| 191 | sglang serve --model-path MiniMaxAI/MiniMax-H3-Image --host 0.0.0.0 --port 30000 --warmup-mode off |
| 192 | ``` |
| 193 | |
| 194 | To use a local copy of the MiniMax-H3 `text_encoder/` folder instead of the Hugging Face |
| 195 | cache, add `--text-encoder-path <path to text_encoder>`. |
| 196 | |
| 197 | For lower latency, split each request over several GPUs with sequence parallelism |
| 198 | (Ulysses), for example on 4 GPUs: |
| 199 | |
| 200 | ```bash |
| 201 | sglang serve --model-path MiniMaxAI/MiniMax-H3-Image --host 0.0.0.0 --port 30000 --warmup-mode off \ |
| 202 | --num-gpus 4 --ulysses-degree 4 |
| 203 | ``` |
| 204 | |
| 205 | #### Request parameters |
| 206 | |
| 207 | Text-to-image requests are JSON sent to `/v1/images/generations`. Image editing requests send |
| 208 | the reference images as multipart form data to `/v1/images/edits`; their order defines |
| 209 | `<Picture 1>`, `<Picture 2>`, and so on in the prompt. Both return the PNG in |
| 210 | `data[0].b64_json`. Runnable examples are in [Full Workflow](#full-workflow). |
| 211 | |
| 212 | | Field | Text-to-image | Image editing | Notes | |
| 213 | | --- | --- | --- | --- | |
| 214 | | `prompt` | Required | Required | Used as given, truncated to 4096 tokens. See [Full Workflow](#full-workflow) | |
| 215 | | `size` | Required | Required | `WIDTHxHEIGHT`. The returned image has exactly this size; see [Sizes](#sizes) | |
| 216 | | `image` | — | One or more files | In upload order they are `<Picture 1>`, `<Picture 2>`, ... in the prompt | |
| 217 | | `seed` | Recommended | Recommended | Integer | |
| 218 | | `generator_device` | `cpu` recommended | `cpu` recommended | Same seed, same initial noise on every machine | |
| 219 | | `response_format` | `b64_json` | `b64_json` | | |
| 220 | | `min_generation_area` | Optional | Optional | Minimum generated area in pixels, 1024-16777216, default 4194304 (2048x2048). A smaller `size` is generated at the same aspect ratio on a larger canvas and resized back with LANCZOS; pass 1024 to turn this off. All sizes in the table below are above the default | |
| 221 | | `num_inference_steps` | Omit or 6 | Omit or 6 | Other values return HTTP 400 | |
| 222 | | `guidance_scale` | Omit or 1 | Omit or 1 | Other values return HTTP 400 | |
| 223 | | `negative_prompt` | Not supported | Not supported | Rejected | |
| 224 | |
| 225 | #### Sizes |
| 226 | |
| 227 | The effective working range of MiniMax-H3-Image is about 4 MP to 20 MP, and it generally gives |
| 228 | the best results at around 16 MP. Common 16 MP sizes are listed below: |
| 229 | |
| 230 | | Aspect ratio | Size (width x height) | |
| 231 | | --- | --- | |
| 232 | | 1:1 | 4096x4096 | |
| 233 | | 4:3 / 3:4 | 4736x3552 / 3552x4736 | |
| 234 | | 3:2 / 2:3 | 5024x3360 / 3360x5024 | |
| 235 | | 16:9 / 9:16 | 5472x3072 / 3072x5472 | |
| 236 | | 21:9 | 6272x2688 | |
| 237 | |
| 238 | ### Full Workflow |
| 239 | |
| 240 | MiniMax-H3-Image expects a detailed English description of the whole image. The MiniMax API |
| 241 | turns a short prompt in any language into such a description with **H3-Image-Context-IR** before |
| 242 | generation. We also provide |
| 243 | [MiniMax-H3-Image-PE](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image-PE), which can be |
| 244 | deployed locally. |
| 245 | |
| 246 | This section reproduces the quality of images generated directly by the MiniMax API with a |
| 247 | locally deployed SGLang server. The scripts are in |
| 248 | [`examples/`](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/tree/main/examples) and use only |
| 249 | the Python standard library; run them in that folder: |
| 250 | |
| 251 | ```bash |
| 252 | hf download MiniMaxAI/MiniMax-H3-Image --include "examples/*" --local-dir . |
| 253 | cd examples |
| 254 | ``` |
| 255 | |
| 256 | Before running examples, set your MiniMax API endpoint and key to use the H3-Image-Context-IR API |
| 257 | and the reference scripts. The scripts read them from the environment: |
| 258 | |
| 259 | ```bash |
| 260 | # MiniMax API endpoint (choose one) |
| 261 | # CN |
| 262 | export MINIMAX_API_BASE="https://api.minimax.cn" |
| 263 | # Global |
| 264 | # export MINIMAX_API_BASE="https://api.minimax.io" |
| 265 | |
| 266 | # API key obtained from the MiniMax platform |
| 267 | export MINIMAX_API_KEY="<api-key>" |
| 268 | ``` |
| 269 | |
| 270 | API docs: |
| 271 | - H3-Image-Context-IR: use /v1/h3_image_context_ir [EN-docs](https://platform.minimax.io/docs/api-reference/image-generation-h3-context-ir), [CN-docs](https://platform.minimax.cn/docs/api-reference/image-generation-h3-context-ir) |
| 272 | - Text-to-image: use /v1/images/generations [EN-docs](https://platform.minimax.io/docs/api-reference/image-generation-h3-generations), [CN-docs](https://platform.minimax.cn/docs/api-reference/image-generation-h3-generations) |
| 273 | - Image editing: use /v1/images/edits [EN-docs](https://platform.minimax.io/docs/api-reference/image-generation-h3-edits), [CN-docs](https://platform.minimax.cn/docs/api-reference/image-generation-h3-edits) |
| 274 | |
| 275 | #### case-T2I |
| 276 | |
| 277 | [`instruction.t2i.txt`](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/instruction.t2i.txt). |
| 278 | |
| 279 | <table> |
| 280 | <thead> |
| 281 | <tr><th colspan="2">Stage</th><th>Request</th><th>Result</th></tr> |
| 282 | </thead> |
| 283 | <tbody> |
| 284 | <tr><td rowspan="2">Prompt Enhancement</td><td>H3-Image-Context-IR (MiniMax API)</td><td><code>python3 context_ir.py t2i</code> (<a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/context_ir.py">script</a>)</td><td><code>revised_prompt.txt</code>, <code>revised_size.txt</code>; our run: <a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/revised_prompt.t2i.txt"><code>revised_prompt.t2i.txt</code></a>, <a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/revised_size.t2i.txt"><code>revised_size.t2i.txt</code></a> (4736x3552)</td></tr> |
| 285 | <tr><td>local alternative: <a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image-PE">MiniMax-H3-Image-PE</a></td><td><code>scripts/expand_prompt.py</code> in the PE repository</td><td>Save <code>expanded_caption</code> as <code>revised_prompt.txt</code> and the size for <code>aspect_ratio</code> as <code>revised_size.txt</code></td></tr> |
| 286 | <tr><td colspan="2">MiniMax-H3-Image generation: local SGLang</td><td><code>python3 t2i.py --prompt revised_prompt.txt --size $(cat revised_size.txt)</code> (<a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/t2i.py">script</a>)</td><td><code>t2i.png</code></td></tr> |
| 287 | <tr><td colspan="2">E2E API Reference: text-to-image by directly calling the MiniMax API</td><td><code>python3 api_generate.py t2i</code> (<a href="https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/api_generate.py">script</a>)</td><td><code>api_t2i.png</code></td></tr> |
| 288 | </tbody> |
| 289 | </table> |
| 290 | |
| 291 | <p align="center"> |
| 292 | <img src="assets/workflow_t2i.jpg" alt="Left: t2i.png from local SGLang with the enhanced prompt. Right: api_t2i.png from the MiniMax API. Both show an overhead farm-to-table feast with bread, vegetables and cheeses on a weathered wooden table" width="100%"/> |
| 293 | </p> |
| 294 | <p align="center"><em>t2i.png from local SGLang (left) and api_t2i.png from the MiniMax API (right), 4736x3552 (4:3 picked by H3-Image-Context-IR)</em></p> |
| 295 | |
| 296 | #### case-Edit |
| 297 | |
| 298 | [`instruction.edit.txt`](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/instruction.edit.txt) |
| 299 | with the reference image [`reference1.png`](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/reference1.png) |
| 300 | as `<Picture 1>`. |
| 301 | |
| 302 | | Stage | Request | Result | |
| 303 | | --- | --- | --- | |
| 304 | | Prompt Enhancement: H3-Image-Context-IR (MiniMax API) | `python3 context_ir.py edit` ([script](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/context_ir.py)) | `revised_prompt.txt`, `revised_size.txt`; our run: [`revised_prompt.edit.txt`](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/revised_prompt.edit.txt), [`revised_size.edit.txt`](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/revised_size.edit.txt) (4096x4096) | |
| 305 | | MiniMax-H3-Image generation: local SGLang | `python3 edit.py --prompt revised_prompt.txt --size $(cat revised_size.txt)` ([script](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/edit.py)) | `edit.png` | |
| 306 | | E2E API Reference: image editing by directly calling the MiniMax API | `python3 api_generate.py edit` ([script](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/examples/api_generate.py)) | `api_edit.png` | |
| 307 | |
| 308 | <p align="center"> |
| 309 | <img src="assets/workflow_edit.jpg" alt="Left: reference1.png, the reference image of two podcast hosts with the man wearing headphones. Middle: edit.png from local SGLang with the enhanced prompt. Right: api_edit.png from the MiniMax API. In both results the headphones are gone and the hosts bump fists above the MONEY TALKS sign" width="100%"/> |
| 310 | </p> |
| 311 | <p align="center"><em>Reference image reference1.png (left), edit.png from local SGLang (middle) and api_edit.png from the MiniMax API (right), 4096x4096 (1:1 picked by H3-Image-Context-IR)</em></p> |
| 312 | |
| 313 | ## Showcase |
| 314 | |
| 315 | Text-to-image and image editing results, shown downscaled. |
| 316 | |
| 317 | <p align="center"> |
| 318 | <img src="assets/showcase/photo.jpg" alt="Three photorealistic photographs: a woman hanging white sheets on a rooftop at dusk under a full moon, a close-up profile of a curly-haired young man holding a cup, and flamingos wading in shallow water" width="100%"/> |
| 319 | </p> |
| 320 | <p align="center"><em>Photorealistic photography</em></p> |
| 321 | |
| 322 | <p align="center"> |
| 323 | <img src="assets/showcase/surreal.jpg" alt="Three surreal editorial photographs: a barber trimming a cloud seated in a barber chair, a person slicing the moon on a lawn at night, and a model stepping out of a giant pear" width="100%"/> |
| 324 | </p> |
| 325 | <p align="center"><em>Surreal editorial photography</em></p> |
| 326 | |
| 327 | <p align="center"> |
| 328 | <img src="assets/showcase/scene.jpg" alt="Four environment concepts: ruins of an ancient open-air theater in a forest titled ENVIRONMENT CONCEPT: THE ANCIENT THEATER, a dragon rider on a red rock cliff, an abandoned railway station with an ARRIVALS sign, and shipwrecks frozen in an icy sea" width="100%"/> |
| 329 | </p> |
| 330 | <p align="center"><em>Environment concepts</em></p> |
| 331 | |
| 332 | <p align="center"> |
| 333 | <img src="assets/showcase/char.jpg" alt="Characters, creatures and props: a soldier in a worn helmet and armor, a red-finned dragon-like creature in a forest, a weathered astronaut helmet on Mars, and a three-view character sheet of a figure in winter clothing" width="100%"/> |
| 334 | </p> |
| 335 | <p align="center"><em>Characters, creatures and props</em></p> |
| 336 | |
| 337 | <p align="center"> |
| 338 | <img src="assets/showcase/art.jpg" alt="Four paintings and illustrations: mandarin ducks in a lotus pond in an oil painting style, a Jiangnan garden in light watercolor, tarot card XVIII THE MOON with a rabbit, and a Chinese gongbi painting of a pelican looking into a mirror with four calligraphy characters" width="100%"/> |
| 339 | </p> |
| 340 | <p align="center"><em>Painting and illustration styles</em></p> |
| 341 | |
| 342 | <p align="center"> |
| 343 | <img src="assets/showcase/text.jpg" alt="Two logic puzzle boards: a Chinese candidate-selection puzzle with five conditions and the derived answer, and an English board titled CONFIDENTIAL FILE MISSING with six suspects, true/false stamps and a verification table" width="100%"/> |
| 344 | </p> |
| 345 | <p align="center"><em>Text rendering with reasoning: logic puzzles in Chinese and English, solved correctly on the board</em></p> |
| 346 | |
| 347 | <p align="center"> |
| 348 | <img src="assets/showcase/edit_render.jpg" alt="Left: untextured grey 3D render of a lunar base with habitat modules, a rover and an astronaut. Right: the same scene as a photorealistic film still on the lunar surface" width="100%"/> |
| 349 | </p> |
| 350 | <p align="center"><em>Image editing, one reference: an untextured 3D render (left) turned into a film still of a lunar research base (right)</em></p> |
| 351 | |
| 352 | <p align="center"> |
| 353 | <img src="assets/showcase/edit_map.jpg" alt="Left: illustrated game map with a river, a waterfall, bridges, a castle, a village, a mine and a harbor, and a red arrow on the southern lake. Right: first-person 3D game view from the lake towards the waterfall, the suspension bridge, the mine and the distant castle, with the map as a minimap in the top-left corner" width="100%"/> |
| 354 | </p> |
| 355 | <p align="center"><em>Image editing, one reference: a game map (left) turned into a first-person 3D game view from the red arrow's position, with the map as a minimap (right)</em></p> |
| 356 | |
| 357 | <p align="center"> |
| 358 | <img src="assets/showcase/edit_multi.jpg" alt="Top: ten reference images, an empty sunlit room and nine pieces of furniture. Bottom: the empty room furnished with all nine pieces as a living room" width="100%"/> |
| 359 | </p> |
| 360 | <p align="center"><em>Image editing, ten references: an empty room and nine pieces of furniture (top) composed into one living room (bottom)</em></p> |
| 361 | |
| 362 | ## License |
| 363 | |
| 364 | - MiniMax-H3-Image is released under the [MiniMax-H3-Image Community License Agreement](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/LICENSE). |
| 365 | - The T2I PE Model ([MiniMax-H3-Image-PE](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image-PE)) is part of MiniMax-H3-Image and is accordingly subject to [MiniMax-H3-Image Community License Agreement](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image/blob/main/LICENSE). |
| 366 | - The separate written authorization described in Section IV of the license is requested through [Application form](https://platform.minimax.io/console/license) |
| 367 | |
| 368 | ## Contact Us |
| 369 | |
| 370 | Contact us at [model@minimax.io](mailto:model@minimax.io). |
| 371 | |