<h1 id="minimax-h3-prompt-enhancer">MiniMax H3 Prompt Enhancer</h1>An <a rel="ugc" href="https://ollama.com">Ollama </a>model that turns a short idea, and optionally a reference image, into a full audiovisual prompt for the MiniMax H3.H3 generates video together with sound, so a good prompt describes both what is seen and what is heard. This model writes prompts in the structure from MiniMax's official <a rel="ugc" href="https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_base_en.md">video prompt writing guide</a>: a timeline of shots with camera moves, speaker IDs and dialogue, a soundscape, and background music.It is a prompt and parameter layer on top of an existing vision model. It contains no new weights and no MiniMax code or weights.<h3 id="what-it-does">What it does</h3>Give it an idea (and attach an image if you have one), get back only the finished prompt.Task - You provide - Output starts with |Text to video - Text only - <code>integrated_multimodal_description:</code>Image to video - One image (used as the first frame) - The first-frame alignment line, then the three fieldsFirst and last frame - Two images - The first-and-last-frame alignment lineLast frame - One image, plus "ending frame" in your text - The last-frame alignment lineEvery prompt has the three core fields from the guide:- <code>integrated_multimodal_description</code>: visuals, shots, camera, speech and sounds made by things in the scene, along a timeline. - <code>overall_soundscape</code>: ambience, physical sounds, and non-verbal human sounds. - <code>non_diegetic_music</code>: background music the characters cannot hear.Behavior:- Timeline format: <code>[Shot 1]</code> has no timestamp; later shots read <code>[Shot 2] At 00:04.500, the camera cuts to ...</code> with strictly increasing times inside the duration.- Camera language: motion type, amplitude and speed, for example "The camera pushes in with small amplitude at slow speed".- Speech: speakers get IDs such as <code>(S1)</code>, and the words go in <code>&lt;d&gt;[English] ...&lt;/d&gt;</code>, kept exactly as you wrote them. Voiceover and lip handling follow the guide.- Image tasks: the style, subjects and composition come from your image, and the video develops forward from it (or lands on it, for a last frame).- People: every person is described with an apparent age and ethnicity. If you give them, they are used exactly; otherwise they come from the image or the setting.- No invented text: signs and labels appear only if you asked for them or they are in your image.- Duration: use the one you state ("8 seconds"). The default is 8 seconds. Short clips are a single shot; longer clips get cuts.### In ComfyUIUse the <code>comfyui-ollama</code> node with the model name <code>minimax-prompt-enhancer</code>. For image-to-video, connect the same image to the node and to the H3 reference input.Link: Github: <a target="_blank" rel="ugc" href="https://github.com/royen99/minimax-h3-prompt-enhancer">https://github.com/royen99/minimax-h3-prompt-enhancer</a>HuggingFace: <a target="_blank" rel="ugc" href="https://huggingface.co/DutchIT/minimax-prompt-enhancer">https://huggingface.co/DutchIT/minimax-prompt-enhancer</a>Ollama: <a target="_blank" rel="ugc" href="https://ollama.com/dutchit/minimax-h3-prompt-enhancer">https://ollama.com/dutchit/minimax-h3-prompt-enhancer</a>

Files with SHA256: 1f4a639bf864eb1b41073add29c593527d1b29d9bd7b166bb8776c01a65aaa7e