SOL-H3 is a preview of an accelerated video workflow with native audio. Public generation remains closed while the route and billing checks are prepared.
SOL-H3 / Video with native audio
This SOL-H3 video generator preview is built around native audio. Instead of writing a still image and hoping for movement, describe the scene as a short sequence: what exists, what changes, how the camera follows the action, and what the moment sounds like.
Video + native audioOne focused scene briefPublic generation pendingNo paid requests yet
SOL-H3 scene directions
These original Voor concept studies show how a SOL-H3 brief can combine picture and sound in one direction. They are storyboard studies, not rendered model outputs. Public generation is not open yet, and this page will not start a paid request.
SOL-H3 prompts work best when the scene has one subject, one visible action, and a light source that explains the mood. Keep the camera move simple enough to describe in one sentence.
“A tiny glass bird rests beside a glowing workshop furnace. A slow push-in reveals amber light refracting through its wings. Embers drift upward. Sound: a low furnace hum and delicate glass chimes.”
For a product beat, name the object, the surface, the light movement, and the sound. SOL-H3 is designed to carry picture and sound in the same brief.
“An unbranded brushed-metal watch rests on dark volcanic stone. A thin warm light crosses the bezel as the second hand begins to sweep. Sound: one precise mechanical click and quiet studio ambience.”
A vertical frame can hold an intimate object and a larger world. Describe what changes, what the camera follows, and what the audience should hear at the end.
“An old illustrated atlas opens on a wooden library desk. Tiny paper mountains rise from the page and a warm paper lighthouse begins to glow. A low camera move follows a paper river. Sound: rustling pages, tiny bell tones, soft library room tone.”
Preview status: SOL-H3 generation is not open on Voor yet. The page is designed for prompt preparation, model research, and visual direction while the provider route, output handling, pricing, and billing checks are completed.
How to write a SOL-H3 prompt
Write the scene in the order the viewer experiences it. A structured brief gives SOL-H3 a clear subject, a readable action, a camera path, and a soundscape that belongs to the picture.
Start with one subject and one action. Add the setting, the camera move, the light, and the material detail. Then write what the scene sounds like at the beginning, middle, and end.
Avoid stacking unrelated events. A focused SOL-H3 scene is easier to review, easier to revise, and easier to compare when one decision changes.
Subject + setting
Visible action from first to last frame
Camera movement and framing
Light, material, and atmosphere
Sound: ambience, effect, or dialogue
Ending beat and final imageStart with the thing the camera can follow and the place it exists in. A clear subject gives every later camera or sound choice a reference point.
Describe what changes from the first frame to the last. Keep the action readable instead of listing unrelated events that compete for attention.
Say whether the shot pushes in, tracks, arcs, pulls back, or stays fixed. The camera move tells the model how the action should be framed over time.
Write ambience, a short line of dialogue, or one precise effect. Sound should belong to the scene, not sit on top of it as an unrelated layer.
SOL-H3 use cases
SOL-H3 fits scenes where a single action, a deliberate camera move, and a clear sound cue carry the idea. Start focused, then build a longer sequence from approved beats.
A single clear beat with motion and native audio is the strongest SOL-H3 starting point.
Use macro framing, a moving light, and a precise sound cue to give an object a film-like reveal.
Let the environment transform around a small subject. Describe the camera path and the final reveal.
Build and compare scene briefs now, then run the approved direction when public generation opens.
| Decision | SOL-H3 direction | What to review |
|---|---|---|
| Scene length | One clear beat per generation | Does the action resolve? |
| Camera | One readable move | Does the frame still make sense? |
| Sound | Ambience, effect, or short dialogue | Does the sound have a visible cause? |
| Continuity | Keep identity, wardrobe, and light stable | What changed between takes? |
Live alternatives while SOL-H3 is gated: compare the MiniMax H3 and FLUX 3 routes for available video with native audio. When SOL-H3 generation opens, this page will show the same visible model and credit quote used elsewhere in the workspace.
SOL-H3 is an accelerated video generation workflow developed by NVIDIA Research. It produces video with native audio from a written scene description, including action, camera movement, and sound.
Public generation is not open yet. The Voor page is a preview of the model, examples, and prompting approach. The generator shows a coming-soon state until the live route and billing controls are verified.
Start with one subject and one visible action. Add framing, lighting, a camera move, and the sound the scene should contain. Keep the brief focused instead of listing many unrelated ideas.
Yes. SOL-H3 is designed to generate video and native audio together. Describe the ambience, a specific effect, or a short line of dialogue as part of the same scene brief.
No public date is promised. Access opens only after the provider route, output handling, pricing, and billing checks pass. Until then, the page remains a preview and cannot start a paid request.
No. Voor AI is an independent creative workspace. Model names and trademarks belong to their owners.
Public generation opens only after the route, output handling, and billing checks pass. Until then, keep a clear SOL-H3 brief ready.
Open the SOL-H3 studio previewOptional cookies help us understand usage and measure ads. Essential cookies stay on so Voor works.