Generate models
Generate a model from text, an image or multiple views, then inspect and use the result.
Choose an input
Open AI workspace → Generate → Generate model and choose an input. The result is a model asset; use Agent to create or edit a scene with several objects.
| Input | Use it when |
|---|---|
| Text | You want to describe the object, shape and material in words. |
| Image | You have one photo, reference image or line drawing. |
| Multi-view | You have consistent views of the same object, supplied in the order shown. |
From text
Use specific shapes, openings and proportions instead of vague adjectives. Revise one part of the prompt at a time.
- Choose Text.
- Describe the subject, main shapes, materials and style. State which parts belong to the object.
From one image
Surfaces hidden in the image are inferred. If you have consistent views of the same object, use Multi-view.
- Choose Image. Select Photo / Reference for a photo or rendered object, or Sketch / Line Art for a drawing.
- Show one complete object with a clear silhouette, simple background and readable lighting. A front three-quarter view often reveals more of the form.
- Upload the image. If it is rejected, follow the file validation message and choose a supported file.
From multiple views
- Choose Multi-view.
- Use the same object, pose, proportions and surface appearance in every image, with similar scale and framing. Do not mix designs or mirrored views.
- Match the named slots: Front, Back, Left, then Right. Do not skip an earlier slot.
- Some model versions provide more view slots. Follow the order and requirements shown for that version.
Settings and generation
Available options depend on the input and model version. Choose the generation mode and output type, then expand Output settings to adjust face count, materials and format. Review the displayed credit cost before generating. Submission requires sign-in.
| Controls | What changes |
|---|---|
| Pro / Rapid | Choose a supported quality mode for the current input. Compare a result before committing to more generations. |
| Model version | Controls the available inputs and output options. Changing it may reset incompatible settings. |
| Output type / face count | Choose the intended geometry output. A higher face budget is not a guarantee of a better model and costs more to handle. |
| PBR materials | Includes supported material maps. Geometry-only output does not include a textured surface. |
| Output format | Choose a format supported by the application that will use the model. |
Inspect the result
After adding the result to a scene, continue editing in Modeling and Material. Downloading a model file is separate from saving an editable scene.
- Open the finished task in History and rotate the preview. Check the back, underside, openings and thin parts.
- Check textures and geometry separately. Surface detail may be a texture rather than modeled depth.
- Add the model to your scene or download a supported output. Refine it with modeling or a processing tool when needed.
A real Show3D result
The griffin helmet is an existing Text to 3D result. Inspect the full silhouette as well as its surface details.
Process an existing model
Open the task selector at the top of Generate and choose an operation under Process a model to refine a compatible generated or uploaded asset. Each task creates a separate result and does not automatically replace an object in your scene.
Generate motion
Choose Generate motion to create character motion from a description. It can run without a source model. When retargeting is available, select an eligible previous motion result; this is separate from editing scene keyframes in Animation.