Once trained, your Seelab model is available wherever you create: in the Creation space, in the editor, in Magnify and in Video. The principle is the same everywhere: the model brings the product, the prompt describes the scene.

Generate images

1. In Creation, click the model name under the prompt. Your Seelab models appear at the top of the list. You can also click Use model from the Custom models library.
2. Describe the scene, not the product: the setting, the lighting, the framing, how the visual will be used.
3. Generate, then iterate.

Add reference images to enforce a setting, a mood or a composition: the product comes from the model, the rest from the reference. See Using up to 10 reference images.

Insert your product into an existing image

In the editor, the Replace tool accepts your Seelab model: paint the area where the product should go, choose the model and describe the staging. The inserted product respects the shape and the label the model has learned.

See Insert your product with your Seelab model.

Upscale with precision

In Magnify, select your Seelab model: the upscale relies on what the model knows about your product, for more faithful details up to 4K.

See Magnify: upscale up to 4K.

Generate videos

A single training creates your image model and your video model. In Video, choose your Seelab model to animate your product with the same fidelity.

See Animate your product with your video model.

Tips

• Is your product distorted? Remove any description of the product from the prompt.
• Can't find your model? Check the active project: models belong to each project.
• For a scene with several products that must be pixel-perfect, build it with the Composer.

Glossary

• Use model: the library button that opens Creation with the model selected.
• Replace: the editor tool that replaces or inserts an element in a painted area.
• Magnify: Seelab's upscaling tool.
• Video model: the video version of your Seelab model, created by the same training.

Updated on: 31/10/2026