Images generated as separate and editable elements

Luma Layers generates images as editable layers. Modifying a product or text no longer requires rebuilding the composition.

Layers allows you to modify a specific part of an image without regenerating the entire composition. In Luma, objects, text, and the background become independent elements. This makes it possible to replace a product, adjust a color, or modify a headline while preserving the already approved framing, mood, and layout.

This approach is powered by Uni-1, which generates images directly as separate layers. Unlike tools that attempt to reconstruct an editable structure from a flattened image, the model organizes each element separately right from generation. An imported image can also be analyzed and then broken down using the same logic.

Text remains editable as true graphic elements. Users can change their content, color, or typography—including using a custom font—without shifting the rest of the scene. Layers can also be reordered, hidden, deleted, or repositioned to adapt the creation to different formats.

Each element can be extracted on a transparent background, reused in another composition, or combined with layers from other images. Luma also supports the creation of batch variations, notably to adapt the same visual into multiple languages while maintaining its graphic structure. Finally, elements can be individually upscaled up to 2K.

The entire process remains accessible from Luma Agents. A request made in natural language can be used to extract layers, modify an element, and then recompose the image without leaving the creative environment.