When you upload a depth map it will assist the AI model with asset generation based on the volume and depth of the object.
In a depth map, areas that are whiter are “closer” to the viewer, while images that are blacker are “further” from the viewer. You can use depth maps to more easily establish foreground, midgrounds, and backgrounds in an asset.
When uploading an image, you can use the Image Similarity slider to adjust how much the generated output should match the depth map. A higher % image similarity will cause the output images to more closely match the depth map.