Depth Anything V3 Image
Depth Anything V3 Image estimates scene depth from a supplied photograph, render, or illustration. It captures small structures, including strands of hair and narrow objects, and assigns sky regions to the distant end of the depth range.
The output translates the source composition into spatial information. Grayscale rendering provides a depth signal for downstream image workflows, while colorized rendering offers a way to inspect the scene's depth relationships visually. Both represent depth estimated from the same source image.
What it's good for
A depth map can guide image generation that follows a reference composition while changing its appearance. In compositing, it can drive masks and focus treatments around detailed boundaries, such as a portrait subject against foliage.
Outdoor imagery is also a useful fit: sky-aware depth supports distance-based haze treatments. Motion designers can use the map for parallax animation, and artists can feed it into depth-aware relighting workflows.
