Image-based model that predicts depth maps, surface normals and albedo from a single regular image.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Marigold v2 is a pixel-level computer vision model that takes an ordinary image and produces dense geometric and material estimates: depth maps, surface normals and albedo. It continues the Marigold line of image-analysis models. The only source available here is a creator video and a Hugging Face Space (listed under the huawei-bayerlab namespace), so vendor and details are not fully confirmed.
A creator reported that it outperforms MoGe 3 and InfiniDepth in comparisons, and that it has been released. The same source states a GPU memory requirement of roughly 17 GB VRAM at one resolution and about 29 GB at a higher resolution.
It can be tried through the Hugging Face web demo. Licence, weights availability, and release date were not verified from official pages, so they are left unknown.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Pixel-level model that converts a regular image into depth maps, surface normals and albedo. Beats Moge 3 and InfiniDepth in comparisons and is released; needs about 17 GB VRAM at one resolution and 29 GB at higher.
Closest neighbours by meaning, not just by category.
Model · Google
Predicts depth, surface normals and sky masks from a single 360° panorama using a cubemap-adapted 3D foundation model.
Model · Nvidia
NVIDIA research model that turns a single image into a queryable 3D Gaussian scene with semantics and object instances.
Model · Apple
Apple research model that generates 3D assets with view-dependent appearance from a single input image.
Model · Pruna AI
Open-weight image generation and editing model from Pruna AI, distilled from MiniMax's H3 video model to run in 8 steps.
Research
Post-training method that refines feed-forward 3D reconstruction models (Pi3, DA3, VGGT) on unlabeled video.
Model
Feed-forward 3D Gaussian Splatting model that reconstructs scenes from arbitrary input views, with or without camera poses.