pointdit / README.md
Flor's picture
card: update from fastforks 1cbf1671
34d1bbf verified
|
Raw History Blame Contribute Delete
10.3 kB
metadata
library_name: fastforks
pipeline_tag: depth-estimation
license: other
license_name: dinov3-license
license_link: >-
  https://github.com/google-research/pointdit/blob/11f53a30aeb4d34d49d2beda655ee6f19b2ad7f8/LICENSE
tags:
  - depth
  - mono-depth
  - monocular-depth
  - cuda
  - pytorch
  - fastforks
  - fastforks-card

PointDiT (fastforks)

PointDiT (Xu et al., Google 2026): monocular geometry as generation; a pixel-space diffusion transformer denoises a point map in 3 Euler steps, conditioned on frozen DINOv3 features of the image. Outputs an affine-invariant point map and its depth.

This repo is the model card only: it holds no weights. fastforks downloads the authors' weights from the links below and checks them against the SHA-256 in the model card. Code: https://github.com/kleinicke/fastforks. Kernels for the kernels backend: https://huggingface.co/fastforks/kernels. Original code: https://github.com/google-research/pointdit at 11f53a30aeb4.

Author weight repos: https://huggingface.co/facebook/dinov3-vitb16-pretrain-lvd1689m, https://huggingface.co/facebook/dinov3-vitl16-pretrain-lvd1689m, https://huggingface.co/haofeixu/pointdit.

Task: depth. Tier: experimental. Outputs: points, depth. Backends: portable, kernels.

Input: one RGB image. Output: a depth or inverse-depth map. kind (a card field, result.kind) says what the numbers mean: metric = metres; scale_invariant, affine_invariant, affine_invariant_log (log depth) and affine_invariant_inverse are relative (unknown scale or shift, not metres). Metric models that use the focal length take model(image, intrinsics=K) (K [3, 3], pixels of the input): metric3d-vit-large and da3metric-large need it for correct metres (da3metric-large: metres = output x focal / 300; without intrinsics= a 300 px focal length is assumed, wrong for a real camera). unidepth-v2-vitl14, unik3d-vitl, depth-pro and MoGe estimate the camera themselves (given intrinsics= replace fx); zoedepth-nk and the relative models (da3mono-large too) take none. Normals-only models (kind none) return result.normals [B, H, W, 3] (unit vectors, camera frame: x right, y down, z forward).

Variants

Weights are the authors' files (not hosted here). SHA-256 shortened to 12 hex digits; the full hashes are in card.toml.

variant weights (author link) SHA-256 weights licence commercial use
pointdit-b-512@1 dinov3-vitb16-pretrain-lvd1689m/model.safetensors
pointditb-512-mixdata-nodinov3-1d42aacf.pth
9a21ac3df0c6
1d42aacfd6aa
LicenseRef-DINOv3-License yes
pointdit-l-512@1 dinov3-vitl16-pretrain-lvd1689m/model.safetensors
pointditl-512-mixdata-nodinov3-240c1a4f.pth
dcb2e45127cc
240c1a4fc9a2
LicenseRef-DINOv3-License yes

Licence

Original-code strings, recorded as they are (research-use package; check them before any other use). Code: Apache-2.0. Licence files: LICENSE.

Training data: Mixture of synthetic and real point-map datasets ("mixdata") per the PointDiT paper; the DINOv3 encoder: LVD-1689M (not released).

  • Licence note: The PointDiT code and weights are Apache-2.0; the frozen DINOv3 ViT-B/16 and ViT-L/16 encoders are under the DINOv3 License (Meta: attribution and conditions apply; gated on Hugging Face); read it before commercial use.

Licence evidence: https://github.com/google-research/pointdit/blob/11f53a30aeb4d34d49d2beda655ee6f19b2ad7f8/LICENSE (Apache-2.0); https://huggingface.co/haofeixu/pointdit (license: apache-2.0); encoder: https://huggingface.co/facebook/dinov3-vitl16-pretrain-lvd1689m (DINOv3 License, gated)

Input size

any H x W >= 1, resized or padded inside the model Tested up to (portable, fp16, RTX 5060 Ti 16 GB): pointdit-b-512 3000x4000 (0.8 GiB peak allocated, 1.0 reserved); pointdit-l-512 3000x4000 (1.6 GiB peak allocated, 2.2 reserved)

Speed and memory

Depth, synthetic

model input time (ms): original fp32 original portable kernels kernels speed-up: vs original vs portable peak memory (MiB, allocated / reserved): original portable kernels
pointdit-b-512 518x770 – – 30.7 20.5 – 1.5x – 432 / 484 434 / 590
pointdit-l-512 518x770 – – 94.2 61.0 – 1.5x – 1285 / 1410 1314 / 1656

Depth, Middlebury 2014 half, 994x1482

model time (ms): original fp32 original portable kernels kernels speed-up: vs original vs portable peak memory (MiB, allocated / reserved): original portable kernels
pointdit-b-512 105 48.7 30.7 20.6 2.4x 1.5x 1214 / 1246 450 / 544 466 / 630
pointdit-l-512 335 112 94.6 61.4 1.8x 1.5x 3953 / 4010 1301 / 1410 1314 / 1696

Startup and first call

model load (s): original portable kernels first call (s): original portable kernels
pointdit-b-512 2.57 0.65 (518x770) 0.56 (518x770) 0.11 0.29 (518x770) 0.17 (518x770)
pointdit-l-512 9.47 0.92 (518x770) 0.94 (518x770) 0.12 0.27 (518x770) 0.25 (518x770)

torch.compile and TensorRT

model input time: portable compile TensorRT fp16 kernels preparation: compile (cold first call) ONNX + engine build peak MiB (alloc / reserved): portable compile kernels TensorRT (engine + torch)
pointdit-b-512 994x1482 30.7 ms 23.7 ms fails: BackendUnavailableError: onnx is not installed: pip install 20.6 ms 33.6 s – 450 / 544 465 / 524 466 / 630 –

Fixed-size cells: time (ms) / peak allocated / peak reserved memory (MiB).

Hardware, protocol and caveats: benchmarks (RTX 5060 Ti 16 GB, sm_120). Original = the original code named above at the precision of portable (fp16, or (fp32) where fp16 is unsupported or not finite); original fp32 = its default fp32 run. Peak memory is the torch allocator's.

Accuracy

On public datasets

AbsRel / delta1 after a least-squares scale + shift fit (in disparity space for inverse-depth models), mean over images. Original = the original code named under Details, same checkpoint and protocol.

variant dataset original fp32 original fp16 portable kernels note
pointdit-b-512 NYUv2 test, AbsRel / delta1 0.0453 / 0.9773 0.0454 / 0.9773 0.0453 / 0.9773 0.0453 / 0.9773
pointdit-b-512 iBims-1, AbsRel / delta1 0.0374 / 0.9821 0.0374 / 0.9820 0.0374 / 0.9821 0.0374 / 0.9821
pointdit-b-512 ETH3D two-view train left, AbsRel / delta1 0.0896 / 0.9085 0.0896 / 0.9086 0.0896 / 0.9086 0.0896 / 0.9085 possibly in the training data
pointdit-b-512 Middlebury 2014 trainingH left, AbsRel / delta1 0.0313 / 0.9948 0.0313 / 0.9948 0.0313 / 0.9948 0.0313 / 0.9948 possibly in the training data
pointdit-l-512 NYUv2 test, AbsRel / delta1 0.0410 / 0.9800 0.0411 / 0.9800 0.0410 / 0.9800 0.0410 / 0.9800
pointdit-l-512 iBims-1, AbsRel / delta1 0.0313 / 0.9848 0.0313 / 0.9847 0.0313 / 0.9848 0.0313 / 0.9848
pointdit-l-512 ETH3D two-view train left, AbsRel / delta1 0.0748 / 0.9264 0.0748 / 0.9278 0.0748 / 0.9265 0.0748 / 0.9261 possibly in the training data
pointdit-l-512 Middlebury 2014 trainingH left, AbsRel / delta1 0.0230 / 0.9952 0.0230 / 0.9951 0.0230 / 0.9952 0.0230 / 0.9952 possibly in the training data

Measured on NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 2fb984b3, 2026-11-15.

Parity with the original code

Every variant is checked against reference outputs of the original code at its pinned commit. CPU, float32: max |diff| <= 1.5e-06 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.00029 and p99 <= 0.0012 (absolute error of the output points).

Limitations

  • Affine-invariant output: points up to an unknown scale and 3-D shift, depth (their z) up to scale and shift; align before use.
  • Trained at 512 x 512: inputs are resized to about tokens 16-pixel patches at their aspect ratio (bilinear) and the point map is resized back with nearest, as the original's wild-image path.

Usage

from fastforks import depth

model = depth.load("pointdit-b-512")  # mode="auto": the preferred backend from recorded measurements
result = model("room.jpg")
result.points  # [H, W] float32 tensor; result.kind says what the maps mean
# models that take a camera: model("room.jpg", intrinsics=K), K [3, 3] in pixels
result.backend  # what ran, e.g. Backend(mode='portable', device='cuda:0', dtype='float16')

Install with pip install fastforks (see https://github.com/kleinicke/fastforks). Other variants: pointdit-b-512, pointdit-l-512.

Kernels backend

depth.load("pointdit-b-512", mode="kernels") runs our CUDA kernels (fastbb, dpt_fused). It needs a prebuilt kernel for your GPU architecture (downloaded from https://huggingface.co/fastforks/kernels on first use) and a recent NVIDIA driver; if there is none it raises BackendUnavailableError with the reason and never falls back silently. mode="auto" picks kernels when it can run and logs the choice.