WAFT-Stereo (fastforks)

WAFT-Stereo (Wang and Deng 2025): warping instead of a cost volume; Depth Anything V2 (LoRA) features, a ViT-based disparity proposal and 3 warp-and-refine iterations.

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/princeton-vl/WAFT-Stereo at c534320860f4.

Author weight repos: https://huggingface.co/MemorySlices/WAFT-Stereo.

Task: stereo. Tier: stable. Outputs: disparity. Backends: portable, kernels.

Input: a rectified left / right image pair. Output: a disparity map (pixels, left view).

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
waft-stereo-b@1	DAv2B-4.pth	65c52425722e	CC-BY-NC-4.0	no
waft-stereo-s@1	DAv2S-4.pth	340d6bff17ad	BSD-3-Clause	yes

Licence

Original-code strings, recorded as they are (research-use package; check them before any other use). Code: BSD-3-Clause. Files: LICENSE.

Training data: Freiburg SceneFlow (FlyingThings3D, Driving, Monkaa), not stated by the authors (SceneFlow, TartanAir, CREStereo, Falling Things, Sintel, Virtual KITTI 2, UnrealStereo4K, WMGStereo, CARLA, Spring, TartanGround, FSD (configs/SynLarge/*.yaml); only SceneFlow is in datasets.toml).

• Licence note: WAFT-Stereo-B is labelled CC-BY-NC-4.0 conservatively although the Hugging Face card says BSD-3-Clause: the checkpoint embeds a fine-tuned Depth Anything V2 ViT-B whose original weights are CC-BY-NC-4.0.
• Licence note: Both checkpoints were trained on data that includes NVIDIA's FSD dataset, whose terms for commercial use of trained models are unclear; the status does not depend on the training data.

Licence evidence: https://github.com/princeton-vl/WAFT-Stereo/blob/c534320860f4bbb2c0df896e3ccb95de49284aec/LICENSE; vendored files (net/*.py) checked against c534320: none carries an NVIDIA or FoundationStereo header, original only credits FoundationStereo in its README, so BSD-3-Clause holds

Input size

any H x W >= 17, resized or padded inside the model Tested up to (portable, fp16, RTX 5060 Ti 16 GB): waft-stereo-b 1024x1024 (3.2 GiB peak allocated, 4.7 reserved); waft-stereo-s 1024x2048 (5.3 GiB peak allocated, 8.3 reserved)

Speed and memory

Stereo, 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
waft-stereo-b	1897.5	1139.5	1137.23	392.88	2.9x	2.9x	5743 / 5876	5775 / 5852	1907 / 2316
waft-stereo-s	1251.6	689.3	689.01	283.76	2.4x	2.4x	3314 / 3398	3293 / 3456	1113 / 1286

Startup and first call

model	load (s): original	portable	kernels	first call (s): original	portable	kernels
waft-stereo-b	3.07 (375x1242)	1.65	0.42†	0.44 (375x1242)	1.15†	0.39†
waft-stereo-s	1.60 (375x1242)	0.58	0.30†	0.14 (375x1242)	0.69†	0.28†

Fixed-size speed and peak memory, all models

model	mode	640x480	1280x720	1920x1080	12 MP 4000x3000
waft-stereo-b	portable	SKIPPED	SKIPPED	SKIPPED	SKIPPED
waft-stereo-b	kernels	62.9 ms / 676 / 908	214.9 ms / 1310 / 1880	607.1 ms / 2523 / 3712	OOM
waft-stereo-s	portable	83.4 ms / 941 / 1146	357.7 ms / 2165 / 3356	1234.1 ms / 5368 / 8448	SKIPPED
waft-stereo-s	kernels	54.4 ms / 381 / 516	148.0 ms / 760 / 1120	441.0 ms / 1473 / 2150	6423.1 ms (n=3) / 7556 / 11364

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 authors' code 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

Original = the authors' code, same checkpoint and protocol.

variant	dataset	portable	kernels	note
waft-stereo-s	ETH3D two-view train, bad > 1 px % / EPE	0.95 / 0.177	0.95 / 0.177	possibly in the training data
waft-stereo-s	Middlebury 2014 trainingH, bad > 2 px % / EPE	4.75 / 0.895	4.77 / 0.897	possibly in the training data
waft-stereo-b	ETH3D two-view train, bad > 1 px % / EPE	0.74 / 0.163	0.74 / 0.163	possibly in the training data
waft-stereo-b	Middlebury 2014 trainingH, bad > 2 px % / EPE	3.39 / 0.728	3.38 / 0.728	possibly in the training data

Measured on NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, 2026-10-27.

Parity with the original code

Every variant is checked against golden outputs of the original code at its pinned commit. CPU, float32: max |diff| <= 0.002 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.02 px and p99 <= 0.25 px.

Limitations

• Only the two SynLarge checkpoints (S, B) are included; DAv2L-5 and the Real / KITTI / Middlebury fine-tunes are not.
• Portable WAFT-B is slightly slower than the original code and allocates slightly more memory; kernels is about 3x faster with a third of the memory.

Usage

from fastforks import stereo

model = stereo.load("waft-stereo-b")  # portable PyTorch on the GPU if there is one, else the CPU
result = model("left.png", "right.png")
result.disparity  # [H, W] float32 tensor, input resolution
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: waft-stereo-b, waft-stereo-s.

Kernels backend

stereo.load("waft-stereo-b", mode="kernels") runs our CUDA kernels (igev, cutlass_conv, cudnn_hacc, fastbb). 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.