NAFNet (fastforks)

NAFNet (Chen et al., ECCV 2022): a nonlinear-activation-free U-Net for image restoration: blocks of channel LayerNorm, 1 x 1 and depthwise 3 x 3 convs, a gate that multiplies two channel halves (SimpleGate) and simplified channel attention, four 2 x down / up levels. nafnet-gopro removes motion blur (GoPro, width 64, run with the test-time local converter as upstream's NAFNetLocal), nafnet-sidd removes smartphone sensor noise (SIDD, width 64).

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/megvii-research/NAFNet at 2b4af71ebe09.

Task: restoration. Tier: experimental. Outputs: image. Backends: portable, kernels.

Input: one RGB image. Output: the restored (upscaled) image, model.scale x the input size, RGB in [0, 1]. tile= runs large images in overlapping tiles (as ComfyUI).

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
nafnet-gopro@1 NAFNet-GoPro-width64.pth 329d3ab4077b MIT yes
nafnet-sidd@1 NAFNet-SIDD-width64.pth cd685efaae01 MIT yes

Licence

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

Training data: GoPro deblurring (GOPRO_Large, Nah et al. 2017: 3214 blurred / sharp frame pairs) (options/train/GoPro/NAFNet-width64.yml: GoPro train (2103 blurred / sharp pairs), 256 x 256 crops.).

Licence evidence: https://github.com/megvii-research/NAFNet/blob/2b4af71ebe098a92a75910c233a3965a3e93ede4/LICENSE ; https://github.com/megvii-research/NAFNet/blob/2b4af71ebe098a92a75910c233a3965a3e93ede4/docs/GoPro.md ; https://github.com/megvii-research/NAFNet/blob/2b4af71ebe098a92a75910c233a3965a3e93ede4/docs/SIDD.md

Input size

any H x W >= 1, resized or padded inside the model Tested up to (portable, fp16, RTX 5060 Ti 16 GB): nafnet-gopro 1988x2880 (7.0 GiB peak allocated, 9.6 reserved); nafnet-sidd 1988x2880 (7.8 GiB peak allocated, 10.1 reserved)

Speed and memory

Restoration, synthetic, 512x512

model time (ms): original portable kernels kernels speed-up: vs original vs portable peak memory (MiB, allocated / reserved): original portable kernels
nafnet-gopro 63.7 60.4 55.3 1.2x 1.1x 484 / 612 451 / 532 451 / 550
nafnet-sidd 71.6 60.6 56.9 1.3x 1.1x 608 / 730 575 / 670 575 / 650

Startup and first call

model load (s): original portable kernels first call (s): original portable kernels
nafnet-gopro 1.50 0.35 0.64 0.39 0.28 0.23
nafnet-sidd 1.16 0.79 0.70 0.29 0.37 0.24

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)
nafnet-gopro 512x512 60.4 ms 35.4 ms 52.6 ms 55.3 ms 38.1 s 66.5 s 451 / 532 422 / 470 451 / 550 722 / 734
nafnet-sidd 512x512 60.6 ms 30.9 ms 24.5 ms 56.9 ms 30.3 s 33.8 s 575 / 670 450 / 508 575 / 650 650 / 668

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

Each degraded image at its own size, one pass without tiles (the model pads internally); PSNR and SSIM of the output rounded to uint8 on RGB, no border crop, averaged over the images (the authors' test scripts). Original = the original code named under Details, same checkpoint and protocol.

variant dataset original fp32 portable kernels note
nafnet-gopro GoPro test (every 4th: 278 frames), deblurring, PSNR / SSIM (RGB) 33.73 / 0.9468 33.73 / 0.9468 33.73 / 0.9468
nafnet-sidd SIDD validation (1280 crops), denoising, PSNR / SSIM (RGB) 40.21 / 0.9243 40.21 / 0.9243 40.21 / 0.9243

Measured on NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks f104019b, 2026-11-14.

Parity with the original code

Every variant is checked against reference outputs of the original code at its pinned commit. CPU, float32: max |diff| <= 0.003 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.0015 and p99 <= 0.005 (absolute error of the output image).

Usage

from fastforks import restoration

model = restoration.load("nafnet-gopro")  # mode="auto": the preferred backend from recorded measurements
result = model("photo.png")  # tile=512 for large images
result.image  # [3, sH, sW] float32 RGB in [0, 1], s = model.scale
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: nafnet-gopro, nafnet-sidd.

Kernels backend

restoration.load("nafnet-gopro", mode="kernels") runs our CUDA kernels (cudnn_hacc). 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.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Collection including fastforks/nafnet