SeedVR2
SeedVR2 (Wang et al., ByteDance Seed, 2025): upscales and restores photos and video frames x2 with a large diffusion transformer run for a single step; 3 B and 7 B versions, the 7 B runs on 16 GB with weight streaming (default, ~17 GB free host RAM). Apache-2.0 code and weights.
experimental Fork of ByteDance-Seed/SeedVR@e4de8c2 · backends: portable, kernels
- Paper SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training arXiv · first released 5 Jun 2025
Paper 2025 Added in 0.2.0 Evaluated with 0.2.0
Use it
pip install fastforks Lean install with only what this model needs: pip install fastforks-core
from fastforks import restoration
model = restoration.load("seedvr2-3b") # kernels on a supported GPU, else portable PyTorch
result = model(image)
result.image
result.backend # what ran: mode, device, dtype, kernels fastforks bench seedvr2-3b Times the model on this machine. load(…, mode="portable") or mode="kernels" picks a backend; usage guide.
Speed and memory
Every backend on one GPU (RTX 5060 Ti 16 GB) and the same input, the authors’ code included. Kernels runs bf16. Portable runs bf16 on CUDA.
| Backend | dtype | Time per call (ms) | vs original fp16 | Peak memory allocated / reserved | Load (s) | First call (s) |
|---|---|---|---|---|---|---|
| Original fp32 | fp32 | not measured | – | – | – | – |
| Original (own default)load: the authors' fp32 .pth files from the page cache (read once before) | bf16 | 13150 | – | 7,598 MiB/14.9 GiB | 96.2 | 13.89 |
| Portable fp32 | fp32 | not measured | – | – | – | – |
| Portable | bf16 | 2295 | 5.7× | 6,140 MiB/6,772 MiB | 2.6 | 2.45 |
| Kernels | bf16 | 2296 | 5.7× | 6,140 MiB/6,772 MiB | 3.3 | 2.46 |
| Portable + torch.compile | 1941 | 6.8× | 5,237 MiB/8,144 MiB | 238 (build) | – | |
| TensorRT | not measured here | – | – | – | – |
Measured on RTX 5060 Ti 16 GB, torch 2.14.1+cu130, 2026-11-14 to 2026-11-15. First call, 3 warm-up, peak reset, median of 3 rounds x median of 5. Ratios divide two
times of this table (same GPU, same input). Original fp16 is the authors’ own precision option, else torch.autocast. torch.compile and TensorRT: build time instead of load time. TensorRT memory (engine, not comparable) is the GPU memory the engine took, read from the driver because TensorRT allocates outside PyTorch, so it can't be compared with the PyTorch peaks of the other rows.
Other devices
| Device | Backend | dtype | ms / call | Peak memory | Output vs reference | Input |
|---|---|---|---|---|---|---|
| Apple M1 Pro 16 GB MPS, torch 2.14.1, 2026-11-15 | Original | fp32 | cannot run here | – | 512x512 | |
| Portable | fp32 | cannot run here | – | 512x512 | ||
| Kernels | bf16 | not implemented | – | 512x512 | ||
CUDA memory is torch’s peak allocated / reserved; Apple has no such counter, so the Mac rows show the process footprint (weights, Python and torch included). “Output vs reference” compares the device’s output with the stored golden output under the card’s tolerance; – = not checked on that device. Kernels “not implemented” = no kernels for this model on that backend yet: it runs portable there.
Licence and weights
Commercial use allowed
- Code
- Apache-2.0 LICENSENOTICE
- Weights
- Apache-2.0 source
- Training data
- Paper (arXiv 2506.05301): adversarial post-training of SeedVR on large-scale image and video data with synthetic degradations; the data sources and terms are not stated.
- The converted weight files (bf16 transformer with the fixed prompt embedding, bf16 VAE) are hosted by fastforks under the originals' Apache-2.0 licence; NOTICE keeps the attribution and lists the changes.
Our reading of the licences, not legal advice.
Accuracy
The test set through the authors’ code and through ours: same checkpoint, same protocol. The distance between the original’s fp32 and fp16 runs is what half precision alone costs; our runs should stay within it.
DIV2K validation
4 centre crops 512 px, bicubic x2 LR -> HR, PSNR / SSIM (Y) / PSNR to the original
| Run | dtype | psnr ↑ | ssim ↑ | psnr_orig ↑ | images | ms / image |
|---|---|---|---|---|---|---|
| Original fp32 | not measured | |||||
| Original fp16 | bf16 | 22.02 | 0.5865 | – | 4 | – |
| Portable fp32 | not measured | |||||
| Portable | bf16 | 22.00 | 0.5863 | 43.13 | 4 | – |
| Kernels | bf16 | 22.00 | 0.5863 | 43.13 | 4 | – |
The original code was not run on this dataset, so the table cannot show the distance to it.
SeedVR2 is a generative one-step restorer, so PSNR is only a fidelity check. Each of the centre 512 px crops of DIV2K validation HR images (a tiny subset: generation is slow) is downscaled with bicubic x2 (PIL) and restored x2 by the model, one pass; PSNR and SSIM (Gaussian 11 x 11, sigma 1.5) on the Y channel of the uint8 output with a 2 px border cropped, averaged over the crops; PSNR to the original = RGB uint8 PSNR of our output against the authors' output of the same input and settings. The original is the authors' inference code, bf16. No perceptual metric (LPIPS) is given: the lpips package and its network weights are not part of the test environment. Original = the authors’ code, same checkpoint and protocol. Measured on original fp16: RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-16; portable, kernels: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 33b2432c, 2026-11-16.
DIV2K validation
25 bicubic x2 center crops of 64 x 64 LR pixels, RGB PSNR
| Run | dtype | psnr ↑ | images | ms / image |
|---|---|---|---|---|
| Original fp32 | not measured | |||
| Original fp16 | not measured | |||
| Portable fp32 | not measured | |||
| Portable | bf16 | 20.2542 | 25 | 81.6 |
| Kernels | not measured | |||
The original code was not run on this dataset, so the table cannot show the distance to it.
Every fourth DIV2K validation image. A 64 x 64 center crop of the official bicubic x2 input is restored at 128 x 128 and compared with the corresponding original pixels. PSNR uses RGB in [0, 1] and is averaged over the 25 crops. Both modes use resident BF16 weights and seed 666. Measured on RTX 5060 Ti, torch 2.14.1+cu130, fastforks 29b3bbad, 2026-11-18.
4 video clips (first frame)
PSNR to the original's output
| Run | dtype | psnr_orig ↑ | clips | ms / clip |
|---|---|---|---|---|
| Original fp32 | not measured | |||
| Original fp16 | not measured | |||
| Portable fp32 | not measured | |||
| Portable | bf16 | 44.56 | 4 | – |
| Kernels | bf16 | 44.56 | 4 | – |
The original code was not run on this dataset, so the table cannot show the distance to it.
First frame of each clip (clips of 2 to 10 frames, x2); PSNR of our output (RGB, uint8) against the authors' output of the same clip; there is no ground truth for these clips. Measured on RTX 5060 Ti, torch 2.14.1+cu130, fastforks 33b2432c, 2026-11-16.
Closeness to the original
Beyond the test-set scores: the raw output of ours against the original’s fp32 output on the same input, pixel by pixel. CPU, float32: max |diff| <= 0.02 vs the original. GPU, bfloat16 (the default), portable and kernels: mean |diff| <= 0.01 and p99 <= 0.05 (absolute error of the output image).
| Check | Tolerance | Measured |
|---|---|---|
| CPU fp32, max |diff| | ≤ 0.02 | not recorded |
| GPU default dtypes, mean |diff| | ≤ 0.01 | 0 |
| GPU default dtypes, p99 |diff| | ≤ 0.05 | 0 |
Recorded for seedvr2-3b on RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-14, with these kernels: nf4
Details
SeedVR2 (Wang et al., ByteDance Seed 2025): one-step video restoration and super-resolution by a diffusion transformer (NaDiT, adversarially post-trained) on the latents of a causal 3-D VAE. `seedvr2-3b` and `seedvr2-7b` are the 3 B and 7 B transformers (same VAE); frames are restored one at a time.
- Input / output
- Input: one RGB image. Output: the restored (upscaled) image,
model.scalex the input size, RGB in [0, 1].What the numbers mean, and which models need the camera focal length
tile=runs large images in overlapping tiles (as ComfyUI). - Input size
- any H x W >= 8, padded to a multiple of 8; tested up to 512x512 (11.7 GiB allocated)
- Options
seed=666· changes the outputstream=true· changes the outputquant="none"one of 'none', 'nf4'· changes the output
- Outputs
image- CUDA kernels
nf4
Limitations
- Generative restoration: the output depends on `seed` (one CPU generator draws the VAE posterior noise and the diffusion noise, upstream's default 666); other seeds give other plausible details.
- One frame per call: the clip mode of the original (temporal attention over 4n + 1 frames) is not used by this image API.
- The input is padded on the bottom and right (replicate) so that `scale x` its size is a multiple of 16, and the output cropped back; the original centre-crops the resized frame to a multiple of 16 instead.
- No colour fix: the authors' scripts optionally match the output colours to the input with StableSR's wavelet reconstruction (non-commercial code), which is not included.
- The 3 B model needs ~7 GB for its weights plus the VAE's activations (~14 GB peak at 960 x 1696 output); `stream=False` keeps them resident; by default (`stream=True`, CUDA) they stay in ~6.6 GB of page-locked host RAM (~9 GB peak at that size).
- The 3 B TensorRT measurement uses `stream=False` and fixed 64 x 64 input. Export with `export_onnx`, then build the engine in a fresh process. On 25 DIV2K crops the largest per-crop PSNR change was 0.109 dB. The largest single-pixel RGB channel difference was 0.627 on the [0, 1] scale.
- The 7 B model's weights are 16.5 GB in bf16: it runs with the default `stream=True` (blocks in ~16.5 GB of page-locked host RAM; load fails with an error when less is free) or `quant="nf4"` (~4.6 GB resident); with `stream=false` and no quant it runs out of GPU memory.
- `quant="nf4"` changes the generated detail: 23-27 dB PSNR against the bf16 output (3 B, 4 clips; 7 B: 26-31 dB, 2 clips), about as far as another `seed` (18-24 dB), with ~3.5x smaller transformer blocks.
Full reference, generated from the repository
SeedVR2 (Wang et al., ByteDance Seed, 2025): upscales and restores photos and video frames x2 with a large diffusion transformer run for a single step; 3 B and 7 B versions, the 7 B runs on 16 GB with weight streaming (default, ~17 GB free host RAM). Apache-2.0 code and weights.
Commercial use: yes.
| variant | input | backend | time (ms) | peak memory (MiB, allocated / reserved) |
|---|---|---|---|---|
seedvr2-3b | synthetic (512x512) | original portable kernels portable + torch.compile | 13150 2295 2296 1941 | 7598 / 15268 6140 / 6772 6140 / 6772 5237 / 8144 |
seedvr2-3b | 64x64 | original portable kernels TensorRT bf16 | – – – 46.8 | – – – 7662 / 7664 |
seedvr2-7b | synthetic (512x512) | original portable kernels portable + torch.compile TensorRT fp16 | OOM 2712 2713 fails: load: OutOfMemoryError: CUDA out of memory. Tried to allocat fails: RuntimeError: Inplace update to inference tensor outside Inf | – 6402 / 7034 6402 / 7034 – – |
Weight streaming
Warm inference time and peak reserved GPU memory for the same input with both settings. Streamed weights use pinned host memory.
| Model | Mode | Input | GPU | PyTorch | Resident (stream=False) | Streamed (stream=True) |
|---|---|---|---|---|---|---|
| seedvr2-3b | portable | 512x512 | RTX 5060 Ti | 2.14.1+cu130 | 2,281 ms / 15,348 MiB | 2,294 ms / 8,876 MiB |
| seedvr2-7b | portable | 512x512 | RTX 5060 Ti | 2.14.1+cu130 | Out of memory | 2,715 ms / 8,880 MiB |
Details
- Original code: https://github.com/ByteDance-Seed/SeedVR at
e4de8c24441a - Code licence: Apache-2.0 (permissive); files:
LICENSE,NOTICE - Weights licence: Apache-2.0
- Licence evidence: https://github.com/ByteDance-Seed/SeedVR/blob/e4de8c24441a67e1b7df56abea10645059bb1185/LICENSE ; https://huggingface.co/ByteDance-Seed/SeedVR2-3B (license: apache-2.0)
- Training data: Paper (arXiv 2506.05301): adversarial post-training of SeedVR on large-scale image and video data with synthetic degradations; the data sources and terms are not stated.
- Licence note: The converted weight files (bf16 transformer with the fixed prompt embedding, bf16 VAE) are hosted by fastforks under the originals' Apache-2.0 licence; NOTICE keeps the attribution and lists the changes.
- Options:
seed= 666,stream= True,quant= none (one of 'none', 'nf4') - Kernel libraries:
nf4 - Variants:
seedvr2-3b,seedvr2-7b - Outputs: image
- Input size: any H x W >= 8, padded to a multiple of 8
- Tested up to (portable, fp16, RTX 5060 Ti 16 GB):
seedvr2-3b512x512 (11.7 GiB peak allocated, 15.0 reserved);seedvr2-7b512x512 (6.3 GiB peak allocated, 8.7 reserved) seedvr2-3b512x512, first call: portable + torch.compile 238 s (cold Inductor cache).seedvr2-3b64x64, first call: TensorRT bf16 324 s (ONNX + engine).seedvr2-3boriginal 512x512: load: the authors' fp32 .pth files from the page cache (read once before)seedvr2-3bload / first call: original 96.2 s / 13.89 s, portable 2.65 s / 2.454 s, kernels 3.27 s / 2.462 sseedvr2-7bload / first call: portable 222.77 s / 3.174 s, kernels 118.43 s / 3.176 s- Protocol, first call, 3 warm-up, peak reset, median of 3 rounds x median of 5
Measured
SeedVR2 is a generative one-step restorer, so PSNR is only a fidelity check. Each of the centre 512 px crops of DIV2K validation HR images (a tiny subset: generation is slow) is downscaled with bicubic x2 (PIL) and restored x2 by the model, one pass; PSNR and SSIM (Gaussian 11 x 11, sigma 1.5) on the Y channel of the uint8 output with a 2 px border cropped, averaged over the crops; PSNR to the original = RGB uint8 PSNR of our output against the authors' output of the same input and settings. The original is the authors' inference code, bf16. No perceptual metric (LPIPS) is given: the lpips package and its network weights are not part of the test environment. Every fourth DIV2K validation image. A 64 x 64 center crop of the official bicubic x2 input is restored at 128 x 128 and compared with the corresponding original pixels. PSNR uses RGB in [0, 1] and is averaged over the 25 crops. Both modes use resident BF16 weights and seed 666. First frame of each clip (clips of 2 to 10 frames, x2); PSNR of our output (RGB, uint8) against the authors' output of the same clip; there is no ground truth for these clips. Original = the original code named under Details, same checkpoint and protocol.
| variant | dataset | original fp16 | portable | kernels | TensorRT | note |
|---|---|---|---|---|---|---|
seedvr2-3b | DIV2K validation, 4 centre crops 512 px, bicubic x2 LR -> HR, PSNR / SSIM (Y) / PSNR to the original | 22.02 / 0.5865 / – | 22.00 / 0.5863 / 43.13 | 22.00 / 0.5863 / 43.13 | ||
seedvr2-3b | DIV2K validation, 25 bicubic x2 center crops of 64 x 64 LR pixels, RGB PSNR | 20.2542 | 20.2671 | |||
seedvr2-3b | 4 video clips (first frame), PSNR to the original's output | 44.56 | 44.56 | |||
seedvr2-7b | DIV2K validation, 2 centre crops 512 px, bicubic x2 LR -> HR, PSNR / SSIM (Y) / PSNR to the original | 23.16 / 0.6228 / – | 23.13 / 0.6227 / 44.85 | 23.13 / 0.6227 / 44.85 | ||
seedvr2-7b | 2 video clips (first frame), PSNR to the original's output | 46.14 | 46.14 |
Measured on portable: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 29b3bbad / 33b2432c, 2026-11-16 to 2026-11-18; kernels: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 33b2432c, 2026-11-16; tensorrt: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 29b3bbad, 2026-11-18.
Parity
CPU, float32: max |diff| <= 0.02 vs the original. GPU, bfloat16 (the default), portable and kernels: mean |diff| <= 0.01 and p99 <= 0.05 (absolute error of the output image).
| variant | kernels error vs original, mean / p99 | GPU, torch | kernels used |
|---|---|---|---|
seedvr2-3b | 0.0 / 0.0 | RTX 5060 Ti, torch 2.14.1+cu130 | nf4@e236d80ba78b:sm_120 |
seedvr2-7b | 0.0 / 0.0 | RTX 5060 Ti, torch 2.14.1+cu130 | nf4@e236d80ba78b:sm_120 |
Limitations
- Generative restoration: the output depends on
seed(one CPU generator draws the VAE posterior noise and the diffusion noise, upstream's default 666); other seeds give other plausible details. - One frame per call: the clip mode of the original (temporal attention over 4n + 1 frames) is not used by this image API.
- The input is padded on the bottom and right (replicate) so that
scale xits size is a multiple of 16, and the output cropped back; the original centre-crops the resized frame to a multiple of 16 instead. - No colour fix: the authors' scripts optionally match the output colours to the input with StableSR's wavelet reconstruction (non-commercial code), which is not included.
- The 3 B model needs ~7 GB for its weights plus the VAE's activations (~14 GB peak at 960 x 1696 output);
stream=Falsekeeps them resident; by default (stream=True, CUDA) they stay in ~6.6 GB of page-locked host RAM (~9 GB peak at that size). - The 3 B TensorRT measurement uses
stream=Falseand fixed 64 x 64 input. Export withexport_onnx, then build the engine in a fresh process. On 25 DIV2K crops the largest per-crop PSNR change was 0.109 dB. The largest single-pixel RGB channel difference was 0.627 on the [0, 1] scale. - The 7 B model's weights are 16.5 GB in bf16: it runs with the default
stream=True(blocks in ~16.5 GB of page-locked host RAM; load fails with an error when less is free) orquant="nf4"(~4.6 GB resident); withstream=falseand no quant it runs out of GPU memory. quant="nf4"changes the generated detail: 23-27 dB PSNR against the bf16 output (3 B, 4 clips; 7 B: 26-31 dB, 2 clips), about as far as anotherseed(18-24 dB), with ~3.5x smaller transformer blocks.
Example output
No example rendered for this checkpoint yet.