Skip to content

[SIGGRAPH Asia 2026] 4DAnyone: Create Anyone in 4D from a Casual Monocular Video

Release: compare 4DAnyone-Base and 4DAnyone-Turbo (#24)

Yudong Jin4159ba01f8b5
diff --git a/README.md b/README.md
index 84a2f61..fb81f9d 100644
--- a/README.md
+++ b/README.md
@@ -10,7 +10,7 @@
## News
-- **2026-09-02**: Achieved a **5.58×** denoising speedup over the base model using the Wan2.2 5B Turbo LoRA. See [Inference performance](docs/inference_performance.md) for details.
+- **2026-09-02**: Released **4DAnyone-Turbo**, achieving a **5.58×** denoising speedup over the base 4DAnyone model. See [Inference performance](docs/inference_performance.md) for details.
- **2026-08-28**: Achieved a **1.42×** end-to-end speedup for the complete 24-view generation pipeline.
- **2026-08-28**: Reduced peak GPU memory to **25.4 GiB** while slightly improving speed.
@@ -36,6 +36,8 @@ python scripts/download_example.py
## Inference
+This release provides two models: **4DAnyone-Base**, corresponding to the original model described in the paper, and **4DAnyone-Turbo**, the default accelerated model. Set `--enable_turbo=False` to use **4DAnyone-Base**.
+
4DAnyone supports flexible target-view counts, pitch layers, and yaw coverage. Here are several common camera configurations:
### 6-View Full Orbit
@@ -95,7 +97,7 @@ Run `python inference.py --help` for the full list.
- `start_yaw`: horizontal angle of the first view, in degrees. Yaw `0` is the front view.
- `yaw_span`: horizontal range covered by each camera layer, in degrees.
- `gpu_ids`: GPU IDs used for parallel pose/VAE view stages and target denoising. Defaults to all visible GPUs.
-- `enable_turbo`: enable the Wan2.2 5B Turbo LoRA. Defaults to `True`; set it to `False` to use the base model.
+- `enable_turbo`: whether to use 4DAnyone-Turbo. Enabled by default.
### Output
@@ -142,7 +144,7 @@ See the [nerfstudio guide](docs/nerfstudio.md) for details.
### Inference Acceleration
- [x] Achieve up to 1.42× end-to-end speedup by parallelizing pose encoding and VAE processing across GPUs.
-- [x] Accelerate inference with the Wan2.2 5B Turbo LoRA.
+- [x] Achieve a **5.58×** denoising speedup with 4DAnyone-Turbo.
### Reconstruction
diff --git a/docs/inference_performance.md b/docs/inference_performance.md
index d378867..9ad08ef 100644
--- a/docs/inference_performance.md
+++ b/docs/inference_performance.md
@@ -1,6 +1,21 @@
# Inference performance
-This page reports matched inference efficiency for the default Wan2.2 5B Turbo LoRA and the unfused base model on the [6-view full-orbit](../README.md#6-view-full-orbit) example:
+We compare **4DAnyone-Turbo**, the default accelerated model, with **4DAnyone-Base**, the original model described in the paper.
+
+## Quality comparison
+
+We evaluate 32 validation scenes: 16 from DNA-Rendering, 8 from MVGameHuman, and 8 from SynCamVideo. Both models use identical target cameras, 25-frame clips, seed 42, and foreground-masked metrics; 4DAnyone-Turbo uses 4 denoising steps with scheduler shift 5, while 4DAnyone-Base uses 50 steps, serving as an upper-bound quality reference for our method.
+
+| Model | Denoising steps | PSNR ↑ | SSIM ↑ | LPIPS ↓ |
+| --- | ---: | ---: | ---: | ---: |
+| 4DAnyone-Turbo | 4 | 25.48 | 0.895 | **0.072** |
+| 4DAnyone-Base | 50 | **25.69** | **0.899** | 0.073 |
+
+The two models achieve comparable quality across PSNR, SSIM, and LPIPS.
+
+## Denoising performance
+
+We measure 4DAnyone-Turbo with 4 denoising steps and 4DAnyone-Base with 24 steps on the [6-view full-orbit](../README.md#6-view-full-orbit) example:
```bash
python inference.py \
@@ -8,18 +23,16 @@ python inference.py \
--views_per_layer 6
```
-The command above uses Turbo by default. Add `--enable_turbo=False` to run the base model. Each denoising configuration below is the median of three complete runs from the same source revision.
-
-## Denoising
+Each reported denoising time is the median of three complete runs from the same source revision.
-These tables compare denoising time across attention backends for the Turbo and base models.
+These tables compare denoising time across attention backends for both models.
- `SDPA`, `FlashAttention-3`, and `SageAttention`: denoising time using each attention backend.
-- `Peak CUDA allocated`: maximum live tensor allocation observed across the three runs and available attention backends for each GPU and model profile.
+- `Peak CUDA allocated`: maximum live tensor allocation across all runs and available attention backends.
- `GPU memory`: total device memory capacity.
- `—`: backend unavailable on that GPU.
-### Turbo (default)
+### 4DAnyone-Turbo
| GPU | SDPA (min) | FlashAttention-3 (min) | SageAttention (min) | Peak CUDA allocated (GiB) | GPU memory (GiB) |
| --- | ---: | ---: | ---: | ---: | ---: |
@@ -28,7 +41,7 @@ These tables compare denoising time across attention backends for the Turbo and
| RTX 5880 Ada | 3.51 | — | 3.04 | 25.47 | 47.37 |
| RTX A6000 | 3.80 | — | 3.71 | 24.91 | 47.53 |
-### Base
+### 4DAnyone-Base
| GPU | SDPA (min) | FlashAttention-3 (min) | SageAttention (min) | Peak CUDA allocated (GiB) | GPU memory (GiB) |
| --- | ---: | ---: | ---: | ---: | ---: |
@@ -37,9 +50,9 @@ These tables compare denoising time across attention backends for the Turbo and
| RTX 5880 Ada | 20.40 | — | 17.58 | 25.47 | 47.37 |
| RTX A6000 | 22.02 | — | 21.42 | 24.91 | 47.53 |
-### Base / Turbo speedup
+### Speedup
-Each value divides the Base median by the Turbo median for the same GPU and attention backend.
+Speedup is calculated by dividing the 4DAnyone-Base median by the 4DAnyone-Turbo median for the same GPU and attention backend.
| GPU | SDPA | FlashAttention-3 | SageAttention |
| --- | ---: | ---: | ---: |
@@ -49,11 +62,11 @@ Each value divides the Base median by the Turbo median for the same GPU and atte
| RTX A6000 | 5.79× | — | 5.77× |
> [!NOTE]
-> FlashAttention-3 is available on the tested SM90 GPUs and unavailable on RTX A6000 (SM86) and RTX 5880 Ada (SM89). SageAttention results use the official `sageattn` entry point, which dispatches an architecture-specific kernel.
+> SageAttention results use the official `sageattn` entry point, which automatically dispatches an architecture-specific kernel for each GPU.
-## Preprocessing
+## Preprocessing performance
-Preprocessing is shared by Turbo and Base, so it is reported once. These profile-independent values are retained from the preceding release benchmark; this refresh remeasured the matched denoising path.
+Preprocessing is identical for both models and therefore reported once.
- `GVHMR`: tracking, ViTPose, image-feature extraction, and motion prediction.
- `Skeleton extraction`: target-view skeleton-condition rendering.
diff --git a/inference.py b/inference.py
index 86de4a4..0a7ad11 100644
--- a/inference.py
+++ b/inference.py
@@ -46,8 +46,8 @@ def inference(
enable_rcp: Use proposal views before generating more than six targets.
The proposal count follows views_per_group.
enable_tcr: Shift view groups cyclically between denoising steps.
- enable_turbo: Use the Wan2.2 5B Turbo LoRA for accelerated denoising.
- Disable it to use the unfused base model.
+ enable_turbo: Whether to use 4DAnyone-Turbo for accelerated denoising.
+ Disable it to use the base 4DAnyone model.
data_dir: Root for reusable GVHMR motion and final 4DAnyone outputs.
model_dir: Model root; missing public checkpoints download here.
checkpoint_path: Local 4DAnyone checkpoint override.