Add paper and code links, and sample usage

README.md

@@ -1,22 +1,24 @@
 ---
 license: cc-by-4.0
+size_categories:
+- 1K<n<10K
 task_categories:
-  - image-to-3d
-tags:
-  - physical-ai
-  - nurec
-  - 3d-reconstruction
-  - novel-view-synthesis
-  - radiance-fields
-  - photometric-calibration
-  - ncore
+- image-to-3d
 pretty_name: PPISP Dataset
-size_categories:
-  - 1K<n<10K
+tags:
+- physical-ai
+- nurec
+- 3d-reconstruction
+- novel-view-synthesis
+- radiance-fields
+- photometric-calibration
+- ncore
 ---
 
 # PPISP Dataset
 
+[**Project Page**](https://research.nvidia.com/labs/sil/projects/ppisp/) | [**Paper**](https://huggingface.co/papers/2601.18336) | [**Code**](https://github.com/nv-tlabs/ppisp)
+
 ## Dataset Description:
 The PPISP dataset accompanies the work "PPISP: Physically-Plausible Compensation and Control of Photometric Variations in Radiance Field Reconstruction". It contains object-centric scene captures of four outdoor scenes, each captured with three different cameras, for multi-view 3D reconstruction and novel view synthesis. The photos were captured with exposure bracketing of +/-2 EV and re-processed with automatic exposure and color corrections to create a challenging benchmark for methods that compensate for photometric inconsistencies.
 
@@ -26,6 +28,49 @@ The dataset is provided in two formats: COLMAP (`colmap/`) and NCore V4 (`ncore/
 
 This dataset is ready for commercial/non-commercial use.
 
+## Sample Usage
+
+The following snippet demonstrates how to integrate the PPISP module into a radiance field reconstruction pipeline, as described in the official repository:
+
+```python
+from ppisp import PPISP, PPISPConfig
+
+# 1. Initialize
+ppisp = PPISP(num_cameras=3, num_frames=500)
+
+# 2. Create optimizers and scheduler
+ppisp_optimizers = ppisp.create_optimizers()
+ppisp_schedulers = ppisp.create_schedulers(ppisp_optimizers, max_optimization_iters)
+
+# 3. Training loop
+for step in range(max_optimization_iters):
+    # Render raw RGB from your radiance field
+    rgb_raw = renderer(camera_idx, frame_idx)  # [..., 3]
+
+    # Apply PPISP post-processing
+    rgb_out = ppisp(
+        rgb_raw,
+        pixel_coords,           # [..., 2]
+        resolution=(W, H),
+        camera_idx=camera_idx,
+        frame_idx=frame_idx,
+    )
+
+    # Add PPISP regularization loss to other losses
+    loss = reconstruction_loss(rgb_out, rgb_gt) + ppisp.get_regularization_loss()
+    loss.backward()
+
+    # Step optimizers and scheduler
+    for opt in ppisp_optimizers:
+        opt.step()
+        opt.zero_grad(set_to_none=True)
+    for sched in ppisp_schedulers:
+        sched.step()
+
+# 4. Novel view rendering: pass camera_idx as usual, frame_idx=-1
+rgb_out = ppisp(rgb_raw, pixel_coords, resolution=(W, H), camera_idx=camera_idx, frame_idx=-1)
+```
+
 ## Dataset Owner(s):
 NVIDIA Corporation
 
@@ -58,9 +103,20 @@ We used COLMAP to reconstruct camera poses and a sparse point cloud for each seq
 
 Measurement of total data storage: About 14.3 GB in compressed form (8.1 GB COLMAP format + 6.2 GB NCore V4 format).
 
-## Reference(s):
-[Project Page](https://research.nvidia.com/labs/sil/projects/ppisp/)
+## Citation
+
+```bibtex
+@misc{deutsch2026ppispphysicallyplausiblecompensationcontrol,
+      title={PPISP: Physically-Plausible Compensation and Control of Photometric Variations in Radiance Field Reconstruction}, 
+      author={Isaac Deutsch and Nicolas Moënne-Loccoz and Gavriel State and Zan Gojcic},
+      year={2026},
+      eprint={2601.18336},
+      archivePrefix={arXiv},
+      primaryClass={cs.CV},
+      url={https://arxiv.org/abs/2601.18336},
+}
+```
 
 ## Ethical Considerations:
-NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications.  When downloaded or used in accordance with our terms of service, developers should work with their internal team to ensure this dataset meets requirements for the relevant industry and use case and addresses unforeseen product misuse.   
-Please report quality, risk, security vulnerabilities or NVIDIA AI Concerns here.  
+NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal team to ensure this dataset meets requirements for the relevant industry and use case and addresses unforeseen product misuse.   
+Please report quality, risk, security vulnerabilities or NVIDIA AI Concerns here.