Upload README.md

README.md

@@ -94,6 +94,12 @@ homepage: https://vlar-group.github.io/PhysInOne.html
     line-height: 1.16;
     font-weight: 700;
   }
+  .physinone-card pre {
+    margin-top: 0.25em;
+  }
+  .physinone-card p + pre {
+    margin-top: 0.15em;
+  }
 </style>
 
 <h1  align="center" style="font-size: 42px;">PhysInOne: Visual Physics Learning and Reasoning in One Suite</h1>
@@ -280,7 +286,7 @@ This design guarantees that training, validation, and test sets are fully separa
 <div style="margin-left: 24px;">
 
 
-<h3 id="physical-phenomena" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.35em;">&bull;</span>Physical Phenomenon and Abbreviations</h3>
+<h3 id="physical-phenomena" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.18em;">&bull;</span>Physical Phenomenon and Abbreviations</h3>
 <div style="margin-left: 24px;">
 
 
@@ -381,7 +387,7 @@ This naming convention allows users to quickly parse scene metadata and link it
 
 PhysInOne provides synchronized visual and physical annotations for each dynamic 3D scene.
 
-<h3 id="depth" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.35em;">&bull;</span>Depth</h3>
+<h3 id="depth" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.18em;">&bull;</span>Depth</h3>
 <div style="margin-left: 24px;">
 
 
@@ -389,7 +395,7 @@ Depth maps are provided in `.npz` format and are expressed in meters.
 
 </div>
 
-<h3 id="segmentation" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.35em;">&bull;</span>Segmentation</h3>
+<h3 id="segmentation" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.18em;">&bull;</span>Segmentation</h3>
 <div style="margin-left: 24px;">
 
 
@@ -405,7 +411,7 @@ Expected encoding:
 
 </div>
 
-<h3 id="captions" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.35em;">&bull;</span>Captions</h3>
+<h3 id="captions" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.18em;">&bull;</span>Captions</h3>
 <div style="margin-left: 24px;">
 
 
@@ -413,7 +419,7 @@ Each scene includes a `caption.txt` file containing an English paragraph that de
 
 </div>
 
-<h3 id="trajectories" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.35em;">&bull;</span>Trajectories</h3>
+<h3 id="trajectories" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.18em;">&bull;</span>Trajectories</h3>
 <div style="margin-left: 24px;">
 
 
@@ -421,7 +427,7 @@ TODO
 
 </div>
 
-<h3 id="cameras" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.35em;">&bull;</span>Cameras</h3>
+<h3 id="cameras" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.18em;">&bull;</span>Cameras</h3>
 <div style="margin-left: 24px;">
 
 
@@ -430,7 +436,7 @@ Each scene includes multiple camera viewpoints to support 3D perception, reconst
 - Taichi-based MPM simulations (elastic solids, plasticine, granular substances, some Newtonian and Non-Newtonian fluids): 15 static cameras + 1 dynamic camera
 - Other scenes: 12 static cameras + 1 dynamic camera
 
-<h5 id="camera-json-structure" style="font-size: 1.02em; margin-top: 0.55em; margin-bottom: 0.12em; line-height: 1.16; font-weight: 700; margin-left: 0;"><span style="display: inline-block; width: 1.25em;">&#9702;</span>Camera JSON structure</h5>
+<h5 id="camera-json-structure" style="font-size: 1.02em; margin-top: 0.55em; margin-bottom: 0.12em; line-height: 1.16; font-weight: 700; margin-left: 0;"><span style="display: inline-block; width: 1.10em;">&#9702;</span>Camera JSON structure</h5>
 
 <div style="margin-left: 18px;">
 
@@ -444,7 +450,7 @@ Per camera:
 
 </div>
 
-<h5 id="dynamic-camera-behavior" style="font-size: 1.02em; margin-top: 0.55em; margin-bottom: 0.12em; line-height: 1.16; font-weight: 700; margin-left: 0;"><span style="display: inline-block; width: 1.25em;">&#9702;</span>Dynamic camera behavior</h5>
+<h5 id="dynamic-camera-behavior" style="font-size: 1.02em; margin-top: 0.55em; margin-bottom: 0.12em; line-height: 1.16; font-weight: 700; margin-left: 0;"><span style="display: inline-block; width: 1.10em;">&#9702;</span>Dynamic camera behavior</h5>
 
 <div style="margin-left: 18px;">
 
@@ -453,7 +459,7 @@ Per camera:
 
 </div>
 
-<h3 id="point-clouds" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.35em;">&bull;</span>Point Clouds</h3>
+<h3 id="point-clouds" style="font-size: 1.35em; margin-top: 0.75em; margin-bottom: 0.22em; line-height: 1.2; margin-left: 24px;"><span style="display: inline-block; width: 1.18em;">&bull;</span>Point Clouds</h3>
 <div style="margin-left: 24px;">
 
 
@@ -487,6 +493,7 @@ The `filter_cases.py` script supports selection by:
 
 Please download the `assets` folder, which contains:
 
+<div style="margin-top: -0.35em;">
 ```text
 metadata/
   repo_assignment.txt
@@ -495,6 +502,7 @@ scripts/
   filter_cases.py
   download.py
 ```
+</div>
 
 </div>
 
@@ -502,9 +510,11 @@ scripts/
 <div style="margin-left: 24px;">
 
 
+<div style="margin-top: -0.35em;">
 ```bash
 pip install huggingface_hub tqdm
 ```
+</div>
 
 </div>
 
@@ -545,6 +555,7 @@ python scripts/filter_cases.py \
 
 By default, phenomenon matching uses `contains` mode. For example, the following command selects all double-physics cases that contain `FrictionStop`:
 
+<div style="margin-top: -0.35em;">
 ```bash
 python scripts/filter_cases.py \
   --split train \
@@ -553,9 +564,11 @@ python scripts/filter_cases.py \
   --match_mode contains \
   --output selected_cases.json
 ```
+</div>
 
 For exact matching, use `--match_mode exact`. Order does not matter. For example, this command selects cases whose phenomenon set is exactly `{AccelConcaveSpin, AccelSurfaceSpin}`:
 
+<div style="margin-top: -0.35em;">
 ```bash
 python scripts/filter_cases.py \
   --split train \
@@ -564,6 +577,7 @@ python scripts/filter_cases.py \
   --match_mode exact \
   --output selected_cases.json
 ```
+</div>
 
 <div style="margin-left: 24px;" data-subsection="random-sampling">
 
@@ -574,6 +588,7 @@ python scripts/filter_cases.py \
 
 `--num` is the global number of cases sampled after filtering. The default random seed is `42`.
 
+<div style="margin-top: -0.35em;">
 ```bash
 python scripts/filter_cases.py \
   --split train \
@@ -583,6 +598,7 @@ python scripts/filter_cases.py \
   --seed 42 \
   --output selected_cases.json
 ```
+</div>
 
 </div>
 
@@ -595,32 +611,39 @@ python scripts/filter_cases.py \
 
 Each selected case is downloaded as a zip file from the corresponding Hugging Face shard repository.
 
+<div style="margin-top: -0.35em;">
 ```bash
 python scripts/download.py \
   --selection selected_cases.json \
   --output_dir ./PhysInOne
 ```
+</div>
 
 If the dataset repositories are gated or private, pass a Hugging Face token:
 
+<div style="margin-top: -0.35em;">
 ```bash
 python scripts/download.py \
   --selection selected_cases.json \
   --output_dir ./PhysInOne \
   --token YOUR_HF_TOKEN
 ```
+</div>
 
 By default, downloaded zip files are saved into a flat output folder. To preserve shard/split/activity structure:
 
+<div style="margin-top: -0.35em;">
 ```bash
 python scripts/download.py \
   --selection selected_cases.json \
   --output_dir ./PhysInOne \
   --keep_shard_structure
 ```
+</div>
 
 The preserved structure is:
 
+<div style="margin-top: -0.35em;">
 ```text
 PhysInOne/
   physinone_part1/
@@ -628,6 +651,7 @@ PhysInOne/
       DoublePhysics/
         AccelConcaveSpin_AccelSurfaceSpin__bg070__K5ER39_trajectory.zip
 ```
+</div>
 
 </div>
 
@@ -656,6 +680,7 @@ In total, assets comply with licenses including CC BY-NC, CC BY-SA, CC BY-NC-SA,
 
 If you use PhysInOne in your research, please cite:
 
+<div style="margin-top: -0.35em;">
 ```bibtex
 @article{zhou2026physinone,
   title={PhysInOne: Visual Physics Learning and Reasoning in One Suite},
@@ -664,6 +689,7 @@ If you use PhysInOne in your research, please cite:
   year={2026}
 }
 ```
+</div>
 
 </div>