Update index.html

docs/index.html

@@ -6,7 +6,7 @@
         content="Compact3D reduces the storage memory requirements of 3D Gaussian Splatting models.">
   <meta name="keywords" content="3D Gaussian Splat, Compact3D, Comp3D">
   <meta name="viewport" content="width=device-width, initial-scale=1">
-  <title>Compact3D: Compressing Gaussian Splat Radiance Field Models with Vector Quantization</title>
+  <title>Compact3D: Smaller and Faster Gaussian Splatting with Vector Quantization</title>
 
   <!-- Global site tag (gtag.js) - Google Analytics -->
   <script async src="https://www.googletagmanager.com/gtag/js?id=G-PYVRSFMDRL"></script>
@@ -49,7 +49,7 @@
     <div class="container is-max-desktop">
       <div class="columns is-centered">
         <div class="column has-text-centered">
-          <h1 class="title is-1 publication-title">Compact3D: Compressing Gaussian Splat Radiance Field Models with Vector Quantization</h1>
+          <h1 class="title is-1 publication-title">Compact3D: Smaller and Faster Gaussian Splatting with Vector Quantization</h1>
           <div class="is-size-5 publication-authors">
             <span class="author-block">
               <a href="https://klnavaneet.github.io/">KL Navaneet</a><sup>*</sup>,</span>
@@ -140,18 +140,19 @@ <h2 class="title is-3">Abstract</h2>
           <p>
             3D Gaussian Splatting is a new method for modeling and rendering 3D radiance fields that achieves much faster 
             learning and rendering time compared to SOTA NeRF methods. However, it comes with a drawback in the much larger 
-            storage demand compared to NeRF methods since it needs to store the parameters for several 3D Gaussians.
+            storage demand compared to NeRF methods since it needs to store the parameters for millions of 3D Gaussians.
           </p>
           <p>
-            We notice that many Gaussians may share similar parameters, so we introduce a simple vector quantization method
-             based on kmeans algorithm to quantize the Gaussian parameters. Then, we store the small codebook along with the
-              index of the code for each Gaussian. Moreover, we compress the indices further by sorting them and using a 
-              method similar to run-length encoding. 
+            We notice that large groups of Gaussians share similar parameters and introduce a simple vector quantization 
+            method based on K-means algorithm to quantize the Gaussian parameters. Then, we store the small codebook along 
+            with the index of the code for each Gaussian. We compress the indices further by sorting them and using a method 
+            similar to run-length encoding. Moreover, we use a simple regularizer that encourages zero opacity 
+            (invisible Gaussians) to reduce the number of Gaussians, thereby compressing the model and speeding up the rendering. 
           </p>
           <p>
-              We do extensive experiments on standard benchmarks as well as a new benchmark which is an order of magnitude larger
-               than the standard benchmarks. We show that our simple yet effective method can reduce the storage cost for the 
-               original 3D Gaussian Splatting method by a factor of almost 20× with a very small drop in the quality of rendered images.
+            We do extensive experiments on standard benchmarks as well as an existing 3D dataset that is an order of magnitude 
+            larger than the standard benchmarks used in this field. We show that our simple yet effective method can reduce the 
+            storage costs for 3DGS by 40 to 50x and rendering time by 2 to 3x with a very small drop in the quality of rendered images.
           </p>          
         </div>
       </div>
@@ -168,28 +169,31 @@ <h2 class="title is-3">How It Works</h2>
     <div class="w3-display-container w3-row w3-white w3-margin-bottom w3-center">
         <embed src="./static/images/teaser_new.png" style="width:70%">
     </div>
-    <p>We compress the 3D Gaussian Splatting (3DGS) model using vector quantization of the parameters of the Gaussians. The quantization
+    <p>We compress the 3D Gaussian Splatting (3DGS) model by (1) compressing the parameters of each Gaussian and (2) reducing the total number of 
+      Gaussians. To compress the parameters, we use a simple K-means based vector quantization. The quantization
       is performed along with the training of the Gaussian parameters. Considering each Gaussian as a vector, we perform K-means clustering 
-      on the covariance and color parameters of all Gaussians to represent the N Gaussians in the model with k cluster centers (codes). 
+      on the covariance and color parameters to represent the N Gaussians in the model with k cluster centers (codes). 
       Each Gaussian is then replaced by its corresponding code for rendering and loss calculation. The gradients with respect to centers are copied 
       to all the elements in the corresponding cluster and the non-quantized versions of the parameters are updated. Only the codebook 
       and code assignments for each Gaussian are stored and used for inference. Our method, CompGS, maintains the real-time rendering property of 
-      3DGS while compressing it by an order of magnitude.
+      3DGS while compressing it by an order of magnitude. During training, we also encourage the Gaussians to be transparent by regularizing the 
+      opacity parameter. Highly transparent Gaussians are regularly pruned, resulting in fewer total Gaussians at the end of training. This further 
+      reduces the storage and also greatly speeds up the training and rendering.  
     </p>
 
     <hr>
     <div class="w3-left-align w3-margin-bottom">
         <h2 class="title is-3">Comparison with SOTA Methods</h2>
     </div>
     <div class="w3-display-container w3-row w3-white w3-margin-bottom w3-center">
-        <embed src="./static/images/table_1.png" style="width:80%">
+        <embed src="./static/images/table_1_v2.png" style="width:80%">
     </div>
     <p>3DGS performs comparably or outperforms the best of the
       NeRF based approaches while maintaining a high rendering speed during inference. Trained NeRF models are significantly smaller than
       3DGS since NeRFs are parameterized using neural networks while 3DGS requires storage of parameters of millions of 3D Gaussians.
-      Our method, CompGS, is a vector quantized version of 3DGS that maintains the speed and performance advantages of 3DGS while being an order of
-      magnitude smaller. We report the averaged FPS and memory over all datasets. CompGS is identical to 3DGS during inference and thus
-      has the same FPS. ∗ Reproduced using official code. † Reported from 3DGS.
+      Our method, CompGS, is a vector quantized version of 3DGS that maintains the speed and performance advantages of 3DGS while being 40-50x 
+      smaller. ∗ Reproduced using official code. † Reported from 3DGS. Our timings for 3DGS and CompGS are reported using a RTX6000
+      GPU while those with † used A6000 GPU. We boldface entries for emphasis.
     </p>
 
 
@@ -242,14 +246,13 @@ <h2 class="title is-3">Results on the Large-scale ARKit Dataset</h2>
     <div class="w3-left-align w3-margin-bottom">
         <h2 class="title is-3">Conclusion</h2>
     </div>       
-    <p>3D Gaussian Splatting efficiently models
-      3D radiance fields, outperforming NeRF in learning and
-      rendering efficiency at the cost of increased storage.
-      To reduce storage demands, we apply k-means-based
-      vector quantization, compressing indices and employing a
-      compact codebook. Our method cuts the storage cost of
-      3D Gaussian Splatting by almost 20×, maintaining image
-      quality across benchmarks.
+    <p>3D Gaussian Splatting efficiently models 3D radiance fields, outperforming NeRF
+       in learning and rendering efficiency at the cost of increased storage. To reduce 
+       storage demands, we apply opacity regularization and K-means-based vector quantization, 
+       compressing indices and employing a compact codebook. Our
+       method cuts the storage cost of 3DGS by almost 45×, increases rendering FPS
+       by 2.5× while maintaining image quality across benchmarks.
+
     </p>
 
   </div>
@@ -262,7 +265,7 @@ <h2 class="title is-3">Conclusion</h2>
         <h2 class="title is-3">BibTeX</h2>
     </div>
     <pre><code>@article{navaneet2023compact3d,
-          title={Compact3D: Compressing Gaussian Splat Radiance Field Models with Vector Quantization},
+          title={Compact3D: Smaller and Faster Gaussian Splatting with Vector Quantization},
           author={Navaneet, KL and Meibodi, Kossar Pourahmadi and Koohpayegani, Soroush Abbasi and Pirsiavash, Hamed},
           journal={arXiv preprint arXiv:2311.18159},
           year={2023}
@@ -276,7 +279,7 @@ <h2 class="title is-3">BibTeX</h2>
     <h2 class="title">BibTeX</h2>
     <pre><code>@inproceedings{navaneet2023compact3d,
  author = {Navaneet, K L and Pourahmadi Meibodi, Kossar and Koohpayegani, Soroush Abbasi and Pirsiavash, Hamed},
- title = {Compact3D: Compressing Gaussian Splat Radiance Field Models with Vector Quantization},
+ title = {Compact3D: Smaller and Faster Gaussian Splatting with Vector Quantization},
  year = {2023}
 }</code></pre>
   </div>