Refine on text

Author: Tom Notch

src/components/MotivationFigures.tsx

@@ -85,7 +85,7 @@ interface FigureDef {
 }
 
 const figures: FigureDef[] = [
-  { src: "Tailored_Planar_Model.svg", label: "Tailored Planar Model" },
+  { src: "Tailored_Planar_Model.svg", label: "Lens-Specific Planar Model" },
   {
     src: "Equivariance_vs_Invariance.svg",
     label: "Equivariance vs Invariance",

src/paper.mdx

@@ -99,7 +99,7 @@ USF scales efficiently to high-resolution spherical imagery and maintains *less
 
 ## Motivation
 
-Most vision pipelines apply planar CNNs that assume a pinhole camera and operate on regular 2D grids. Each camera type (pinhole, fisheye, panoramic) has a unique projection geometry, forcing practitioners to train **separate, lens-specific models**. This is **mathematically unavoidable**: by Gauss's *Theorema Egregium*, no flat projection preserves the intrinsic curvature of the sphere, so any 2D representation inevitably distorts geometry.
+Most vision pipelines apply planar CNNs that assume a pinhole camera and operate on regular 2D grids. Each camera type (pinhole, fisheye, panoramic) has a unique projection geometry, forcing practitioners to train **separate, lens-specific models** that inevitably overfit onto the distortion patterns they see. This is **mathematically unavoidable**: by Gauss's *Theorema Egregium*, no flat projection preserves the intrinsic curvature of the sphere, so any 2D representation inevitably distorts geometry.
 
 <MotivationFigures client:idle />
 
@@ -167,7 +167,7 @@ All benchmarks use a 960×480 panorama input with batch size 8, RGB channels, av
   </TabsContent>
 </Tabs>
 
-**Geometry caching is the key enabler**: once neighborhood structures and interpolation weights are precomputed, sustained runtime drops by **orders of magnitude**. Without caching, geometric preprocessing would dominate and make spherical pipelines infeasible at scale. With caching, spherical networks run at roughly **2×** the wall-clock time of planar counterparts, which we consider acceptable given the added geometric consistency, rotation equivariance, and lens-agnostic processing.
+**Geometry caching is the key enabler**: once neighborhood structures and interpolation weights are precomputed, sustained runtime drops by **orders of magnitude**. Without caching, geometric preprocessing would dominate and make spherical pipelines infeasible at scale. With caching, spherical networks train at roughly **2×** the wall-clock time of planar counterparts, which we consider acceptable given the added geometric consistency, rotation equivariance, and lens-agnostic processing.
 
 ## Citation
 

src/styles/global.css

@@ -168,3 +168,8 @@ main .katex-display {
     @apply cursor-pointer;
   }
 }
+
+.prose p {
+  hyphens: auto;
+  text-align: justify;
+}