Renumber book 3 images

Note that image 4 and 5 both use img-3.04-cornell-refactor2.jpg

Author: hollasch

books/RayTracingTheRestOfYourLife.html

@@ -1660,7 +1660,7 @@
 see.
 
   ![Image 4: Cornell box, with imperfect PDF
-  ](../images/img-3.02-cornell-refactor2.jpg class=pixel)
+  ](../images/img-3.04-cornell-refactor2.jpg class=pixel)
 
 </div>
 
@@ -1711,7 +1711,7 @@
 material, we'll get a slightly different image.
 
   ![Image 5: Cornell box, with uniform hemispherical sampling
-  ](../images/img-3.02-cornell-refactor2.jpg class=pixel)
+  ](../images/img-3.04-cornell-refactor2.jpg class=pixel)
 
 </div>
 
@@ -2174,7 +2174,7 @@
 Which produces:
 
   ![Image 6: Cornell box, with orthonormal basis scatter function
-  ](../images/img-3.03-cornell-ortho.jpg class=pixel)
+  ](../images/img-3.06-cornell-ortho.jpg class=pixel)
 
 Let’s get rid of some of that noise.
 </div>
@@ -2330,7 +2330,7 @@
 With 10 samples per pixel this yields:
 
   ![Image 7: Cornell box, sampling only the light, 10 samples per pixel
-  ](../images/img-3.04-cornell-sample-light.jpg class=pixel)
+  ](../images/img-3.07-cornell-sample-light.jpg class=pixel)
 
 </div>
 
@@ -2384,7 +2384,7 @@
 This gives us:
 
   ![Image 8: Cornell box, light emitted only in the downward direction
-  ](../images/img-3.05-cornell-lightdown.jpg class=pixel)
+  ](../images/img-3.08-cornell-lightdown.jpg class=pixel)
 
 </div>
 
@@ -2532,7 +2532,7 @@
 `cosine_pdf` class:
 
   ![Image 9: Cornell box with a cosine density PDF
-  ](../images/img-3.06-cornell-cos-pdf.jpg class=pixel)
+  ](../images/img-3.09-cornell-cos-pdf.jpg class=pixel)
 
 </div>
 
@@ -2723,7 +2723,7 @@
 At 10 samples per pixel we get:
 
   ![Image 10: Cornell box, sampling a hittable light, 10 samples per pixel
-  ](../images/img-3.07-hittable-light.jpg class=pixel)
+  ](../images/img-3.10-hittable-light.jpg class=pixel)
 
 </div>
 
@@ -2842,7 +2842,7 @@
 1000 samples per pixel yields:
 
   ![Image 11: Cornell box, mixture density of cosine and light sampling
-  ](../images/img-3.08-cosine-and-light.jpg class=pixel)
+  ](../images/img-3.11-cosine-and-light.jpg class=pixel)
 
 </div>
 
@@ -3201,7 +3201,7 @@
 not sampled with more density.
 
   ![Image 12: Cornell box with arbitrary PDF functions
-  ](../images/img-3.09-arbitrary-pdf.jpg class=pixel)
+  ](../images/img-3.12-arbitrary-pdf.jpg class=pixel)
 
 </div>
 
@@ -3377,7 +3377,7 @@
 expensive!
 
   ![Image 13: Cornell box with glass sphere, using new PDF functions
-  ](../images/img-3.10-cornell-glass-sphere.jpg class=pixel)
+  ](../images/img-3.13-cornell-glass-sphere.jpg class=pixel)
 
 </div>
 
@@ -3462,7 +3462,7 @@
 And we get a decent image with 1000 samples as before:
 
   ![Image 14: Cornell box using a mixture of glass & light PDFs
-  ](../images/img-3.11-glass-and-light.jpg class=pixel)
+  ](../images/img-3.14-glass-and-light.jpg class=pixel)
 
 </div>
 
@@ -3525,7 +3525,7 @@
 Happily, the black specks are gone:
 
   ![Image 15: Cornell box with anti-acne color function
-  ](../images/img-3.12-book3-final.jpg class=pixel)
+  ](../images/img-3.15-book3-final.jpg class=pixel)
 
 </div>