CIS565-Fall-2018 · hanbollar · Oct 10, 2018 · Oct 16, 2018 · Oct 18, 2018 · Oct 18, 2018
diff --git a/README.md b/README.md
@@ -1,21 +1,93 @@
 CUDA Rasterizer
 ===============
 
-[CLICK ME FOR INSTRUCTION OF THIS PROJECT](./INSTRUCTION.md)
 
 **University of Pennsylvania, CIS 565: GPU Programming and Architecture, Project 4**
 
-* (TODO) YOUR NAME HERE
-	* (TODO) [LinkedIn](), [personal website](), [twitter](), etc.
-* Tested on: (TODO) Windows 22, i7-2222 @ 2.22GHz 22GB, GTX 222 222MB (Moore 2222 Lab)
 
-### (TODO: Your README)
 
-*DO NOT* leave the README to the last minute! It is a crucial part of the
-project, and we will not be able to grade you without a good README.
+Hannah Bollar: [LinkedIn](https://www.linkedin.com/in/hannah-bollar/), [Website](http://hannahbollar.com/)
 
 
-### Credits
+
+Tested on: Windows 10 Pro, i7-6700HQ @ 2.60GHz 15.9GB, GTX 980M (Personal)
+
+____________________________________________________________________________________
+
+![Developer](https://img.shields.io/badge/Developer-Hannah-0f97ff.svg?style=flat) ![CUDA 8.0](https://img.shields.io/badge/CUDA-8.0-yellow.svg) ![Built](https://img.shields.io/appveyor/ci/gruntjs/grunt.svg) ![Issues](https://img.shields.io/badge/issues-none-green.svg)
+
+[//]: # (https://img.shields.io/badge/implementation-in%20progress-orange.svg)
+
+[Visuals](#visuals) - [Features](#all-current-features) - [Optimizations](#optimizations) - [Bloopers](#bloopers) - [References](#references) 
+
+# Rasterizer
+
+### Visuals
+
+Triangles | Normals | Lines | Points
+:-------------------------:|:-------------------------:|:-------------------------:|:---------------------:
+![](images/duck_tex.png)| ![](images/duck_nor.png)| ![](images/duck_lines.png)| ![](images/duck_point.png)|
+![](images/cow_nor.png)| ![](images/cow_nor.png)| ![](images/cow_lines.png)| ![](images/cow_points.png)|
+![](images/milktruck_tex.png)| ![](images/milktruck_nor.png)| ![](images/milktruck_lines.png)| ![](images/milktruck_points.png)
+
+`*` note that the cow had no texture nor a base color, so it's texture was rendered by normals based on my implementation.
+`**` not all textures are double sided, so I have the normals acting as a base color in place for no texture, just for visual appeal.
+
+### All Current Features
+
+The flags of all toggleable features can be updated in `rasterize.cu`.
+
+Graphics Features
+- Vertex Shading
+- Primitive Assembly
+	- Triangles
+	- Lines*
+	- Points*
+- Rasterization
+- Fragment Shading
+	- Lambert
+	- Regular Texture Mapping*
+	- Bilinear Interpolated Texturing*
+- Use of Depth Buffer for Frag Comparisons
+	- Mutex to prevent race conditions
+- BackFace Culling*
+
+`*` = additional features.
+
+### Optimizations
+
+![](images/raster_runtimes.png)
+
+![](images/raster_runtimes_data.png)
+
+Here it's noticeable that the Primitive assembly is most often the costliest action. Having Back Face Culling and Mutex are the most beneficial optimizations, and adding in the overhead for bilinear interpolated texturing is extremely costly in comparison, even with the optimizations it is still slower than the runtime that doesnt have bilinear interpolation but also doesnt have any optimizations.
+
+![](images/trivslinevspoint.png)
+
+![](images/trivslinevspoint_data.png)
+
+There's heavy overhead for my line vertex transforms and primitive assembly because I have a helper method there for readability and the others dont have this.
+
+![](images/demo_persp_div.gif)
+
+The above gif demos perspective divide - a rasterizing technique to force proper visual perspective based on 1/z values. Notice how the top of the truck, though geometrically is the same width x height dimensions as the bottom, when the top is farther away than the bottom to the viewer, you can see it's square outline fits inside the bottom's square outline.
+
+![](images/nomutex.gif)
+
+The above gif demos what happens when mutex is turned off. Race conditions create a flickering that shouldnt always be there. It occurs because without the atomic mutex check, the fragment threads are running simultaneously and doing depth checks based on the value currently in the fragment buffer at a location. Since theyre running simulatiously, it's not guaranteed that when one checks a depth at that index, that that is the same depth being overwritten by the time that thread gets to filling in that fragment buffer's index - aka a race condition. Using mutex allows us to wait until the fragment is available to be checked, then it does a depth check as a normal rasterizer should.
+
+### Bloopers
+
+Bad Back Face Culling | Improper Line Rasterizing | Overblown Bilinear Texturing
+:-------------------------:|:-------------------------:|:-------------------------:
+![](images/incorrectbackface.png)| ![](images/linerasterizingwrong.png)| ![](images/correcttexturing_shadedglaretoolarge.png)
+
+- bad back face culling - my directional check was reversed mistakenly
+- improper line rasterizing blooper - I had clamped to an int too late, so my indexing was out of the appropriate bounds after I had already done my bounds check.
+- overblown bilinear texturing - never properly casted to int, yielded same mathematical indexing error as improper line rasterizing blooper but visual was different because of differeing visualization scheme
+
+### References
 
 * [tinygltfloader](https://github.com/syoyo/tinygltfloader) by [@soyoyo](https://github.com/syoyo)
 * [glTF Sample Models](https://github.com/KhronosGroup/glTF/blob/master/sampleModels/README.md)
+* [bilinear filtering](https://en.wikipedia.org/wiki/Bilinear_filtering)
diff --git a/renders/2-cylinder-engine-load-correct.png → ...enders/2-cylinder-engine-load-correct.png b/renders/2-cylinder-engine-load-correct.png → ...enders/2-cylinder-engine-load-correct.png
diff --git a/renders/duck-diffuse-texture.png → given_renders/duck-diffuse-texture.png b/renders/duck-diffuse-texture.png → given_renders/duck-diffuse-texture.png
diff --git a/renders/triangle.png → given_renders/triangle.png b/renders/triangle.png → given_renders/triangle.png
diff --git a/images/correcttexturing_shadedglaretoolarge.png b/images/correcttexturing_shadedglaretoolarge.png
diff --git a/images/cow_lines.png b/images/cow_lines.png
diff --git a/images/cow_nor.png b/images/cow_nor.png
diff --git a/images/cow_points.png b/images/cow_points.png
diff --git a/images/demo_persp_div.gif b/images/demo_persp_div.gif
diff --git a/images/duck_lines.png b/images/duck_lines.png
diff --git a/images/duck_nor.png b/images/duck_nor.png
diff --git a/images/duck_point.png b/images/duck_point.png
diff --git a/images/duck_tex.png b/images/duck_tex.png
diff --git a/images/incorrectbackface.png b/images/incorrectbackface.png
diff --git a/images/linerasterizingwrong.png b/images/linerasterizingwrong.png
diff --git a/images/milktruck_lines.png b/images/milktruck_lines.png
diff --git a/images/milktruck_nor.png b/images/milktruck_nor.png
diff --git a/images/milktruck_points.png b/images/milktruck_points.png
diff --git a/images/milktruck_tex.png b/images/milktruck_tex.png
diff --git a/images/nomutex.gif b/images/nomutex.gif
diff --git a/images/raster_runtimes.png b/images/raster_runtimes.png
diff --git a/images/raster_runtimes_data.png b/images/raster_runtimes_data.png
diff --git a/images/trivslinevspoint.png b/images/trivslinevspoint.png
diff --git a/images/trivslinevspoint_data.png b/images/trivslinevspoint_data.png