The small toy 3D software rasterizer using SDL2 library. This project seems to be built well only on Windows yet.

You can build this project with Makefile. To build it, just type

make all

in command.

By modifying main.cc source file and rebuilding project, you can change what is rendered. The user(programmer) is intended to write the code with including only "mimic/include/mimic.hh" header file which provides enough operations for rendering. For more details about the operations this project provides, see mimic/include/mimic.hh and its comments.

The whole rendering pipeline, from vertex processing stage to per-sample operation stage, is mimicked by calling DrawArray(const uint32_t start, const uint32_t number, DrawingType type) function in mimic/include/mimic.hh, with data of context instance(global) which is set through other function calls before calling this function.

Behavior of this function varies slightly with the type parameter, but overall behavior is almost similar.

In this stage, the function calculates the position of vertex, by calling vertex shader, per each vertex. Vertex shader is descripted by user and stored in context instance previously by user calling void set_shaders(const uint32_t out_size, void (*vertex_shader)(float*[], mmath::Vec4<float>*, float[]), color_t (*fragment_shader)(struct Fragment*)) function.

In this stage, some operations are conducted in order to fit vertex shader output for the pixel of output window. Perspective division(i.e. dividing xyz-coordinate by w-coordinate) is conducted in this stage.

In case of lines, the function clips the line preemptively and adjusts the coordinates and attributes.

And in case of triangles, back-face culling is conducted.

The primitives, composed of vertices, are converted into fragments(pixel) in this stage. Namely, the rasterizer determines which pixel is contained in each shape, such as triangle or line. For this, it conducts DDA algorithm for lines, and conducts naive algorithm(checking for all pixel if it is in primitive) for triangles.

I plan to improve it more efficiently soon.

The fragments, created in Scan Conversion stage, are shaded in this stage, in the way which fragment shader describes. Like Vertex Processing stage, fragment shader is descripted by user and stored in context instance previously by user calling void set_shaders(const uint32_t out_size, void (*vertex_shader)(float*[], mmath::Vec4<float>*, float[]), color_t (*fragment_shader)(struct Fragment*)) function.

Lastly, the shaded fragments are filtered by z-buffering, so that only the frontmost fragment are displayed at rendering window in this stage.

Implementation of phong shading, through using this project.