Basic support for intersecting against hard-coded analytic shapes for spheres and cubes. Lighting was direct illumination from point lights with shadows.

Basic support for intersecting against hard-coded analytic shapes for spheres and cubes. Lighting was point lights with shadows with Whitted raytracing for reflection.

Basic support for intersecting against triangle meshes (but without acceleration structures yet!) for generated cubes and planes, with analytical spheres, pointlights and Whitted raytracing.

Initial support for importing arbitrary triangle polygon meshes (from .3ds files) without smoothed normals and intersecting against them (still without acceleration structures yet, so very slow!).

Support for smooth mesh normals on arbitrary triangle meshes (calculated automatically), and added basic KDTree acceleration structure support for each polygon mesh to speed up ray/triangle intersection.

Initial support for quad area light sampling, allowing soft-shadows.

Added scene-level KDTree (on top of shape-level for polymeshes), providing much faster ray intersection of multiple objects.

Converted to use bucketed (tiled) rendering, instead of previous scanline approach, which improved interactivity a bit. Also added initial support for procedural textures (checkerboard).

Added support for procedural gridlines texture.

Initial refraction support with Whitted raytracing, but without any Fresnel attenuation (just fixed-function).

Initial (and slightly broken: occlusion ray distribution was wrong/uneven, which can be clearly seen in the render with bands of occlusion rather than the expected falloff with distance) Ambient Occlusion support.

Fixed ambient occlusion support, and added support for compound (grouped) shape hierarchies.

Basic support for image texturing and bump-mapping, and added initial support for .obj format loading.

Better support for reading .obj files and .mtl material files.

Full support for geometry and object (compound) instancing, with trivial ref-counting/COW duplication in the UI.

Full uni-directional pathtracing light transport integration, also with DOF via camera circle-of-confusion sampling.

Better, more physically-correct BSDFs for materials, along with initial Environment (IBL) light support (but without importance sampling at this point).

Further tweaks to BSDFs, adding basic fixed-function layered materials under either clearcoat or microfacet lobe. Model is Stanford Dragon.

Added support for luminance-based importance sampling for Environment lights, reducing variance quite considerably. Model is Stanford Dragon.

Added plausible material presets for different metals (not totally accurate at this point) with conductor BSDFs, and other tweaks to other BSDFs and presets. Models are Stanford Dragon and Headus Killeroo (http://www.headus.com.au/).

Added ability in the UI to easily generate blocks of instances in the shape of meshes. Source model is Stanford Dragon.

Not visible in render, but used to generate it: better UI, with improved transform manipulators and more robust undo/redo support. Model is Katana DCC example robot.

Added (uniform, non-grid based) subdivision surface support (Catmull-Clark) with a custom half-edge mesh structure.

Added displacement support to allow modifying point positions based on texture inputs (image and procedural).

Fixed (code had existed for years, but was wrong beforehand) Oren Nayar rough diffuse model, allowing better representation of rough diffuse surfaces which are more 'realistic' than standard Lambertian models.

Initial basic homogeneous volume scattering support, with single scattering (I think?) with Henyey Greenstein phase function, and a basic diffuse material medium boundary. Model is Stanford Dragon.

Slightly more correct homogeneous volume scattering to approximate Sub Surface Scattering, although material medium boundary setup still isn't physically-correct at this point. Model is Stanford Dragon.

Initial support for intersecting against a curve primitive, allowing hair/fur rendering.

Initial support for heterogeneous volume scattering support with voxel grids. At this stage with just biased ray-marching for density integration.

Added support for multiple scattering (although this render was still quite limited for cloud scattering, and the IBL luminance was clamped, as MIS hadn't been implemented yet), and sparse voxel grid support, allowing much better memory efficiency for large voxel sims.

An attempt to render the Corona Render demo scene with Imagine.

Integration of Imagine into the Katana DCC as a Renderer plugin, allowing side-by-side comparison with Arnold and PRMan on production scenes with the same input geometry / layouts.

Support for complex shading, allowing per-hitpoint BSDF lobe selection and layering. Also added lazy image loading and UDIM support.