Roadmap

This document describes the hight level roadmap of the Aqsis product line. We have gathered user and developer input via a survey, this roadmap defines the upcoming releases based on that feedback.

This is subject to change.

The following information details an overview of the area's of improvement for each planned release. This information is meant only to provide a high level picture of the release, please refer to the various design pages for more detail on the planned changes.

• AqsisTex
• Piqsl
• Eqsl
• Smooth shading interpolation
• OSX bundle deliverable
• Linux desktop integration
• Internal shader preprocessor

• Autoshadows
• Miqser
• Performance improvements
• Bug fixing
• Layered Shaders
• OpenEXR support
• Blobbies support
• Facevertex support

Now that 1.2.0 is released, see here for more details of the final changes, or the ChangeLog.txt for even finer detail.

Proposed features can be found below, which should ultimately find their way into a planned/sheduled release…

1. Outstanding bugs (which also need to be fixed in the 1.0 tree).
   1. inter-grid cracking problem
   2. adaptive mesh refinement when rendering subdivision surfaces (must handle large base meshes with large variation in polygon size without exhibiting cracks)
   3. Du, Dv when shading polygon primitives
2. Refactor major code sections. A lot of the codebase has evolved into it's current state, now is a good time to refactor a lot of the internals, cleaning up interfaces and implementation details.
3. Sampling performance. This includes lots of profiling and looking at the representation of micropolygons, buckets and grids.
4. Shading performance. This will probably involve a partial rewrite of the shader compiler and shader execution engine.
5. Win32 version compiling with the latest MS toolset. Microsoft Visual Studio .NET 2003.
6. Raytracing. This is a necessary prerequisite for global illumination (which is a big job on its own), but is interesting enough in its own right that a working raytracer would be enough reason to release a new major version release.
7. RiGeometricApproximation(RI_FLATNESS). This is a small job, but it's very important.
8. RiShadingInterpolation(RI_SMOOTH). Again, a small job but important. This is something to look at while we're looking at sampling performance.
9. Global illumination and raytraced ambient occlusion. These two kind of go together, because irradiance caches and occlusion caches are so similar. Once you've implemented one, you kind of have the other for almost free.
   1. This assertion may not be true on reflection. Current thinking is that BrickMaps(?) are superior to k-d trees for a combined irradiance cache/photon map, but this may not be true for occlusion caches. But then again, a brick map-like structure for occlusion caches might be better. We'll see when we get there, I guess.
10. Deep Shadow Maps
11. Soft shadows.
12. Better shader compiler, compiling to C++.
13. New RIB parser, supporting inline archives and RIB filters.
14. Better support for Image Based Lighting.
15. librx support for providing feedback for DSOs.
16. “Layer Rendering.” Using some mechanism like AOVs, it would be nice to render different objects into different images in a way that could be composited later.
17. It would be handy if different AOVs could be sampled differently, say, using different PixelSamples settings. An example is rendering a “Beauty Pass” and a depth map in one render. The depth map should be sampled 1×1 without jittering.
18. Multi-threading.
19. 64-bit port. (Opteron?)
20. It might be handy if motion vectors could be output as an AOV. Not sure if this is useful at all.
21. OpenGL/GPU based rendering, for super fast render times, like Parthenon.
22. OpenGL previewer?
23. Support for more output file formats, such as OpenEXR and Cineon/DPX.
24. Support for Attribute “dice” “motionfactor” - increase shading rate based on motion blur/DoF.