critical

check raytracer ongoing energy if direction is scattering-weighted, should the ongoing energy be weighted too? what does this mean for diffuse rain?

Raytracer is slower when waveguide cutoff is high???

Combine fields in condensed_node struct.

Is it worth checking all paths in the image source tree?

Soft source without solution growth.

Only ambisonic output?

app stuff

Finish adding the help panel info. switch to normal tooltips

Benchmark gpu vs cpu performance.

documentation

Compare to previous results from undergrad final commit. Compare code too?

Set a bunch of listening tasks. Connect expectations of the algorithm and compare them to the outputs. Match the audible output to the changes in the code, where possible.

normal

Fuzzing. File loading is especially likely to be vulnerable.

nice-to-have

app

Improve source/reciever + microphone direction models.

Built-in auralization.

Better visualisation colour map.

Save/load presets from file.

Variable speed of sound.

Undo/redo.

Better comparative visualisation of rays + waves

library

Speed up compile times.

Can I use something like a spatial sinc kernel to place the initial impulse at an exact location?

Directional sources.

Real-time simulation.

Ambisonic output/microphone capsules.

Use Boost Units to validate physical quantities.

Look into other approaches for microphone modelling, which don't affect modal response as much.

Closest triangle algo is really dumb + slow. It's fine for the time being, but might be a bottleneck in the future.

Call from the commandline.

Use proper BRDF instead of lambertian diffusion.

Trace properly in non-closed scenes. Possible with anechoic waveguide boundaries? Probably not because 'ghost points' would end up coinciding with air points.

Air absorption - Better calculation method based on temperature, humidity, ambient density, speed of sound. Implement as a post-process in all models.

Automatically align waveguide mesh to model's minimal bounding box.