Why is the IOR for volumes scalar?

[thomasmoore5]

The index of refraction of many real world volumes varies by wavelength. How come there is an IOR value, but no dispersion parameter or RGB IOR?

[CodeFHD]

LuxCore is currently not a full spectral engine, i.e. the wavelength is not carried as a property of the sample across multiple bouncing rays.

The dispersion feature offered by the glass material is a workaround that enables some artistic applications like a spectrum from a prism, but it has side effects which are documented in here: LuxCoreRender/LuxCore#262

The only thing a volume IOR does is overwrite a glass materials IOR. For other matierals, it is ignored:
https://wiki.luxcorerender.org/LuxCoreRender_Volumes

[CodeFHD]

There is no distinct priority order yet, although based on the recent user survey (where new core features came out at the top), and personal interests of mine I would say rather high.

Full spectral conversion is a significant amount of work. it requires changes to pretty much all rendering, material, texture, ... functions. Some analysis of performance (degradation) compared to classic RGB-rendering, and then a decision how to implement it (i.e. how to have both without creating too much duplicate code).

Wavelengths dependent IORs as such already exist for the current implementation, though the formula is very simple (2-term Cauchy) and should be replaced with something more complex (e.g. 6-term Sellmeier). This part alone is not difficult to implement. Some thought should be put how to expose this in BlendLuxCore in particular, so that it won't be too complex for the average user.

[thomasmoore5]

What about wavelength dependent IORs for volumetrics?

[CodeFHD]

That is a whole new feature. Unless you mean simply adding a dispersion setting to volume nodes that gets used by glass nodes, the same as with IOR as outlined in my first response.

Assuming you are thinking further: Rather than considering wavelength-dependence, volume-IOR effects need to be implemented. In other words, gradient index optics: https://en.wikipedia.org/wiki/Gradient-index_optics

Recently, someone on the Discord server started working on this but I cannot speak for the progress of that.
For my personal ToDo-list, I would put regular spectral rendering first before considering gradient-index.

[thomasmoore5]

Thanks for getting back to me on this. So, just to confirm, Gradient Index Optics would be required to accurately simulate rainbows?

[CodeFHD]

No, gradient index is not something required for all spectral effects.
If you mean the actual rainbow that appears during bad weather, that is simple dispersion and reflection in water drops: https://de.wikipedia.org/wiki/Regenbogen#/media/Datei:Rainbow1.svg

But it would be the same for a "rainbow" created with a prism.

Gradient index IOR I would see rather in special applications, really going into the scientific domain, i.e. accurately modelling some effect that can't be achieved otherwise.