feat: change the number of depth samples in volume rendering to a controllable UI parameter

[seankmartin]

Goal

Change the number of depth samples from a fixed 64 to a user-controlled variable. Named "samples per ray".

Primary changes

1. Create a UI controllable int, default 64, min 1, max 8096 that defines the samples per ray.
2. On update of samples per ray we trigger a redraw and a recompute of the chunk priority, in case this change switches the desired data resolution level. The data resolution level is currently based on the depth of the view frustum divided by the number of samples.
3. Pass the samples per ray into both the chunk worker and the shader to accommodate this.
4. Control the samples per ray in Python.

[google-cla[bot]]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[jbms]

This definitely seems like a good option, but what do you think about using the existing "Resolution" control (that currently does nothing) but modifying it to display the different resolutions in units of "voxel size / view frustum depth"? That way this "Resolution" control would behave similarly to the other Resolution controls and would no longer be a "display only" control.

[seankmartin]

I really like that idea. My hope was that the implementation here would be a stepping stone to something more robust, and that is a good suggestion.

We'd essentially be changing up the logic from the resolution being determined by a user setting of the near and far planes and samples per ray, to the number of samples per ray being determined by a user setting of the near and far planes and resolution level. Am I understanding that right?

Edit: Maybe we want to still leave the number of samples modifiable so that one can over/under sample the actual data resolution if they want to.

[jbms]

To be clear, I was imagining that the actual user setting would still be number of samples, as you have it now, it would just be controlled using the (modified) "Resolution (3d)" control rather than a separate control. In order for that to make sense, the "Resolution (3d)" control would need to display the number of samples in log scale, and plot each resolution according to the number of samples that would be optimal for it, so that it displays in the correct position relative to the vertical bar that indicates the current "number of samples" selection.

[seankmartin]

Sure, I understand better now - I still like the idea!

I'll get to making that change in the coming days

[jbms]

Please resolve merge conflicts and then should be ready.

[seankmartin]

@jbms merge conflicts fixed!