search

RenderKit / openvkl

Intel(R) Open Volume Kernel Library
Apache License 2.0
200 stars 26 forks source link

Upper bound on structuredRegular volume dimensions? #20

Open nastronaut opened 1 year ago

nastronaut commented 1 year ago

Hello,

I am attempting to render an isosurface in a 5120^3 structuredRegular volume on OSPRay. The volume has a scalar field where the data are provided by floats (doing the math, that's half a terabyte of data). I use mmap since I do not have half a terabyte of CPU memory. When I render I see artifacts where it appears blocks are missing or are rendered differently at the high end of the X and Y dimensions (the more rapidly changing dimensions in my data object). Is there an upper bound on how large the dimensions can be in a structuredRegular volume? I could not find such a bound in the OpenVKL or OSPRay docs.

The attached image is from an early time step in the simulation. The isosurface should have a relatively uniform texture across the domain in X and Y. If I down sample the data to 1024^3 and render that I do get a relatively uniform isosurface.

Thanks,

Patrick 5120-isosurface-example

gregjohnson commented 1 year ago

Hi Patrick, Thanks for reporting the issue. There should be no upper bound on volume dimensions, so this does appear to be a bug. Can you tell me which specific Open VKL and OSPRay versions you're using? We can then try to reproduce this ourselves and look for a fix. Thanks!, Greg

nastronaut commented 1 year ago

I'm using relatively recent versions of the libraries:

Embree version: 3.13.5

OpenVKL version: 1.3.1

OSPRay version: 2.10.0

Thanks,

Patrick