Approximation of the signed distance fields

[Jeremie G.]

Good afternoon, 

I am using the scikit-fmm library to calculate signed distance fields to identify the parts of the domain that are a certain distance away from the 0 value level set (in 3D).

In my case: 

1. accuracy is not very important, 

2. using a narrow band is perfectly OK.
I am dealing with potentially large volumes, and I am wondering if is there is a trick to make the computations faster, besides using the narrow band feature of the distance function? 

Thanks and regards, 
Jeremie

[Jason Furtney]

Hi Jeremie,

Thanks for the message and for your interest in the software. 

I cannot think of an immediate solution to what you are asking for. scikit-fmm allocates an array large enough for the entire domain and the narrow band feature just stops the calculation at a certain distance. You may be able to get a small speed up by using the first-order point update (see the optional arguments for the distance function). Other than that, if you need to go faster, you may need to look into breaking the domain up into parts and solving them in parallel and somehow stitching the domains back together into the global solution. scikit-fmm does not have any support for parallel processing, so I am not sure a good way to do that.

Sorry I can't be of more help and good luck with your work,

Jason

--
You received this message because you are subscribed to the Google Groups "scikit-fmm" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scikit-fmm+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/scikit-fmm/2f35bfae-9951-404f-85a2-b6e0889f3e71n%40googlegroups.com.

[Jeremie G.]

Hi Jason, 

Thank you for your quick response. Thanks also for the suggestion about breaking the domains into parts - it could indeed be an avenue to look at. 

Best regards, 

Jeremie

To view this discussion on the web visit https://groups.google.com/d/msgid/scikit-fmm/CAOU2ZSU2NntgqHfyb7KTAWSLLwCg9QOJOGZGkAvoLjyh%3D7Bktw%40mail.gmail.com.