You may want to take a careful look at the pbrt-v3 license before you get too involved. I'm not sure it's GPL2+ compatible. It may be but you should get clarification. Also, pbrt-v3 is not packaged in Debian and likely not packaged for any other platform either so that needs to be considered as well. I am not allowing any more CMake dependency library build code in KiCad so that is not an option.
On 10/28/2015 7:06 AM, Nick Østergaard wrote: > 2015-10-28 10:29 GMT+01:00 Mário Luzeiro <[email protected]>: >> >>>> - I have in mind to try to make it as fast / interactive as possible, so >>>> it will need some specific implementation.. some hacks .. that will target >>>> this specific propose. >> >>> I understand. Speed is definitely an issue with big boards. But I am >>> no export in ray tracing characteristics, but what is the scalability >>> of your implementation in reagard to board complexity (lots of holes >>> and lots of other objects) and seperately about the screensize? >> >> Actually, as I experienced, the size of the boards will not make a big >> impact in the loading/rendering. From the 'size' it means 'how much >> items/objects it will load/render'. >> ( Btw, anyone have a real real big board project for kicad? something like >> the complexity of a PC motherboard? ) >> >> Regarding the scalability, raytracing algos are very good to handle >> scalability. The limit of objects, are the u32 size and your RAM. >> I am using some 'state-of-art' algos for ray traversal (frustum packets >> over a bounding volume hierarchy). (BVH is based on the PBRT 3 book [1], and >> traversal is based on [2]) >> I means that it can adapt very well to any number of objects or >> configurations without suffering much penalty. >> >> Loading the board is also very fast, creating the BVH will take some 200ms >> on complex boards. >> The thing that will take more time is still some tessellation need for the >> polygon filled planes, but I have in mind some optimizations. >> It means at the moment that loading a board like 'video' (from kicad's demo >> boards) or 'hackRF' takes about 3 seconds + render. >> I plan to try to cache somehow the tesselation, so this time can be much >> more reduced. >> >> Resolution is at moment set to the windows size, but for export rendering to >> image file, we can add a dialog to choose different resolutions. >> Of-course, more resolution means it will be more slow. >> >> At this moment on my machine, rendering for quality takes about >> 1..2.3seconds (depend of the view) with all eye-candy features set. If some >> of that features are not used (eg: post shaders effects, refractions or >> shadows), the render can take less than 1s >> >> Cheers, >> Mario >> >> >> [1] https://github.com/mmp/pbrt-v3/ >> >> [2] Wald, S. Boulos, and P. Shirley. Ray Tracing Deformable Scenes using >> Dynamic Bounding Volume Hierarchies. ACM TOG , 26(1), 2007 >> http://www.cs.cmu.edu/afs/cs/academic/class/15869-f11/www/readings/wald07_packetbvh.pdf >> implementation of the algo based in the description from: >> https://cseweb.ucsd.edu/~ravir/whitted.pdf >> > > Ok, I see. Thank you for the information. > > _______________________________________________ > Mailing list: https://launchpad.net/~kicad-developers > Post to : [email protected] > Unsubscribe : https://launchpad.net/~kicad-developers > More help : https://help.launchpad.net/ListHelp > _______________________________________________ Mailing list: https://launchpad.net/~kicad-developers Post to : [email protected] Unsubscribe : https://launchpad.net/~kicad-developers More help : https://help.launchpad.net/ListHelp
