[Soc-2009-dev] Raytrace api

[André Pinto]

Hi,

I probably wont be able to be online this sunday meeting.
Anyway my SoC plans for this first week are:

===comming week===
    *find the bug that makes some black dots.. before I only changed the API
but it seems some bugs appeared.. I would like to have those sorted out
before continuing. I already spend some time trying. but its kinda hard to
debug it.
   *plugging the bvh-tree available on blenlib as a tree structure
   *if time allows add also a raytrace structure to instance-support

===questions===
 none

===schedule===
 I am still a under school works, but hopefull the advanced I gave before
will allow me not to get behind schedule.


2009/5/11 Brecht Van Lommel <brecht at blender.org>

> Hi André,
>
> On Mon, 2009-05-11 at 16:12 +0100, André Pinto wrote:
> > Hi all,
> >
> > ==Questions==
> > this week I have been coding a more generic raytrace api and I have
> > some questions:
> >
> > * In the new files (rayobject.h, rayobject*.c), I copied the GPL
> > license block from other files and changed copyright to "2009 Blender
> > Foundation". I also set the properties svn:keywords and svn:eol-style.
> > Is this all that needs to be done for created files?
> > And Is there any sample GPL licence block? i noticed those sligtly
> > change between files.
>
> I don't think we have a policy on svn properties and the license block
> details, just do the same as another file in the directory.
>
> > * The exported functions from render, should all started with RE_ ?
> > That means the functions I added should be:
> > RE_rayobject_create/raycast/done/add..
>
> Yes, use the RE_ prefix.
>
> > * Is it safe to assume that all MEM_callocN are aligned to 4bytes?
>
> Yes (though obviously make a comment in the raytracing code that you
> rely on this).
>
> > ==API details==
> > Details about the ray trace design implementation
> > (render/intern/include/rayobject.h):
> >
> > I denoted RayObject as everything where its possible to raycast
> > including a single face or a complex data structure.
> > Thus allowing to create compound structures (eg.: a bvh of octrees
> > (like 1 octree per object)).
> >
> > In order to allow an eficient type-detectation mechanism, all
> > RayObjects are aligned to 4bytes, leaving 2bits of adress to determine
> > the type, so far 2 types exist:
> >
> > RayFace - struct with coords, object and face.
> > RayObject - struct to an abstract raytrace aceleration, with callbacks
> > to generic stuff like: add faces, free struct and raycast
> >
> > All callbacks like get coordinates/transform_coords, check_hit, where
> > removed for now.. as the functionality of those will be coded as
> > needed in a better way.
> >
> > For now only API and Isect structure were modified.
> > Daniel Salazar(ZanQdo) reported an 9% improve on his quad core2  :)
> > At the moment only shadow rays work/were tested.
>
> OK, design sounds good, and 9% is a nice start :).
>
> Brecht.
>
>
> _______________________________________________
> Soc-2009-dev mailing list
> Soc-2009-dev at blender.org
> http://lists.blender.org/mailman/listinfo/soc-2009-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/soc-2009-dev/attachments/20090522/43ca5c7d/attachment.htm