[Bf-committers] GL_SELECT

[Jason Wilkins]

My opinion about picking is that it is OK that picking is coupled to the
drawing code because we are picking what is being drawn.  That is generic
programming.  It seems like otherwise you'd have twice as much code with
two branches that need to be kept in sync.

I'd rather explore features that could make both drawing and picking
better, while still reusing the code for specifying what is drawn/picked.

Maybe the problem isn't that picking/drawing code is coupled, but that both
are coupled to other code?  Maybe it is that geometry specification needs
to be decoupled from state that has only to do with either picking or
rasterization.



On Sun, Jul 28, 2013 at 5:40 AM, Antony Riakiotakis <kalast at gmail.com>wrote:

> On 28 July 2013 10:01, Jason Wilkins <jason.a.wilkins at gmail.com> wrote:
>
> > Psy-Fi did an implementation of Occlusion query picking.  I'd like to
> know
> > why that wasn't followed up on.
> >
> >
> Nothing really, the code is rotting peacefully here
> https://github.com/Psy-Fi/psy-fi_blender/tree/occlusion-selection and it's
> been a while since I rebased the branch to trunk.
>
> As far as I remember the consensus was that ray picking would be
> advantageous because it would allow tools to be decoupled from drawing
> code. Also I'm not sure if OpenGL ES supports occlusion queries, if not it
> might be problematic later on.
>
> There is really one issues with ray picking in my mind, which is that for
> big scenes it would require tremendous processing for pixel exactness,
> unless a hierarchical scheme is used. If some bvh scheme is used then
> possibly these could take quite less with the disadvantage of increased
> memory footprint. However such queries could even be threaded, making this
> more effective.
>
> In any case, the occlusion code does work but requires two passes for a
> first hit collision to do depth peeling. Since display is usually fast this
> is OK for most cases. I haven't tested it at all with the offscreen buffer
> selection that blender uses now but I hope it should work.
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>