Hi Aurelien,

On 3 June 2013 11:10, Aurelien Albert <[email protected]> wrote:
> I had his example to implement :
>
> - 100 instances to draw, with 100 different instances data stored into a 
> VertexAttrib using VertexAttribDivisor.
> - "normal" draw => ok
> - "odd/even" draw => here I need to draw each odd instance in red, each even 
> instance in green
>   => I'm still using the same data array, but with a glVertexAttribDivisor of 
> 2, and I use 2 drawables :
>         first with "red" shader and 0 offset
>         second with "green" shader and 1 instance offset
>
> I know this is a very special case, and I can live by handling this by myself 
> using draw callbacks

Playing such reuse games sound like a recipe for running out of array
bounds by accident for one of the usages.  Avoiding such issue really
does require a level of mental dexterity to understand and keeping
understanding throughout the lifecycle of the application and by all
those who might review/modifying it.  Sometimes being "clever" like
this isn't clever overall.


>> It probably is time for a general discussion on osg-users about this topic.
>
>
> I think so. For the moment, I see instancing as an "openGL trick" and "user 
> should exactly know the consequences to use it".

Instancing is still pretty new, but we've been using vertex shaders
for a long time now and still the OSG has rather cludged over it - the
bounding volume issue is one big cludge that we really need to sort
out.

Robert.
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Reply via email to