yade-dev team mailing list archive

[Bruno Chareyre <bruno.chareyre@xxxxxxxxxxx>]

sega a écrit :
>    
>   
>> The problem is somebody else will want to draw velocity, stiffness, 
>> volume of Voronoi cells, local deformation, or anything else I don't 
>> even suspect. So, saving all possible modelling results in periodic 
>> snapshots is impossible.
>> Off-screen rendering sounds more realistic to me, only problem is you 
>> will have to set camera position before the simulation.
>>     
>
> I think that really when planning the simulation to select parameters that 
> will be needed in the future for postprocessing. It is clear that in the 
> future if I want to work only with the Voronoi cells, I need only to save 
> these values and nothing more. And in the future I always be able to 
> visualize this Voronoi cells and see it from a new angle (clipping plane, 
> color scheme, etc...). (By the way, not necessarily to save velocity. It can 
> be obtained from two consecutive positions snapshots.). 
> In doing so, off-screen rendering sounds fine and in this case: you can 
> configure a new look and also start to render the scene in background.
> If I will have to set camera only before the simulation, I will have to set 
> _all_ cameras (all foreshortening, clipping planes, color shemes).To me, this 
> is not so easy... In my opinion, it is better to separate the simulation and 
> visualization. How do you think?
>
>   
I agree that separating simulation and visualisation would be better 
(besides it needs more file management before actually getting the 
animation), but the problem is (I think) it needs more coding.

Offscreen rendering needs no new method. Anything drawable in Yade can 
be seen on snapshots, directly converted from graphic card output to 
image file (for setting camera and other QGL parameters, I guess we can 
use files like the preferences.xml we have in ~/.yade).

If we go for separating simulation/rendering, we need 2 more methods for 
each type of drawable : writing GL object to file + loading GL object 
from file, and we also need the code and interface for tuning which 
types of objects we want to save in snapshots.
Sounds more difficult, but of course it has the advantage of being a lot 
more versatile.

Actually, both approaches can co-exist in Yade. So, if you feel like 
coding the saving/loading of velocity field in text snapshots, it will 
be usefull. And if later we have offscreen rendering, it will be a quick 
solution for all those objects that are not included in text snapshots.

Bruno

-- 
 
_______________
Chareyre Bruno
Maitre de conference

Institut National Polytechnique de Grenoble
Laboratoire 3S (Soils Solids Structures) - bureau E145
BP 53 - 38041, Grenoble cedex 9 - France
Tél : 33 4 56 52 86 21
Fax : 33 4 76 82 70 43
________________

_______________________________________________
yade-dev mailing list
yade-dev@xxxxxxxxxxxxxxxx
https://lists.berlios.de/mailman/listinfo/yade-dev