That's all I do, and yes, it's not great. You could build a table of object
space error (as determined by QEM or whatever your edge-collapse function
is) vs number of tris. Then from the screen-space BB and a screen-space
error value, and the object-space bounding sphere radius, you can find the
object-space error required. Look that up in the table, and you know how
many tris you should draw. The table doesn't have to be very precise - my
feeling is that 16 entries is going to be enough - maybe a few more - and
then linearly interpolate between those values.
I've not tried this, but I think I will - the current method (num_tris =
tweak_factor * BB_screen_area) is not wonderful.
Tom Forsyth - Muckyfoot bloke.
Whizzing and pasting and pooting through the day.
> -----Original Message-----
> From: Timur Davidenko [mailto:ti...@3d...]
> Sent: 19 July 2000 19:50
> To: gda...@li...
> Subject: [Algorithms] ProgMesh LOD function.
>
>
> I`m trying to come up with good lod function for progressive
> mesh reduction,
> i have all progressive mesh collapses precalculated so can`t
> use per vertex
> screen error function,
> i need something more general for complete mesh, function
> which says how
> many percents of mesh should be cut down
> currently i just taking screen space area of mesh bounding
> box, dividing by
> some kludge const, but it works badly for different mesh scales.
> any ideas?
>
> regards.
>
> _______________________
> Timur Davidenko.
> 3Dion Inc. (www.3dion.com)
> e-mail: ti...@3d...
>
>
>
> _______________________________________________
> GDAlgorithms-list mailing list
> GDA...@li...
> http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list
>
|
