Thread: RE: [Algorithms] View-Depend vs. View-Independent LOD
Brought to you by:
vexxed72
Menu
▾
▴
gdalgorithms-list
|
From: Tom F. <to...@mu...> - 2000-08-09 19:11:28
|
VIPM is where your collapse order is fixed - the only thing the observer's position does is choose how many tris to expand/collapse to. VDPM varies the order of collapse depending on where the view is, so bits further away from the camera get collapsed before bits near the camera - stuff like that. That's it. Tom Forsyth - Muckyfoot bloke. Whizzing and pasting and pooting through the day. > -----Original Message----- > From: Pai-Hung Chen [mailto:pa...@ac...] > Sent: 09 August 2000 19:57 > To: gda...@li... > Subject: [Algorithms] View-Depend vs. View-Independent LOD > > > Hi, > > I've heard that there are two kinds of LOD: view-dependent and > view-independent. Could somebody briefly explain what the essential > difference between the two? > > Thanks in advance, > > Pai-Hung Chen |
|
From: Daniel R. <dan...@ho...> - 2000-08-09 20:13:29
|
(think about progressive mesh / mmr lod)
if you would crate a big landscape an choose its triangle
count via lod and change it completely it's View Independant
if you use a algorithm to calculate which landscpae
point should be colapsed you will do that mostly
via: far away => many collapses
near => few collapes
and so this far/near is View Dependant
hope this helps, re...@ma...
>From: Tom Forsyth <to...@mu...>
>Reply-To: gda...@li...
>To: gda...@li...
>Subject: RE: [Algorithms] View-Depend vs. View-Independent LOD
>Date: Wed, 9 Aug 2000 20:05:51 +0100
>
>VIPM is where your collapse order is fixed - the only thing the observer's
>position does is choose how many tris to expand/collapse to.
>
>VDPM varies the order of collapse depending on where the view is, so bits
>further away from the camera get collapsed before bits near the camera -
>stuff like that.
>
>That's it.
>
>Tom Forsyth - Muckyfoot bloke.
>Whizzing and pasting and pooting through the day.
>
> > -----Original Message-----
> > From: Pai-Hung Chen [mailto:pa...@ac...]
> > Sent: 09 August 2000 19:57
> > To: gda...@li...
> > Subject: [Algorithms] View-Depend vs. View-Independent LOD
> >
> >
> > Hi,
> >
> > I've heard that there are two kinds of LOD: view-dependent and
> > view-independent. Could somebody briefly explain what the essential
> > difference between the two?
> >
> > Thanks in advance,
> >
> > Pai-Hung Chen
>
>_______________________________________________
>GDAlgorithms-list mailing list
>GDA...@li...
>http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list
________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com
|
|
From: Daniel R. <dan...@ho...> - 2000-08-09 20:41:15
|
>From: "Pai-Hung Chen" <pa...@ac...> >Reply-To: gda...@li... >To: <gda...@li...> >Subject: Re: [Algorithms] View-Depend vs. View-Independent LOD >Date: Wed, 9 Aug 2000 13:26:06 -0700 > >Hi, > > > VIPM is where your collapse order is fixed - the only thing the >observer's > > position does is choose how many tris to expand/collapse to. > > VDPM varies the order of collapse depending on where the view is, so >bits > > further away from the camera get collapsed before bits near the camera - > > stuff like that. > >Does this mean VIPM collapses polys *evenly* across the whole mesh so that >polys get the same degree of collapse regardless of their distance to the >viewpoint, while in VDPM farer polys are collapsed more heavily than nearer >polys? exactly, this is one of the typical behaviours ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com |
|
From: Pai-Hung C. <pa...@ac...> - 2000-08-09 21:07:20
|
In a HW T&L setting (GeForce), are VIPM/VDPM really worth the extra management cost omposed on CPU? Or a simple Quadtree/Octree paradigm will suffice, in which I may well just send the un-LODed polys to the HW T&L pipeline that is so fast at handling huge amount of polys that it still beats the LOD optimization? Thank you, Pai-Hung Chen |
|
From: Tom F. <to...@mu...> - 2000-08-10 08:49:53
|
If you look at the amount of CPU work required for VIPM, it's tiny. And you don't touch any vertex data, just indices. It certainly is useful for scenes with lots of people - my characters go up to around 120ktris each. Draw twenty of them at full rez, and you'll kill even the best chip. VIPM is absoloutely required for this sort of stuff, and it's very fast. Tom Forsyth - Muckyfoot bloke. Whizzing and pasting and pooting through the day. > -----Original Message----- > From: Pai-Hung Chen [mailto:pa...@ac...] > Sent: 09 August 2000 22:03 > To: gda...@li... > Subject: Re: [Algorithms] View-Depend vs. View-Independent LOD > > > In a HW T&L setting (GeForce), are VIPM/VDPM really worth the extra > management cost omposed on CPU? Or a simple Quadtree/Octree > paradigm will > suffice, in which I may well just send the un-LODed polys to > the HW T&L > pipeline that is so fast at handling huge amount of polys > that it still > beats the LOD optimization? > > Thank you, > > Pai-Hung Chen > > > > _______________________________________________ > GDAlgorithms-list mailing list > GDA...@li... > http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list > |
|
From: Daniel R. <dan...@ho...> - 2000-08-10 17:14:23
|
yes, you got it. 100 buble gum donut points for you. you asked the question of the year. many,many discussed this topic in the last weeks/months and the results vary extremly. depending detailed on what you're goind to do. ... this will not help, but clarify, re...@ma... >From: "Pai-Hung Chen" <pa...@ac...> >Reply-To: gda...@li... >To: <gda...@li...> >Subject: Re: [Algorithms] View-Depend vs. View-Independent LOD >Date: Wed, 9 Aug 2000 14:03:04 -0700 > >In a HW T&L setting (GeForce), are VIPM/VDPM really worth the extra >management cost omposed on CPU? Or a simple Quadtree/Octree paradigm will >suffice, in which I may well just send the un-LODed polys to the HW T&L >pipeline that is so fast at handling huge amount of polys that it still >beats the LOD optimization? > >Thank you, > >Pai-Hung Chen > > > >_______________________________________________ >GDAlgorithms-list mailing list >GDA...@li... >http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com |
|
From: Daniel R. <dan...@ho...> - 2000-08-10 17:14:23
|
yes, you got it. 100 buble gum donut points for you. you asked the question of the year. many,many discussed this topic in the last weeks/months and the results vary extremly. depending detailed on what you're goind to do. ... this will not help, but clarify, re...@ma... >From: "Pai-Hung Chen" <pa...@ac...> >Reply-To: gda...@li... >To: <gda...@li...> >Subject: Re: [Algorithms] View-Depend vs. View-Independent LOD >Date: Wed, 9 Aug 2000 14:03:04 -0700 > >In a HW T&L setting (GeForce), are VIPM/VDPM really worth the extra >management cost omposed on CPU? Or a simple Quadtree/Octree paradigm will >suffice, in which I may well just send the un-LODed polys to the HW T&L >pipeline that is so fast at handling huge amount of polys that it still >beats the LOD optimization? > >Thank you, > >Pai-Hung Chen > > > >_______________________________________________ >GDAlgorithms-list mailing list >GDA...@li... >http://lists.sourceforge.net/mailman/listinfo/gdalgorithms-list ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com |
|
From: Corrinne Y. <cor...@ho...> - 2000-08-09 19:24:16
|
----- Original Message ----- From: "Tom Forsyth" <to...@mu...> To: <gda...@li...> Sent: Wednesday, August 09, 2000 2:05 PM Subject: RE: [Algorithms] View-Depend vs. View-Independent LOD -- Tom F. described both PM's very well below. I am just adding a little. > VIPM is where your collapse order is fixed - the only thing the observer's > position does is choose how many tris to expand/collapse to. -- VIPM generally has a higher degree of precomputability. > VDPM varies the order of collapse depending on where the view is, so bits > further away from the camera get collapsed before bits near the camera - > stuff like that. -- VDPM theoretically has fewer screenspace collapse artifact. It generally has higher realtime computation costs. -- For both PM's evaluate your PM marching costs versus your rasterization costs. If marching costs you more than just rendering them, render them out. :) |
|
From: Pai-Hung C. <pa...@ac...> - 2000-08-09 20:31:09
|
Hi, > VIPM is where your collapse order is fixed - the only thing the observer's > position does is choose how many tris to expand/collapse to. > VDPM varies the order of collapse depending on where the view is, so bits > further away from the camera get collapsed before bits near the camera - > stuff like that. Does this mean VIPM collapses polys *evenly* across the whole mesh so that polys get the same degree of collapse regardless of their distance to the viewpoint, while in VDPM farer polys are collapsed more heavily than nearer polys? Thank you, Pai-Hung Chen |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X