Re: [Flightgear-devel] Texture compression experiments in plib

2006-02-23 Thread Tiago Gusmão
Hi. I'm still waiting for response from the plib-devel list! But until then maybe we could at least include an improved patch in the pre-compiled releases? Cheers Chris Metzler wrote: On Tue, 10 Jan 2006 16:17:32 + Tiago Gusmão wrote: I will post in the plib in the plib devel list jus

Re: [Flightgear-devel] Texture compression experiments in plib

2006-02-22 Thread Chris Metzler
On Tue, 10 Jan 2006 16:17:32 + Tiago Gusmão wrote: > > I will post in the plib in the plib devel list just to make sure we can > go ahead with this, but before i would like to know what exactly do we > want and have someone double-check the specs and issues, i'm not an > OpenGL guru or somet

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-10 Thread Melchior FRANZ
On Tue, Jan 10, 2006 at 04:17:32PM +, Tiago Gusmão wrote: > I see that .sgi has a few dummy fields we could use (altough that would > break the spec, they should be NULL), That's not an option. Even using the comment line for that isn't one, although it wouldn't violate the spec. m. --

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-10 Thread Tiago Gusmão
I will post in the plib in the plib devel list just to make sure we can go ahead with this, but before i would like to know what exactly do we want and have someone double-check the specs and issues, i'm not an OpenGL guru or something like that ;) I just found this: http://www.gamasutra.com/f

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-10 Thread Christian Mayer
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Tiago Gusmão schrieb: > The other problem is that this is plib, i'm not sure this is simply > going to be accepted into the tree after being coded properly. > And to make it an option, i think we also need to modify it to get the > option passed to pli

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Tiago Gusmão
Technically, the extensions can be queried in run-time (and there are a few more that can be included easily). I don't see any reason why would anyone with compatible HW want to disable this, but who knows? Perhaps some purist dislikes that s3tc is lossy ;) The other problem is that this is pl

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Christian Mayer
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Paul Surgeon schrieb: > Well there is still lots of merit in using texture compression. It also > allows > us to stuff a lot more textures (or higher resolution ones) into VRAM even on > higher end cards. Definitely. I'm for texture compression (sw

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Jon Stockill
Curtis L. Olson wrote: I'd be interested in working with the Tiger data as well. Did you create any code to parse/chop the data? If I had a chance to work through the process I might be able to track down some of these problems/issues. I got myself a copy of the 2004 Tiger data in shapefil

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Martin Spott
Hello Curt, "Curtis L. Olson" wrote: > I'd be interested in working with the Tiger data as well. Did you > create any code to parse/chop the data? The OGR library and tools happily read TIGER data without further assistance: http://www.remotesensing.org/gdal/ogr/ We currently use it to imp

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Paul Surgeon
On Monday 09 January 2006 22:11, Christian Mayer wrote: > Tiago Gusmão schrieb: > > Altough s3tc is quite widespread, since we are using 2^n sided textures, > > it should be trivial to add support to down-size sides by 2^n if the > > user wishes so, at the expense of longer loading time. We could a

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Christian Mayer
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Tiago Gusmão schrieb: > Altough s3tc is quite widespread, since we are using 2^n sided textures, > it should be trivial to add support to down-size sides by 2^n if the > user wishes so, at the expense of longer loading time. We could also > provide a w

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Tiago Gusmão
Altough s3tc is quite widespread, since we are using 2^n sided textures, it should be trivial to add support to down-size sides by 2^n if the user wishes so, at the expense of longer loading time. We could also provide a way to convert the actual files. Models could also be LODed during loading

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Jon Stockill
Ralf Gerlich wrote: Hi, Jon Stockill schrieb: [SNIP] As an idea of the processing involved, as I'm sure you're aware by default the terragear tile builder has a cpu limit of 120s set - you're not going to get anything at all built with those limits using this amount of data - it takes anythi

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Curtis L. Olson
Jon Stockill wrote: It doesn't seem to be displaying it that's the problem - it's getting terragear to build it. Once you reach a certain polygon density things start to break down, there's obviously a lot more polygons, but the road polys just don't seem to show up, and there's a big increase

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Ralf Gerlich
Hi, Jon Stockill schrieb: [SNIP] As an idea of the processing involved, as I'm sure you're aware by default the terragear tile builder has a cpu limit of 120s set - you're not going to get anything at all built with those limits using this amount of data - it takes anything between 45 minutes

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Christian Mayer
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 AJ MacLeod schrieb: >>I was hoping that it might enable me to run Jon's new KSFO scenery with >>Tiger data, but no improvement there at all. Either Jon has to reduce the >>ploy count markedly, or I have to upgrade my video card! > > Same here, unfortu

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Jon Stockill
Martin Spott wrote: Hello Jon, Jon Stockill wrote: http://flightgear.stockill.org.uk/testing/BaseSceneryWithTiger.tgz Thanks for posting the link. Can you share your experience with us how to size the hardware in order to run this scenery ? I know TIGER is _very_ detailed and, analogous to

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Martin Spott
Hello Jon, Jon Stockill wrote: > http://flightgear.stockill.org.uk/testing/BaseSceneryWithTiger.tgz Thanks for posting the link. Can you share your experience with us how to size the hardware in order to run this scenery ? I know TIGER is _very_ detailed and, analogous to Ralf's scenery, certain

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread Jon Stockill
AJ MacLeod wrote: Using this patch and plib-1.8.4 on a 64Mb GeForce 420MX there is a massive improvement; not immediately noticeable, but... Whereas before I couldn't use the Hurricane in most parts of the FG world, and got single-figure fps rates at KSFO with any a/c and Nimitz enabled, I ca

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-09 Thread AJ MacLeod
On Sunday 08 January 2006 10:40, Vivian Meazza wrote: > I've applied this patch to my copy of plib-cvs. With a nVidia FX5200 card > there is no improvement in frame rate around KSFO, or near Nimitz (but I > don't get a drop there anyway). I think there is an improvement around SF > and the bridges.

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-08 Thread Jon Stockill
Martin Spott wrote: Ah, I knew once someone would try TIGER data with FlightGear :-) Is there any public notion of this Scenery ? http://flightgear.stockill.org.uk/testing/BaseSceneryWithTiger.tgz That's using tiger data for major roads, railways, and waterways - the water flowing down vall

Re: [Flightgear-devel] Texture compression experiments in plib

2006-01-08 Thread Martin Spott
"Vivian Meazza" wrote: > I was hoping that it might enable me to run Jon's new KSFO scenery with > Tiger data, [...] Ah, I knew once someone would try TIGER data with FlightGear :-) Is there any public notion of this Scenery ? Martin. -- Unix _IS_ user friendly - it's just selective about who

RE: [Flightgear-devel] Texture compression experiments in plib

2006-01-08 Thread Vivian Meazza
Tiago Gusmão > I'm having a HW problem and i can't enable AGP in linux, which makes the > hunter, b1900d and the Nimitz lower FPSs dramatically, which i believe > is caused by the NVRAM filling up with textures thus forcing texture > transfers to the card (GF4 MX 440 64MB) > > so after Surge ment

[Flightgear-devel] Texture compression experiments in plib

2006-01-07 Thread Tiago Gusmão
Hello I'm having a HW problem and i can't enable AGP in linux, which makes the hunter, b1900d and the Nimitz lower FPSs dramatically, which i believe is caused by the NVRAM filling up with textures thus forcing texture transfers to the card (GF4 MX 440 64MB) so after Surge mentioned texture comp