Hi all! > My personal bias is that the added cost to obtain real QBS hardware with built-in stereo support > is well worth the money in terms of time saved on bugs and hassles such as this.
Yes! Real quad-buffered stereo is IMHO crucial for molecular modelling work ;-) > Under Linux, the nVidia Quadro4 750XGL does a fine job, has a three-pin stereo sync plug, > and works with the Nuvision glasses and emitters. While I haven't tested the higher-end FX series, > I have every reason to believe that they would also work just fine given nVidia's unified driver > architecture. I doesn't work for me :-( I ordered a Quadro4 750XGL, and the vendor sent me a Quadro FX1000 by mistake. I didn't notice at first, and just installed the card. Works fine and very fast with the 1.0-4363 version of the Linux driver - but neither PyMOL or VMD detects any hardware stereo.... I have to get it replaced by a 750XGL or 900XGL. We have some older stereographics glasses, which I expect will work with the Nvidia cards. But we have also bought a set of Eye3D premium glasses. They come with a VGA pass-through cable, which can catch the sync signal from any VGA output. They also support different stereo modes, including interlacing, page-flipping, sync-doubling, and line-blanking. You get two sets of glasses (one wired, one IR) for €179 or $199. This works nice with PyMOL and the Xig X-server on an ATI Radeon 9000 card, and the "Scanline interleave" (interlaced) stereo mode in VMD works fine with these glasses and any graphics adapter. Maybe "scanline interleave" could be a stereo option in PyMOL - it is definitely inferior to quad-buffered stereo, but it works with _any_ graphics card and doesn't require a special driver. (I would still go for QBS, though :-) Cheers Esben