| Author | Message | ||
     Alexander Oest Rating: N/A |
A couple of quotes: "All the cards feature VGA-out for analog displays, as well as DVI-out, which can be used to optionally hook up either a second analog monitor or a digital flat screen. A 3-pin connector is included to connect shutter glasses so that three-dimensional models can be visualized when the relevant application supports them." "3Dlabs currently produces three models based on the P10: the starter model, Wildcat VP760, for $449, the mid-range variant, the VP870, for $599 and the high-end product, the VP970, for just under $1199." http://www.tomshardware.com/graphic/02q3/020807/wildcat_vp-01.html This seems like a decent card at a reasonnable price (the low-end model is around the same price as the expensive GeForce4s) Anybody tried it? Michal, I saw you were active at the 3DLabs forum asking about the opportunities of separate left/right output to dual VGA. Did you find out if it's possible on these cards? Alex | ||
| M.H. Rating: N/A |
Hi Alex I am working on a profesional code for stereo video playbak - the DepthQ project for Lightpeed Design. As a part of this project I am testing Wildcat VP 870 and Quadro 4 900 XL extenisvely just now ... Speed results: In standard OpenGl tests: Wildcat VP 870 is about 1.2 - 1.5 x faster than the Quadro 4 900 XL. In DirectX 8.1 test: Quadro 4 900 XL is about 1.5-2x faster than Wildcat VP The left-right output separation implemented for OpenGl stereo in Quadro is not implemented in Wildcat VP . I am doing cosultation with 3D labs people to solve this ... In additon WildcatVP has problems with pixel fillrate - it is to slow for work with our stereo-video playback code now. I am doing consulations with 3D labs drivers developers to solve this as well ... Results: For normal havy polygon based Opengl applications is Wildcat VP is probably better then Quadro ... For games and a lot of texture Quadro wins now ... OpenGl stereo works correctly on both models. | ||
| M.H. Rating: N/A |
One more problem: On both nVida and Wildcat the left and right connectros are not genlocked. It means that if you are trying to do dual monitor stereoscopy, you see correct page flipping only on one of them. The second shows a black barr in witch the stereo polarity changes (visible ony with shuterglasses). If you redirect the left-right Opengl output on different connectors on Quadro, one of the output is phase shifted as well. Did anybody tested whatever this produce problem with passive stereoscopic projection based on signal from this outputs ? Are the 2 outputs on Matrox desynchronized as well ? | ||
| ItsikW Rating: N/A |
Hi Michal, I am working with Matrox G450, and the corresponding sync signals of the two VGA outputs are delayed one with respect to the other by aprroximately half a line. An electrical engineer who inspected it on a scope found that. In non-stereoscopic applications such delay is probably irrelevant. Itsik | ||
| M.H. Rating: N/A |
Hi Itsik Did you try following test: Sync. shuterglasses to one of the monitors. Look on both monitors ... You shuld see black barr on of the monitor to witch you are not synced ... Position of this black bar is equivalnet to the phase shift (about 300 line for 800x600 res, but it is variable) ... | ||
| Alexander Oest Rating: N/A |
Hi Michal ..I'm a bit confused, so just to be certain: This problem with a black bar on one of the monitors... It's only when you attempt some shutter-glasses based stereoscopy, right? I mean, the Quadro boards are still perfectly capable of outputting left eye's picture to one monitor and right eye's to the other with Direct3D and OpenGL, aren't they? (if not, they aren't of much use for a dual-projector, polarizing setup). If I understand you right, then I've got another question: What's the point of experimenting with shutter glasses when you've got a working dual output? Panoramas? something else? Alex | ||
| ItsikW Rating: N/A |
Hi Michal, I cannot activate page flipping right now, as I have Matrox running with Win2000. I will do the test when I change my setup. Itsik | ||
| M.H. Rating: N/A |
Itsik: You can do the test even without page flipping. Normal output is a page flipping de facto as well and the black bar is clearly visible even in non-stereoscopic modes. The black bar is just the plasce wher the shuterglasses incorrectly switch for the second monitor ... Alexandr: The effect totaly corrupt double monitor shuterglasses stereoscopy ... The target of this experiment is: 1) have a large stereos scene visibe on 2 monitors 2) obtain signal for 4 projectors for passive stereoscopy (2 CyVis on each graphic card output) 3) Check the output singla for 2 projectros synchronization You are reight that the observed effects makes theoreticaly both graphic card for no use for dual projection system ... In praxis it work a bit different, human eye probably does not require precise sync. of left and right image and the data buffering in human brain can comensete the left and right projector desynchronization. Symply: 1) Yes you get on the connectors signal only for the correct eye 2) This ignal are not syncronized (tou can have on connector first connector signal for frame 1 + part of frame 2 and on second signal for frame 2 + part of frame 3). This shift is constant during time ... The question is what will the human brain do with this ... The whole problem is related to HW not-synchronization of the output RAMDAC clocks ... | ||
| ItsikW Rating: N/A |
Hi Michal OK, I see what you mean. I will do the test when i am back in my lab on Sunday. I know that in my Matrox card the outputs are not synced, but I am curious to see the effect. The fact that you describe is very discouraging. I had an exchange of letters last week with technical supports of both 3Dlabs and PNY, asking specifically whether the outputs of the dual-output cards are synchronised. I was even more specific, and asked whether the corresponding sync signals are identical with zero delay. 3Dlabs answered immediately positive. For PNY it took a few days, but finally they answered positive too. This is in contradiction to what you describe. You surely noticed that high-end Wildcats can be genlocked. It may be the case that the genlocking is effective only in the primary output. By the way, if you connect a separate cyviz to each one of the outputs to create a two-channel passive display, why the sync between the outputs is important? You will not use shutter glasses in this case. Itsik | ||
| M.H. Rating: N/A |
Both PNY and 3DLabs people are telling nonsence .. Independent test made by me on several computers and by Ligtpseed shows that the output are out of sync ... With 2 CyViz I need output for 4 projectors. pairs of them shuld be in sync again ... They will not be becouse they are connected to differeny CyViz on different connector ... I do not thing that it is possible to genlock 2 CyViz and do post-synchronization ... | ||
| Anonymous Rating: N/A |
I had the same problem on a GF2MX using 2 screens in stereo, considering that Quadro2MXR is the same with a simple resistor changed, all old GF2MX boards are not genlocked and not suitable for multiscreen stereo. |
