In order to render a 3Di UI in stereo with the NVIDIA 3D Vision Kit, first ensure that the kit and associated hardware has been installed correctly. Go to the NVIDIA control panel and click 'Enable Stereoscopic 3D' in the Stereoscopic 3D section. Setup can be verified by running the Stereoscopic 3D Test from the NVIDIA control panel.
Next, launch your application or the D3D9 Scaleform Player in full screen mode. The automatic 3D Vision support only works in full-screen, and requires the Direct3D Scaleform Player, therefore the application must start up in full-screen mode. In the case of Scaleform Player, this requires passing in the command line argument –f. Depending on the settings in the stereo section of the NVIDIA control panel, the application may start up with stereo 3D disabled. If the 3D effect does not appear immediately, hit (CTRL + T) to toggle stereo mode.
Once the stereo effect is active, it is important to adjust the settings. Objects at screen depth should have zero separation and no stereo effect. Similarly, objects which are relatively closer and farther away must have the appropriate positive and negative separation to appear out of or in to the screen. Set the 3D Vision depth and convergence values that look best for the application and then save them in a registry profile for future use.
Normally when using 3D Vision only control over the level of depth is possible, by either turning the scroll wheel on the IR transmitter or by using (CTRL + F3) /(CTRL + F4) to decrease or increase it. This is normal, since for most games the convergence does not require adjustment because everything has been set in the game’s profile. For custom applications or when using Scaleform Player, the correct convergence must be set the first time. The default keys to change the convergence level are (CTRL + F5) and (CTRL + F6), however note that they are not active by default. Follow the steps below to activate this feature.
In order to control the level of convergence one must first “Enable the advanced in-game settings” from NVIDIA’s Control Panel by going to “Stereoscopic 3D”, choosing “Set up stereoscopic 3D” and then opening the “Set Keyboard Shortcuts”. When the advanced settings are enabled, the convergence can be changed using (CTRL + F5) and (CTRL + F6) and also use the (CTRL + F7) key combination in order to save the selected custom settings. Saving the settings allows users to skip over the convergence configuration step the next time the application is run.
When adjusting convergence, unlike depth there is no graphical representation on-screen showing the convergence level. Therefore when changing convergence one must carefully watch the changes on the screen. The best way to proceed is to view a scene where an object should be at the screen depth (or perhaps a middle/reference depth) and then hold down (CTRL + F5) until the image separation starts changing. This can take 10 to 15 seconds as the convergence is being adjusted in very small increments and will not be noticeable at first. Hold down (CTRL + F5) or (CTRL + F6) until the object at screen depth appears to have no separation of images (when viewed without glasses). This will set the convergence to 0 using the object at screen depth as a reference point. Again, once the convergence and depth amounts selected are deemed correct, pressing (CTRL + F7) will save those settings to the registry for the application to use in the future.
NVIDIA has provided an API to access and control stereo 3D behavior programmatically, as well as some documentation on best practices when designing a game for stereo 3D. Please see the links below for more detail:
API : http://developer.nvidia.com/object/nvapi.html.
Designing for stereo http://developer.download.nvidia.com/presentations/2009/SIGGRAPH/3DVision_Develop_Design_Play_in_3D_Stereo.pdf