Stereoscopic rendering (and image sequence recording) in Stride

Hi lovely people

I need to create stereoscopic content in stride for 3D classes using shutters. (60fps each eye, scene is being rendered from two parallel cameras “eye-positions”, shutters run on 120fps in total).

  1. Is there any helppatch or node for stereoscopic content which I could not find…?

or if I do it myself…

  1. Basically I am following this overview by Nvidia: https://www.nvidia.com/content/pdf/gdc2011/stereoscopy.pdf
    I can render the scene twice like shown in this Help patch
  • HowTo Render into Multible Viewports (simliar forum topic: https://discourse.vvvv.org/t/multiple-camera-with-render-groups-and-masks/20273 )
    But after that I need to render Imagesequences for each eye using textureWriter (mainloop increment and framecounter for up counting filenames) .
    Since I need two different videos (sequences) in the end I need two different textures and two different TextureWriters. (I think… :/ )
    Not sure whats the best way to do this - since if I use multible viewports on SceneTexture I only get one texture with both viewports and not a spread of two.
    So my solution right now is to crop each texture to the final resolution using two crop(UV)[stride.textures.Utils] and then write the textures into two separate sequences.
    → somehow I have the feeling there is a better way to do this? Any suggestions or hints?

Second option I came up with in my case would be to use one viewoport, run everything on 120fps and switch left/right eye every second frame. But for that I would need to pause the animation every second frame and for reasons of my setup that is even more complicated…

PS.: If I want to save sequences with 16 or even 32 bit I still cannot use the stride TextureWriter? is that correct? At least I had to change format to lower bit rate when saving PNGs. Otherwise I got errors and faulty files.
→ following this thread: https://discourse.vvvv.org/t/writing-stride-texture-to-disk/20291

1 Like

correct.

also correct.

PNG doesn’t work with higher bit depths, you need dds, tiff, or bmp. there was a thread about this a few days ago.

have you had any practical success? By the way - I have been interested in 3D stereoscopic content, when sometimes using other softwares, that did not have stereoscopic output or render modes.

In your case if you are using standard HDMI type equipment - like 3D capable projectors or TV - you could try with Frame - packing mode - What is Frame Packing 3D? – 3D Frame Packing Explained which means you create two vertically stacked frames with a blank strip between them (45 pixel for 1080 and 30 pixel for 720) - ie. Full HD 3D frame has a resolution of 1920×2205 (1080+45+1080=2205) (which would render at 24 fps via HDMI 1.4 or 1280 x 1470 (720+30+720 - 1470) at 60 fps. I had to create custom resolution for my graphic card - connect via HDMi cable and then the projector or TV would recognise the signal and convert to a 3D image.

The best would be patch / node for vvvv that accepts 4-5 modes: v and h split (=single frame) / stack (ie.e double width or height), and frame packing modes.
If anybody has created such patch, I would be interested to receive a test patch for www gamma to try out with the 3D projection equipment i have.

2 Likes