I am building a 3D animation for a video background. It will be moving and stacking tower of TVs. It will spawn multiple screens, stack them and then turn them on and off, display noise patterns on some of the screens, turn some of them of etc. but in the majority of time it should show random clips of footage on the screens. The question is: What is the best practise to replace the screens and animate it. I was used to 3DsMax and am using now blender for modelling. Rendering is done in blender and octane. I would normally now create cryptomatte for each screen (just a pure green material for example) and track it in davinci resolve or nuke and transform an image onto it. That would also mean that I would have to manually take care of all obstructions.

Current approach is to render the screens directly in blender and animate them there. That however takes quite the dent in my render time as I have to load 20 video footage files and blender is handling this really bad. …

What is the best practise for this situation? What software (i am totally open as long as it saves me time and gives me the needed flexibility) would you use? How would you approach this?

I am gratefull for any tips:

enter image description here

  • $\begingroup$ Tracking, creating plane deformation and masking and all that into a different software sounds like a convoluted workflow (and not a very precise one). Yes blender's handling of video is clunky, but it will deal with the mapping and keying automatically. Try using eevee as render engine. It sounds like you could use some scripting to automate a lot of the workflow as well. $\endgroup$ – susu Sep 22 '20 at 16:34
  • 2
    $\begingroup$ Rendering it directly in Blender seems like the most straight forward solution, so optimizing this would likely be the best approach. Have you tried scaling down the clips? If they only cover a small part of the screen for the entire animation, you may not need a full resolution video. $\endgroup$ – Robert Gützkow Sep 22 '20 at 16:43
  • $\begingroup$ Hey @susu yes, i think so to. I think also I probably should have setup a different topology for the objects. My original plan was to have a material storing all links to the video footage files in an array and then randomly select from that array within the material for the displays. But my coding is still too shallow for that (super easy) task. $\endgroup$ – Avion32 Sep 23 '20 at 10:10
  • $\begingroup$ @RobertGützkow hey, yepp. I have reduced them actually to tiny snippets. Also I thought it might be good to make one longer clip with all the different footage clips in it and offset only the video frames in the texture. So that blender only has to initialise one video file not 20. BUT i think the real issue is, that blender is re-reading the whole list of video files every frame due to the automatic update. Is there a way to cache the textures and video files for the animation? $\endgroup$ – Avion32 Sep 23 '20 at 10:12
  • $\begingroup$ You could try to split the video into an image sequence. While this isn't elegant it will likely avoid the issue you're having. $\endgroup$ – Robert Gützkow Sep 23 '20 at 19:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.