Sequence Recorder and VR Pawn, recorded transforms not applied when rendering video
I am nearly pulling my hair out again, this time trying to understand sequence recordings.
I recorded a sequence of me moving around in VR with the sequence recorder. The VR Pawn has got tracked motion controllers and a mesh for the head following the tracked HMD transform.
After recording, when I play the sequence from within the sequencer I pretty much see what I recorded: me waving around the controllers and moving my head around.
But when I render the sequence to a movie, the head is stuck at the default scene root location while the controllers still work as before.
Why is what I see in the editor when I play the sequence not what I get when I render it to a movie?? The recorded sequence contains all transform information for the head, why is this not used? What happens during the rendering process that causes this?
So, first of all, sorry about my frustration in the previous post, sometimes dealing with the intricacies of the engine just drives me to the edge.
But, with a clear head and some cool down time I found a solution to my problem and I think I may even have learned something:
In my VR Pawn blueprint the head mesh follows the HMD location and rotation, updated on tick. When I play the recorded sequence in the editor, I see the head in the correct (and always changing) location/rotation throughout the sequence. This works because this is not a runtime situation, the Pawn blueprint is not active and what I see is really just a recording of meshes following transform information (by the way in this situation event calls from the event track will not have any effect either).
But when the sequence is played at runtime, in 'simulate' mode and even when recording it to a movie the situation is different, it's all a runtime situation and the Pawn blueprint is active, doing what it is set up to do.
In my case that included updating the head position to follow the hmd location and orientation on tick, but as there was no hmd active the location and rotation was always 0,0,0 which overwrote/invalidated the recorded transform information. Disconnecting the tick event from the update of the head transform in the Pawn blueprint and THEN rendering the sequence to a movie gives me the head motion as recorded.
So it looks like whatever action happens in the Pawn blueprint in such a runtime situation has priority over their recorded counter parts.
I may still be missing a bigger picture here, because what about the motion controllers? They were never a problem, they followed my hands when playing the sequence in editor and when rendering the sequence to a movie. Why were they not stuck to 0,0,0 like the head when there was no realtime transform information available? They must be working differently.
I can imagine that it could be helpful to others if this was addressed somewhere in the documentation. Trying to solve my problem I read every single page of it and did not see anything specific on that.
Anywhen, that's good enough for me right now. ...looking forward to the next pothole.
answered Oct 12 '18 at 10:45 AM
Follow this question
Once you sign in you will be able to subscribe for any updates here