Page MenuHome

Animated Strip-Time doesnt update
Closed, ResolvedPublic


--- Operating System, Graphics card ---
Windows 7 64bit, Geforce 690 GTX

--- Blender version with error ---
2.68.1 r58588

--- Short description of error ---
When a NLA-Strip is animated over the Strip-Time, the animation is not correctly
updated when the animation is aborted with esc. It seems there is an update call
missing. This was a problem in our pipeline when the scene is send to our renderfarm.

--- Steps for others to reproduce the error (preferably based on attached .blend file) ---
Open the attached blendfile -> play animation -> press esc
The cube should now jump back to its original starting position, but instead it stays
at the last played frame.

Event Timeline

The fact that this even works in most cases is a bit of a miracle/anomaly.

In most cases, when you animate the strip properties, the keyframes for this go directly into to the active action. However, the active action is only really evaluated once "on top of" the NLA Stack - that is, once all the strips have been evaluated = once the strip time has already been used.

A workaround which might work is to pushdown the active action onto the NLA stack, and move the resulting strip to the bottom of the stack (i.e. before any of the strips it affects get animated). This may/may not work, but certainly does have some drawbacks, as it makes it trickier to edit the strip time animation.

What I see is a clear lag on every update, also frame switching (click in nla editor).

Wasn't this just meant to work?

@Joshua Leung (aligorith),
even with your workaround the problem is still there.
Please check

A quick update on this bug.

Indeed, even with the workaround it doesn't solve the problem. As a result, the only possible/reliable fix for this is to reinstate/complete the original design I had for how this feature is supposed to work. That is, each NLA strip would have a number of its own F-Curves, which are used for animating time and influence (among other things). This way, we would be able to isolate the F-Curves which need to be specifically evaluated before any of the other NLA stack evaluation operations take place, getting rid of the delay problems here.

I've got a WIP patch/branch that implements this functionality (and it works flawlessly as expected). The only really "blocking" issues are:

  1. How to expose these in the UI. Specifically, where to put them in the channel-list hierarchy
  2. The new method is incompatible with animation done in the old way. While it shouldn't break old files, trying to insert a new keyframe on previously keyframed time/influence properties will end up adding keyframes the new way (instead of continuing on with the existing F-Curves)

Hey @Joshua Leung (aligorith) thanks for the update!

We just ran into this bug again and I would like to ask if you made any progress on the issues?
Here are some ideas about the blocking issues:

  1. Maybe you could post blocking issue number one to the ui design team?
  2. My understanding is that it has never really worked reliably. So it perhaps wouldn't be such a big deal to break compatibility?