NLA strip's Influence initial keyframe sets to 0 #89339
Operating system: Windows-10-10.0.19041-SP0 64 Bits
Graphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 398.36
Broken: version: 2.93.0 Beta, branch: master, commit date: 2021-05-19 16:30, hash:
Worked: Never (2.80+)
Short description of error
Influence value is resetting to 0 when adding the initial keyframe to the NLA strip's influence channels. This problem gets worse when adding also keyframes to an active action at the same time. The values of the edited action are changed as well according to the new 0 value of the strip's influence. You can see how the action values change in the sample file's graph editor when adding a keyframe to all channels in the timeline editor.
This problem happens only when keyframing the influence channels in the dopesheet and timeline editor, but the Influence property in general has some more issues, also with updating.
Exact steps for others to reproduce the error
Add a keyframe in "All channels" in the Timeline editor or a Dopesheet editor on frame 15.
The influence of the NLA strip sets automatically to 0, even if it had a value of 1.
Changed status from 'Needs Triage' to: 'Confirmed'
@TalHershkovich I am not sure if I understand this:
On top of the NLA there is also an active action with "Add" blend type. You can see in the graph editor that the values of the new keyframes in the active action are modified as well when all the keys are added, it looks like it recalculates and add the values of the current attributes to the active action because of the influence of 0 on the NLA strips.
Feel free to add this to description, but steps to reproduce should be as brief as possible and may contain short description of unexpected outcome.
update to the steps, because it almost made me think it was solved
the bug occurs when the layer incluence receives a key from the graph editor or dope sheet.
With the current demo file, the influence channel is hidden, which means it doesn't get a keyframe so the bug doesn't occur
unhide the channels to get the behaviour
however still an issue in 3.3
Looks like this has been upgraded to a crash. Possibly related to changes in D15003. Will take a look into this.
There are some major underlying issues with the NLA influence & strip_time properties. These properties never get updates, which allows the value to actually get taken into account, since on evaluation, the rna property to actually get written is the fcurve value
Discussed in 2023-09-14 Animation & Rigging module meeting. Since this has been around for a while (based on 6mo bug triaging rules), we're deprioritizing this from a bug to a known issue.
The new animation system will solve the need for this.
No due date set.
No dependencies set.
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?