Yes, I see what you mean. As mentioned, cameras are tricky - to expand, I don’t believe we can treat them like any other object with a transform because of it’s unique interaction with input and character.
In any case, it’s still rather hard to determine the culprit from what you’ve shared, which is plenty, but of course testing things myself would obviously give me full control, this is time I don’t currently have to spend.
Here are my guesses
- Main Camera Position: it looks like you are logging this value, and it may help to log it throughout the stages of its use, i.e. whenever it is set as a value. I see on Tick, it is a relative set, to some child actor (which is?)
- Get Sight Vector function: is getting the socket transform, which I assume is by default in world space, not local/relative space. Curious… Additionally, the coordinates from the break vector into the make vector could simply be wrong. Might be worth changing these around to see what differences occur - again, I’m not there with you, so naturally I would start poking around at places that I feel are influences around the issue.
- The lerp(vector) for your timelines are choosing between these values to set the Main Camera’s location. The Main Camera Location which is always the “relative” location of Child Actor (maybe its shouldn’t be relative?) and the Get Sight Vector function which is using the assumed world location of the pistol. (Although I do not understand what RTS Parent Bone Space specifically does).
Again, I would suggest removing the timeline aspect and simply set the location. This would be immediate, but I would imagine it reduces the number of places to debug this problem. So instead of a lerp, just use one of two location fetching methods, the Main Camera Position or the Get Sights Vector function, one for ON and the other for OFF.
EDIT: And one more thing to note is that of the Camera itself, since there could be other influences based on the camera class, spring arm, and/or character that can add to this.