- Posts: 61
- Joined: Thu Sep 20, 2012 6:58 pm
I just wanted to see if anyone out there had any thoughts on how remove an object from a very specific shot type. Essentially, any scene where the object is in front of varying levels of parallax. Such as horizontal (or dolly out) move of a person standing in front of a rolling hill.
So far my thoughts are:
- tracking the foreground object is easy. As is attaching a clone path to this.
- finding a suitable region to clone FROM however is the difficult part. Any other part of the scene would have a very different amount of parallax/scale movement.
- animating the offset is the only solution I can find to mitigate this to some degree. And then using some “randomizing” to blend the edges of this cloned region. However, even the slightest innacuracy to the movement within the cloned region makes it very apparent.
- I have attempted to track/keyframe the movement of various depths of background. But I can’t quite think of a way of making this apply to varying the clone offset.
- and cloning things in sections separated by background depth like above
- some aspects are solved by the nature of cloning (like overall scale, etc.)…which I think is helpful…but then also hurts at times
- Perhaps someone with more experience or smarter than me can figure out how to do math along the lines of: (starting offset position) - (tracked current position of the element that was under the starting offset position) = (new offset position). This would at least solve part of the “drift” to some degree.
Anyhow, any thoughts are appreciated. Thanks everyone.
Marc
So far my thoughts are:
- tracking the foreground object is easy. As is attaching a clone path to this.
- finding a suitable region to clone FROM however is the difficult part. Any other part of the scene would have a very different amount of parallax/scale movement.
- animating the offset is the only solution I can find to mitigate this to some degree. And then using some “randomizing” to blend the edges of this cloned region. However, even the slightest innacuracy to the movement within the cloned region makes it very apparent.
- I have attempted to track/keyframe the movement of various depths of background. But I can’t quite think of a way of making this apply to varying the clone offset.
- and cloning things in sections separated by background depth like above
- some aspects are solved by the nature of cloning (like overall scale, etc.)…which I think is helpful…but then also hurts at times
- Perhaps someone with more experience or smarter than me can figure out how to do math along the lines of: (starting offset position) - (tracked current position of the element that was under the starting offset position) = (new offset position). This would at least solve part of the “drift” to some degree.
Anyhow, any thoughts are appreciated. Thanks everyone.
Marc