I just watched the 'Beauty Techniques in Flame" Logik Academy video, and I thought @andymilkis did a great job of breaking down the core tools of 2D cleanup.
These get me through the vast majority of my work:
Painting a clean frame
Front Source offset
Blur (Infill / Median / Object Obliterator)
UV Warping
Are there any major cleanup techniques I missed?
What’s your go-to if the standbys don’t work?
-Ted
Are we talking beauty cleanup or just cleanup in general? I do very little beauty, but endless streams of everything else. My immediate go-to is stabilize-paint-unstabilize. Substitute any of the other processes for paint at any given time depending on what works best. I have my own custom batch tools for the stabilize and unstabilize, but sometimes I do motion vectors, either alone or in conjunction with my other tools.
Frequency separation! Being able to work separately on color (low pass) and detail (high pass) and then recombine is a wonderful thing
Cleanup in general.
I mentally file stab+unstab & motion vectors under tracking techniques as opposed to cleanup tools, but they’re great to keep in mind. All hail the Repo-Master!
What other processes would you substitute for Paint?
Good one! Maybe Frequency Separation is usually a subsection of Paint, but could also be used with an offset?
Frequency separation, object obliteration, fluid morph and all the other tools mentioned, but in general I never do anything without stabilizing it first unless it’s a lock off shot and the subject area is not moving either, so I suppose I think of that as a cleanup tool.
Maybe the same thing as front source offset – but one thing that has become indispensable for me, is using an Action 3D shape with a projection offset. Works great for removing tracking marks off of a screen.
I recently used a combo of UV warping the underside of a laptop, then using 3D Action shapes to remove stickers.
Well, I mean, you can use frequency separation any way you want. Any trick you can use on RGB footage can be used on your high and low passes. I do procedural stuff with it all the time, not just paint by any means. The main difference being that you’re able to separate color and detail. I also give full credit to @suzannemdyer and @Mburdick for opening my eyes to the wide and wonderful world of frequency separation. They slay, and changed my life for the better.
Thanks for bringing up 3D shapes!
What’s the advantage of using a 3D shape over other methods?
In the tutorial you linked, it seems it could have been just as easy to track in a plain surface.
Yes, I recently used frequency separation (blur, subtract) to be able to stabilise a very shaky background plate. which I had to composite as background of a car greenscreen shot. I was not able to find any good tracking points, but with frequency separation and some CC I got some details in the sky, which I could use to stabilise. Magic!
Actually, maybe this was the tutorial I was looking for:
It’s similar to offsetting a front source. Drag a 3d shape into Action, draw the gmask to select the object for removal, do a planar track, add a projection on the 3d shape, then shift the projection axis to offset the selection. Works great on removing any screen markers, and helps maintain reflections built into the screen. I use it a lot when I need to keep some of the built-in light shifts on a removal.
Does anyone else have issues with their projection shifting when you enable Media Projection and copy the keyframe data from the Axis node with the tracking data to the Media Projection’s axis node (05:49)? Can’t figure this out, what am I missing?
If you copy the tracking data, I think you end up doubling up the tracking offsets. Grant was just tracking the 3D shape to have it follow footage. There shouldn’t be a need to copy the tracking data (at least in this example).
At 5:49 he’s just adjusting the tracker position of the outer two tracking boxes, he’s not copying data.
Thanks Jan. I’m referencing 06:06 - 07:00, a little bit of copying and pasting here. Grant emphasizes pasting on the reference frame, and that could be where I missed
I think @rroth42086 is referencing the first video with the trash bin, not the video with the logo removal on the sweatshirt.
Copying and pasting keyframes can be tricky. The frame you are on when copying and pasting will make difference. I tend to go to the first frame after I select a range of keyframes and then do the copy. Also go to the first frame where I want to paste and then paste. This makes me sure I’m copy/pasting the exact thing.
You also have to make sure you’ve selected the same type and number of channels.
Ah yes, I was looking at the wrong video. But I think the relative nature of copying animation is the part to dig into.
Thanks @Sinan! This indeed resolved the issue: the frame you are on when copying and pasting makes all the difference.