Kalleheikki Kannisto
1
COW Coin
Forum Replies Created
-
You need the green screen only behind the person, doesn’t have to cover the image area, just the area where the person moves; everything else get’s masked out and replaced.
If you don’t need a moving camera, it is pretty straightforward, plenty of tutorials on the subject can be found. (Although quality results with greenscreening is not a trivial task, first thing I had to do was get myself a new camera when I started.) Plentiful and even lighting is crucial.
You could make your backdrop in Photoshop as a 2D still in neutral color (shades of grey), and add color lighting in AE.
Or, if you’re more adventurous, create a 3D scene from 3D solids.
-
Right, there’s an opacity setting for individual masks as well…
To do this with one solid you’d use two copies of the same mask, the first one set to subtract 100% opacity creating the hole, the other one set to add, changing opacity, filling it back in. Also, you could now fade the outside without affecting the inside by changing the opacity of the first mask.
-
Changing the last line to “s” instead of “[s,s]” should do it. Scale is a two-dimensional value (x,y) while rotation is one-dimensional (angle).
(Actually, deleting the last line would do it as well, since the next to the last line gives the same value.)
-
For lens Blur Radius:
linear(transform.opacity, 0,100,11,2)this takes the opacity value and linearly interpolates it from 0-to-100 range to 11-to-2 range.
-
By reversing I mean literally time reversing the action for the particles coming together. I took your render and used the beginning part, time reversed it for the coming together. (You’d want to ease the motion so it isn’t this rigid.) Particle life is pretty important here. You don’t want to have particles being born and disappearing if you want it to look like they are forming the word.
In that way Form is a better option as the particles are persistent. Also, you can fill the area evenly with less particles in Form.
But it doesn’t look half bad this way either.
-
Looks pretty good to me with the combination of single-point tracking, position keyframing, precomposing and Warp Stabilizer. The lens flares are the main thing that jump around now. Maybe some sort of a frame bend would help lessen that effect some.
-
Based on John’s theory you might be able to counteract the contrast effect by adding a Brightness/Contrast Effect with the settings -8,6 contrast, -1 brightness, Use Legacy Mode box checked. While not an answer as to why the shift is happening, maybe some such workaround will allow you to get through the current job and then research the matter further when not under a deadline.
-
(That is going to look quite wicked, if not scary…)
The key here is Face Tracker. You can attach other facial elements over the footage once you’ve tracked the features.
The next thing is figuring out threshold value at which a feature changes (such as an eye blinking or mouth movements and expressions) and assigning different images based on it, which will be a matter of expressions.
The other option is to always use the same overlay features but use the Puppet tool to animate them based on the face tracker points.
(I certainly hope you’re not tryng to convey positive emotions, for this approach will have a Frankenstein’s monster feel.)
-
The point was that AE doesn’t evaluate only the slider that changes value, it evaluates all three. So you end up with the logical operation (OR) between the outputs of all three slider values, NOT the output of a single slider.
Hence, you need to tell AE what you want to do with all three slider values, i.e. you need a mathematical operation with the three value, not a logical operation. (I was trying to explain what the result of the logical operation between the values would be.)
If you want just one of the sliders to affect the transform, you will need a selector that tells which one. OR doesn’t do that. Another slider with the value 1, 2 or 3 could be used to determine which slider’s value you want.
Now, if you really want to try and do what you describe without adding another slider, the way to select one of the three slider values would be a While loop that checks backward in time frame by frame for all three slider values to see which of the three slider values was changed most recently, and you would then use only that slider’s value. It can be done, but I can’t recommend this approach as it will likely slow down rendering.
-
The Face Tracker is 2D, so I don’t know how you would convert that to 3D data. Probably you’ll need a slider or two wherein you add the 3D rotation information by manually keyframing the slider(s).
I guess one could try and figure out an expression to extract the head rotation from a triangle of the nose and eyes somehow (the nose turns more than the eyes, so you’d be able to tell which side and how far the head is being turned to with some trigonometry).
Personally, I’d rather use the 3D camera tracker to attach something 3D, and use the Face Tracker for 2D effects.