More Info. Needed On Clips
TheMusicShow
3 Mar 2008 20:35
by the way, if anyone wants to know, I just composite added the spotlights over a video and the lights appeared over but the strobe was so "powerful" that it covered the person in the frame. u could just see their legs, like a blinding city spotlight on them he
klipper
3 Mar 2008 21:05
Composite modes are either on or off, there is no parameter for how much to add etc. However, you can still change the opacity of the clip normally.
Add, difference, and multiply are the easiest to understand. They do exactly what they sound like. Add simply adds the R, G, & B values (or YUV, I can't remember which) of the top layer to those of the bottom layer pixel by pixel. So if there is a pure white pixel in the top layer, the result will be pure white. If the pixel is pure black, it wont alter the bottom layer. etc.
Difference simply subtracts.
Multiply is how some keying is done. Basically it multiplies the values of the two layers pixel by pixel, R x R, G x G, B x B etc. Imagine the top layer is greyscale, no color, with values from 0% (Black) to 100% (white). When you multiply the layers, whereever the top layer was 0% the result is black (0 times anything is always 0); whereever the top layer was white the bottom layer is unchanged (1 times anything is the same thing). Gray values scale the bottom layers intensity proportional to how bright they are.
Here's how its used to key one layer on top of another using an alpha channel: Multiply the top layer by the alpha channel. Multiply the bottom layer by the inverse of alpha channel (1-alpha); then ADD the two intermediate results together.
What you have to be careful of is multiplying the top layer twice. For example, after effects renders and 3d renders are often 'pre-multiplied'. In this case you only want to multiply the background, then add. You usually don't have to worry about that anymore though, because most applications that can read the alpha channel in a Quicktime file know how to handle it correctly.
I hope this is helpful to you. If you already know most of this - sorry! - but perhaps it might be useful to other Pond5 users.
Add, difference, and multiply are the easiest to understand. They do exactly what they sound like. Add simply adds the R, G, & B values (or YUV, I can't remember which) of the top layer to those of the bottom layer pixel by pixel. So if there is a pure white pixel in the top layer, the result will be pure white. If the pixel is pure black, it wont alter the bottom layer. etc.
Difference simply subtracts.
Multiply is how some keying is done. Basically it multiplies the values of the two layers pixel by pixel, R x R, G x G, B x B etc. Imagine the top layer is greyscale, no color, with values from 0% (Black) to 100% (white). When you multiply the layers, whereever the top layer was 0% the result is black (0 times anything is always 0); whereever the top layer was white the bottom layer is unchanged (1 times anything is the same thing). Gray values scale the bottom layers intensity proportional to how bright they are.
Here's how its used to key one layer on top of another using an alpha channel: Multiply the top layer by the alpha channel. Multiply the bottom layer by the inverse of alpha channel (1-alpha); then ADD the two intermediate results together.
What you have to be careful of is multiplying the top layer twice. For example, after effects renders and 3d renders are often 'pre-multiplied'. In this case you only want to multiply the background, then add. You usually don't have to worry about that anymore though, because most applications that can read the alpha channel in a Quicktime file know how to handle it correctly.
I hope this is helpful to you. If you already know most of this - sorry! - but perhaps it might be useful to other Pond5 users.
TheMusicShow
5 Mar 2008 01:19
holy shit, I'm going to come back to this in a few days klipper, my mind almost exploded thank you
klipper
5 Mar 2008 03:12
I'm glad I could help. But there are much more knowledgeable people out there. For example on this great page there is a great detailed explanation of compositing modes (they are also known as blend modes):
http://www.simpelfilter.de/en/grundlagen/mixmods.html
The FCP manual has a good explanation as well, without the math, starting on pg 394.
http://www.simpelfilter.de/en/grundlagen/mixmods.html
The FCP manual has a good explanation as well, without the math, starting on pg 394.