I've been fiddling around with Photoshop CS4 Extended's video capabilities, and I thought I'd try this out.

I've taken three video files, converted each to black-and-white (grayscale if you want to be pedantic), and then made each video one separate color channel of the above video.

The "Mix to gray" adjustment layers all do the same thing: convert a color video layer to black-and-white. This is effectively the same as desaturating the layer, but as experienced professionals are often heard to say, using the channel mixer can give you much nicer grayscale effects.

The "Only *" adjustment layers take that grayscale video, and take away two of the color channels. For example, the "Only red" layer removes the the green and blue channels from that layer. Of course, since the video is grayscale, it's not losing any information—I'm effectively colorizing the whole video layer red.

There is one final step in all this, however: Since each layer has its opacity set to 100%, only the top layer (the red video) would be showing. To fix this, I set the top layer to 33% opacity, the middle to 50%, and left the bottom one at 100%. This is a trick I picked up a long time ago from some tutorial. Because we want to mix together every layer evenly, we need to "average" their values.

If, for example, I only had two layers, I would set the top one to 50%, and leave the bottom one at 100%. This would mix them evenly—50/50. If I add another layer to the mix, I can't make it 50%, because that would effectively make the bottom two layer 25% each. So, I make this new 3rd layer 33%. That way, all 3 layers are an even 1/3 of the mix. (Now, this is ignoring rounding. However, 0.33% is insignificant in this.)

Why do all of this? No particular reason. I just thought it'd be cool to mix multiple video files into the same file, and see if they can be easily extracted on the other end.

Of course, I've detailed all the steps for mixing them together, but what about the extraction step? That's easy—just open the output video in Photoshop again, and, using another Channel Mixer adjustment layer, do the same grayscale conversion mentioned earlier, but this time, use only one channel, and output that channel at 100%. You can see an example of the output with those settings to the right.

Now, you'll notice some strange lines in that frame. They're especially noticeable at the bottom, spelling out the text "Western Washington University". As best as I can tell, this is an artifact caused by video compression. Generally, video compression is least-noticeable in large areas of flat color. Where it fails the most (as you may have experienced with JPEG files) is around edges and in areas with fine detail. To save space, it will flatten out areas that have lots of color change. In this case, this compression is affecting other color channels—and understandably so. The video compressor doesn't know what our intent is here.