[Darby Edelen]

[Zvi_Twersky] “How are music videos made? Would it be correct to create an entire music video in AE?”

That depends on what you want your music video to look like. Different goals will favor different techniques.

For example, a music video could be shot entirely on a Sony Handycam and edited in iMovie if that’s the look you’re going for =) The video is there to add to the feel of the music (which should probably be the focus), if you want a homegrown cheap & dirty feel then the above technique sounds reasonable.

If you’re looking to stylize your music video or rely heavily on motion graphics or even if you just want to color correct the footage then After Effects can definitely be of help. I would stop short of saying that you’d want to create an entire music video in AE though.

[Darby Edelen]

You could use an expression to convert the object’s 3D coordinates to 2D screen (Comp) space and then rotate the layer based on how far it has traveled beyond a threshold value ‘left’ in this case:

left = 150; //The distance in pixels from the left side of the comp when the layer begins to rotate
right = thisComp.width – left; //The distance in pixels from the right side of the comp when the layer begins to rotate back
duration = 50; //The distance in pixels that the layer will need to travel beyond the threshold for the rotation to reach the ‘end’ angle
start = -90; //The angle to begin the rotation from
end = 0; //The angle to rotate to

angle = start;

x = thisLayer.toComp(anchorPoint)[0]; //Transform this layer’s 3D Coords into its 2D coords and use the X value only

//If we’ve passed the ‘left’ side but haven’t reached the ‘right’ side yet
if((x >= left)&&(x <= right)){ angle = ease(x, left, left + duration, start, end); //ease from the 'start' to 'end' angle as the x value increases from 'left' to 'left + duration' } if(x > right){
angle = ease(x, right, right + duration, end, start); //ease from the ‘end’ to ‘start’ angle as the x value increases from ‘right’ to ‘right +duration’
}

angle; //return the angle

That should work!

[Darby Edelen]

So you’re saying you want the audio levels to depend on the difference in Z values between the camera and the layer? i.e. the camera is closer so the sound gets louder?

The difference in Z value between the layer and the camera should be (layerZ – cameraZ) This complicates things only slightly as your new ‘close’ and ‘far’ values should take the camera’s potential position into account as well (if the camera varies between Z values of -1000 and -300, for example, the closest the camera could ever be to a layer would be 100 units (-200 – (-300)) and the farthest would be (1400 – (-1000)) 2400 units, in this case your new close and far assignments would be:

close = 100;
far = 2400;

Your new distance formula should be:

distance = thisComp.layer(index+1).transform.position[2] – thisComp.layer(“yourcamnamehere”).transform.position[2]; //The Z value of the layer beneath this one minus the Z value of your animated camera

If the camera is animated in the X direction it will make the panning effect more difficult to achieve (unless the camera isn’t panning).

[Darby Edelen]

I was having so much fun playing around with this expression that I decided to render out a sample of the (newly altered) expression in action. =) Thanks for inspiring me:

https://video.google.com/videoplay?docid=-1751789126326910391

The effect is most dramatic in the middle portion.

Here’s the [final?] code used:

loudMax = -12; //The loudest that the audio file will ever be in dB
loudMin = -20; // The quietest that the audio file will ever be in dB
close = -200; // The smallest Z value expected
far = 1400; // The largest Z value expected
distance = thisComp.layer(index+1).transform.position[2]; //The Z value of the layer beneath this one

myLevel = linear(distance, close, far, loudMax, loudMin);

myWidth = thisComp.width;

//The following section is new and required for the change to the ‘balance’ equation:
motionLayer = thisComp.layer(index+1);
halfWidth = motionLayer.width / 2;
halfHeight = motionLayer.height / 2;

balance = motionLayer.toComp([halfWidth,halfHeight])[0]; //This is a new form of this equation that converts the XYZ coords of the layer to screen X coordinates for more realistic panning

leftMult = linear(balance, 0, myWidth, 1, 2);

rightMult = linear(balance, 0, myWidth, 2, 1);

myLeftLevel = myLevel * leftMult;
myRightLevel = myLevel * rightMult;

[myLeftLevel, myRightLevel];

[Darby Edelen]

[ramil] ”
What if I color correct first the image – remove the warm tones of the footage? Could I then apply the one you suggested wuzelwazel?”

I don’t think this would be very helpful as whatever you apply to the image will affect both the horse and the person bringing you back to square one.

You could make the horse and the person blue, but then they’d both be blue… how do you separate one from the other?

[Darby Edelen]

Ah yes, I forgot that Audio Levels are stored in a 2 dimensional array (like 2D position data) the first entry in the array is the left channel’s level and the second is the right channels so the expression should actually be:

loudMax = -5; //The loudest that the audio file will ever be in dB
loudMin = -80; // The quietest that the audio file will ever be in dB
close = -200; // The smallest Z value expected
far = 1400; // The largest Z value expected
distance = thisComp.layer(index+1).transform.position[2]; //The Z value of the layer beneath this one

myLevel = linear(distance, close, far, loudMax, loudMin); //As the Z-position of the layer below this one varies from ‘close’ to ‘far’ vary the return value from ‘loudMax’ to ‘loudMin’

[myLevel, myLevel]

Thanks for catching my error, this is what I get for not testing my expression in AE (; This will give the audio the same level in the left and right stereo channels. You could improve upon this by lowering the amount coming out of the left channel if the ‘bug’ is on the right side of the screen:

myWidth = thisComp.width; //The largest X value visible in the Comp window
balance = thisComp.layer(index+1).transform.position[0]; //The X value of the layer below this one
leftMult = linear(balance, 0, myWidth, 1, 2);//As the bug moves to the right we increase the left multiplier to 10 (dB are measured in negative values… 2 * -5 = -10 = quieter)

rightMult = linear(balance, 0, myWidth, 2, 1);//As the bug moves to the right we decrease the right multiplier to 1 (1 * -5 = -5 = louder)

myLeftLevel = myLevel * leftMult;
myRightLevel = myLevel * rightMult;

[myLeftLevel, myRightLevel];

This would go just above the final statement [myLevel, myLevel] in the previous expression (which would now be deleted). Here is the final expression, and I promise I tested this one! I even tweaked some of the numbers to my liking, you can mess with them more:

loudMax = -5; //The loudest that the audio file will ever be in dB
loudMin = -80; // The quietest that the audio file will ever be in dB
close = -200; // The smallest Z value expected
far = 1400; // The largest Z value expected
distance = thisComp.layer(index+1).transform.position[2]; //The Z value of the layer beneath this one

myLevel = linear(distance, close, far, loudMax, loudMin);

myWidth = thisComp.width;

balance = thisComp.layer(index+1).transform.position[0];

leftMult = linear(balance, 0, myWidth, 1, 2);

rightMult = linear(balance, 0, myWidth, 2, 1);

myLeftLevel = myLevel * leftMult;
myRightLevel = myLevel * rightMult;

[myLeftLevel, myRightLevel];

[Darby Edelen]

Fractal noise should work fine for the orange flaming glowy body look. It’s not the ideal solution obviously, I’m sure that in Fantastic Four they used footage of a lava flow, glowing embers or some other real world firey glow and used this as a base to texture a 3D model of their actor. Fractal noise will look comparatively flat, which can be assuaged somewhat by creative use of blending modes, but it’s an AE only solution if you’re short on lava at the moment.

Particles will work better for any flame emitting from the body. I think a combined approach to build an animated glowy base (Fractal Noise) and more a dynamic shroud of flames (Particle Effects) should work well.

[Darby Edelen]

In that case you’d have something like this applied to the Audio Levels of the sound file (assuming that your audio file is directly above the associated moving layer in the timeline):

loudMax = -5; //The loudest that the audio file will ever be in dB
loudMin = -40; // The quietest that the audio file will ever be in dB
close = -200; // The smallest Z value expected
far = 1400; // The largest Z value expected
distance = thisComp.layer(index+1).transform.position[2]; //The Z value of the layer beneath this one

linear(distance, close, far, loudMax, loudMin); //As the Z-position of the layer below this one varies from ‘close’ to ‘far’ vary the return value from ‘loudMax’ to ‘loudMin’

Unless there’s something more complicated you’re looking for this should do it.

[Darby Edelen]

Whenever you compress something you’re going to lose quality (unless it’s lossless compression, which doesn’t compress your file very much). H.264/AVC is a good codec (it’s what is being used for most HD/Blu-ray DVDs) for quality:file size ratio but you’ll still lose quality.

Depending on where your video is headed I’d recommend H.264 (Web, HD-DVD), MPEG-2 (DVD) or Uncompressed (Anything Else).

Note that you can change the data rate of most compression codecs. Higher data rate = bigger file but generally better quality.

[Darby Edelen]

If you’re trying to link the Audio Level to the Z values of the layers that should be fairly straightforward. I think the only thing you need to think about is how you want it to be carried out, do you want only one instance of the sound that takes an average Z value of all the layers? Do you want multiple instances of the sound that react to specific layers? The end result you’re looking for will determine how to carry it out.