[Reynolds Strother]

I’m sorry, up above in my first example of what to do. I told you the wrong place to put the expression. I said place it on the cameras point of interest and that is wrong. Place it on the cameras focal distance.

Again this forces the camera to focus on what it’s looking at, the POI (point of interest). Put the POI on the text not the particle emitter. Remember that Particular is a 3D plug-in. That means it is just a 2D layer with the “3D looking” particle displayed on it. This means that the 3D text cannot interact with it as it appears in 3D space. (Even though the particle seems like it behind the text, the “stringy particle arms” could be closer to the camera that the text is. Therefore the particle and text interaction can be misleading when arranging them in the comp. The real spatial relationship between the particle and the text is depending on where the EMITTER is placed in relation to the text. This could be your depth of field problem.

Remember that AE is not true 3D. It is 2D planes distributed in 3D space. Volumetric shapes are only illusions in AE. Plug-ins like Particular help project 3D renderings into AE but how they look is not how they are. Think of Particular as being a movie projector shooting the image of a 3D object onto a flat plane.

I love AE and it’s pseudo 3D abilities. I love that people create incredible plug-ins to help AE users get their 3D fix. People judge AE too hard when it comes to 3D, those guys need to look at what they’ve got. AE is one of the most crucial and powerful tools in the world of graphics. It’s always blowing my mind what people can do with it. Using AE to it’s full potential comes from thinking around and beyond what AE (as well as its plugins) can do on the surface.

All I’m saying is keep an open mind to what is happening in your AE projects. AE doesn’t think like humans in unique situations, and it never will. Especially in 3D situations. It’s the users who can think alongside the program who know what they can manipulate and exploit to get AE to pump out incredible motion gfx.

I write too much in the mornings.

[Reynolds Strother]

Two things you could try.

1. If the prob is depth of field, set your focal distance to line up with your point of interest. This way you know what the camera is looking at and focused on. Apply this expression to the point of interest property for the camera…

length(camera,pointOfInterest)

Then create a new null object to control the cameras point of interest. Name it “POI Control”. Put this expression on the point of interest property for your camera…

thisComp.layer(“POI Control”).position

Now you can easily control where the camera looks and focuses. This is when you can adjust the aperture and blur level to what looks best.

2. If is and issue with using the particular effect and your comp settings, render elements separately and merge them later. Build the comp with all the elements together, text, cameras, the particular and so on. When you like your animations, camera moves, and particular elements, render them without the text. Then create a duplicate comp so you can alter how the text is looking with the depth of field. This is when I like to make slight changes to how the camera sees the text. (blur level, aperture, color, effects, kerning and spelling). When your text is looking good, render it out with alpha transparency to be merged with your other render. I do this because the text needs some T.L.C. so it can look nice and legible.

I’ve played around with this tutorial in the past, but only to get that look with Particular, so no specific advice for text in your situation.

[Reynolds Strother]

That explains the strange behavior. What are some better choices for codecs, and thanks for the help.

[Reynolds Strother]

All of my footage is QT Photo JPG and I’ve tried using H.264 and Mpeg-4. All had the same problem.

I’m on QT 7.4.1 now. This problem was present before the QT 7.4 fiasco so I’m sure it isn’t a version issue.

[Reynolds Strother]

Awesome, thanks Dan. Love your site, already have gotten tons of ideas for expression uses from it.

[Reynolds Strother]

Ok here goes my best shot. The idea is to create a bend using multiple thin & tall rectangles. The smoothness of the curve depends on how thin the rectangles are and how many you make. This is basically making facets similar to 3D polygon modeling done in a real 3D app.

We just need to achieve it with what after effects lets us get away with.

Do this in a new empty project to get the hang of what is going on here. Don’t worry about a specific curve or any other fine details. Once you get one basic curve made you will understand what to do to fit your needs.

–Draw a shape layer or make a new solid to get a rectangle on the stage. I suggest a solid to start with.

–Now turn it into a 3D layer, open the layer properties by twirling down the arrow beside the layer name. Twirl down the contents arrow, then the transform arrow so you can alter the scale,anchor point,and rotations.

–Resize the rectangle with scale into a tall and thin rectangle. (vertically tall, horizontally thin, it should look about the same shape as one of those 12 inch rulers we’ve all used in grade school).

–This is Important! Make sure the anchor point is as close to the center of the rectangle as possible to start with. You most likely will need to adjust the ‘x’ and ‘y’ anchor point if you start with a shape layer. You can easily center the z if it is not already centered by giving it a value of ‘0’(zero)… If you started with a solid, like I recommended, it will automatically center itself with the anchor point when you create it. Solids and Shapes have their different qualities. It’s up to you to know which works best for your needs and your knowhow.

–Ok, now that your anchor point is centered on the middle of the rectangle as close as possible (Don’t worry about being precise, just eyeball it), take the ‘Z’ property of the anchor point and change the value to a negative number. You will notice that this is moving the rectangle back in space. Dont move it back too much, just about the same distance as the rectangle is tall.)

–So you’ve moved the rectangle back in space with the ‘Z’ position of the anchor point, now you’re ready to start “curving”. That anchor point you’ve just move away from the rectangle is now the axis of your curve. The distance between the anchor point and the rectangle is the your radius.

–The easy part… On the rectangle ‘Y’ Rotation property, alt click on the stopwatch beside the ‘Y’ rotation to create an expression. Paste or type this into the expression field..

rotation+[(index-1)*4]

Go ahead and hide all the layer properties by twirling the arrows back up.

–All you have to do is duplicate the layer (control + D) as many times as you need to create the desired curve. Here’s the coolest part, that expression you just put in offsets the ‘Y’ rotation in relation to the layer’s order in the stack. So each time you duplicate the layer it will rotate incrementally on the ‘Y’ axis. A pretty useful and simple expression. You can change the last number in the expression to what ever number will space the rectangles properly. I used the number 4. The larger the number the more spacing you get. The thinner your rectangle the smaller number you’ll need.

–Last thing you need to do is select all the layers in the comp – “after you get the desired curve” – and pre-compose them. Now you can reuse this precomp and duplicate it 4 times to put them at the 4 corners of your hockey rink.

Hope that wasn’t too confusing and hope it helps. Once you get the process understood, you can create your own imagery for your rectangles to get the desired surface and transparency you need. I’m assuming you’ll be depicting glass.

[Reynolds Strother]

Need to clear two things up, sorry.

1. The Null does need to be 2D, motion sketch requires it.

2. It is a good idea to take the default expression you get form the pickwhip and divide the value by a large number (10-100).

Looks like this to start with.
thisComp.layer(“camera_z_bank”).transform.position[0]

Now just add the division to the end. I used 90.
thisComp.layer(“camera_z_bank”).transform.position[0]/90

If you don’t do this the rotation will be very dynamic/sensitive to motion sketching. This allows smoother movements.

[Reynolds Strother]

Figured it out. Just in case someone looks into this post I wanted to tell my findings.

By the way you don’t even need a custom expression to do this… And to clear things up I was looking for a way to independently control a slight wobble or side to side tilting of the camera. This is essentially “wiggling” the camera’s Z_rotation by using motion sketch to have some nice organic control of the rotation. (remember that motion sketch can only be applied to a position value, so the trick is to use it to change a rotation value)

Here are the steps:

Create a Null (name it “camera_Z_bank” or “camera_Z_rotation”, whatever the heck you like).

Leave it as a 2d layer. (you only need the x_position value, a 3d layer still works, just not necessary).

Go back to the camera, twirl down the z_rotation stopwatch to create an expression.(alt + rightclick the stopwatch)

Use the expression pickwhip to link the z_rotation of the camera with the position value of the Null you just created. That’s all you need to get the expression.

Now the camera will rotate on its Z axis when you change the X value of the Null’s position.

Apply motion sketch to animate the Null’s position allowing you to give “awesomely, randomized, organic” movement to a rotation value for the camera.

I’m new to using any expressions even in the simplest ways. There could be another way to achieve this but, this is still pretty simple and easy.