I was working on a little job today with a 2D temporally variant scalar field.

You know, B&W footage.

I needed to find the parts of the data that were changing the most and compare them to the overall data and the maximum delta.

What I ended up with, once Ben pointed it out to me, was a simple example of calculus laid out in a couple tools.   The simplest case is just taking the frames I have and interpolating the same number of frames, so there’s no missing samples.  It’s silly, really.

But you can try it with other sampling, so there’s also an example of a Sobel filter, with a 1D kernel perpendicular to the normal 2D one.  Cute really.

If you checked out my interactive smoothing comp, you can see how I used a Sobel filter to make the forward facing laser pointer by looking at the differentiation of the R and G channels over time.  Same idea, just different way of expressing the temporal dimension.

I’m tossing in a Laplacian filter too, just for fun, it’s not useful for the calculus part, but it was easy to do, and shows how you can change the kernel to make different effects.  It’s possible to also evaluate 2D or 3D kernels this way, too.  The temporal offsets can be combined with spatial offsets so you could make a 3D blur filter, or a 3D sharpen.  Or a 3D Unsharp Mask, as I’ve also included.

Trying out some new datasets and new techniques…

EDIT:  Jim asked for some more details, and I already had some images that I intended to post, but forgot about.  So here’s a breakdown of the three layers used to make the above image…

Chameleon rendered layers

The left layer is an environment map lookup, the middle is a front lit with high opacity, and the right is a backlit with low opacity.   These were then additively composited together.

I also did some tests on this dataset with clipping.

Chameleon culled with spherical gizmo

Chameleon culled with box

The box culling was an accident, but I thought it looked like a cut of meat that had been chewed on by mice.

We’re releasing a simple tool today to help you reduce noise in ’09. We develop many 3D filtering tools at AT, but this one is a good candidate for release because it’s useful for video compositing.

Download Median In Time 1.0

MedianInTime performs a median filter over time for each pixel. The median filter kernel can be selected from sizes 3,5,7 or 9 (indicating the total number of frames in the sort). The tool calculates the median based either on the luminance or the RGB channels separately. This operation will generally reduce noise, but is also great for removing things that appear for just one or two frames such as rain drops, film artifacts, and subliminal advertising.

I rendered two examples to demonstrate the technique. The first video shows how objects moving quickly in relation to other content can be removed with the median filter. In this example the stars moving in this time lapse footage are removed (as well as the odd shooting star).

Read the rest of this entry »