Offset Helper

You can "extract" the offset/distorted space from the smudge by running an XY map into it, i.e. put x (horizontal gradient) into red and y (vertical) into green. You then extract the channels again from the output. You can use it directly with Lookup to remap a texture as if it were processed by the smudge (note that this is not exact, but close).

Here's a visual picture of what Sphinx is describing. No need for math components

The nerdy part is that we're processing the data in a single stream instead of having two smudge components or more. This way we don't add more render time plus it's easier to handle our output provided we only need to control 1 component to get the desired result. You then just need to extract the outputs from each color channel.

Extra tip: you can feed up to 4 separate black and white images into a single stream using this method. The three obvious ones are by connecting the RG and B channels, our last image is the alpha channel. In our example we only need 2 pieces of data, the X and Y output for the distortion.

Very cool image

Thank you Sphinx/Skybase. I did not know about this technique. However, it is not faster than simply running a texture through a smudge component.

Indigo - the point is that we want a method to replicate the distortion of a smudge several times (since it is quite expensive) with different inputs - as if we had multiple smudges with the same settings. Skybase only showed one remapping/lookup - you can hook up as many as you want. Moreover the actual purpose with Burts challenge was to get something for horizontal/vertical offsets or similar (I think)

I reckoned the distortion map (XY map) could be a step in the right direction.

Burt - did you hook the X and Y channels directly to the bomber (where?) or how did you use them?

Think about how coordinate space/a classic coordinate system would "look" if you were able to view the dimension/axis values. Horizontal values near 0 would be dark and values near the end would be bright. In FF we have 0,0 as the top,left position and 1,1 as the bottom,right. So we can easily construct that with gradients and pass them through distortions etc..and use the result for various purposes...