Simple image recognition...

Hi,
I’m trying to create something that will take an image and correct it for any rotations/offsets based upon known markers in the image. Is this something that I could reasonably accomplish with Touch? If so, can anyone point me in the right direction?
Thanks,
Michael

and on that note… what does the Slope TOP actually do? What would you use it for in practice?

There currently isn’t a ready-to-go image tracking/recognition feature in Touch. Some of our users use ARToolKit to track markers, and they send the rotation etc. into Touch via Shared Memory to drive rendering.

The Slope TOP tells you the difference between two pixels. By default the R channel will have the rate of change of luminance for the horizontal pixels. So in the output image, if the R channel is 0.5, that means the previous and next pixels in the source image had the same luminance. If the value is > 0.5 it means the previous pixel had a lower luminance than the next pixel. If the value is < 0.5, that means the previous pixels had a higher luminance thank the next pixel.

The Normal Map TOP is a more focused version of this, where it’s meant to use these types of calculations to create a Normal Map TOP to simulate bumps on rendered geometry.

Thanks for the information!

To clarify a bit more, the Slope TOP tells you the difference between two pixels, for each pixel on the output image.