Sanity check on first OF project - video mirror

Hello OF
I’m about to dive into my first OF project and I would very much appreciate thoughts of more experienced OF folks on the following approach.

I have experience using Processing but OF is new to me. I’ve been looking to do something in OF for a while and it seems like the speed benefit over Processing might be useful for this project.

The basic idea is a program that takes video inputs (people in a room) and projects stylised ‘shadows’ onto the wall based on the video input. There are also a couple of effects based on motion/stillness and overlap between shadows in different sections of the screen that I’d like to incorporate, but for now I’ll just describe the basics.

Hardware setup: Mac Pro with dual video cards; 3 x projectors; IR lighting and IR sensitive webcams (likely Unibrain Fire-I). IR lighting because we want to keep the visible lighting in the space dim.

Blob tracking and processing: out of focus webcam -> threshold in OpenCV -> OpenCV detect blobs -> render shadows based on blob’s outline points. Using an out of focus video source and appropriate threshold levels gives a nice ‘stretchy alien’ look as necks/limbs become thinner.

Projection: the plan is to project onto a series of flat screens placed along a ~25ft wall at varying (but not too dramatic) angles while masking out the spaces between the screen. It looks like Theo’s work on warping is the place to start for this

Any thoughts from people who know their way around this stuff?

Thank you, Patrick

Hello Patrick

Welcome to of. I have aswell done the transition from processing. And all excep one thing sounds pretty straight forward to do. Projecting across multiple graphicscards is not very effective. There is another topic here about that. But the problem is simply that the framerate drops quite alot. It might not be a problem for you, but you have to consider it. It might be easier to simply use 1 grapicscard, and a triplehead2go. But i might be wrong.

Jonas // HalfdanJ

yes, i have to agree with Jonas. I recently had a big disappointment with a mac pro with 4 graphic cards. With more than one graphics card, the image rendering is only being processed by one of the cards, then the necessary information is copied to the rest of the cards which causes a lot of overhead and slows everything!!
Read the thread linked by Jonas there is a lot more information there.
other than that OF seems perfect to what you want to do.

Thank you for that advice. I’ve read through the linked thread and will likely try the combination of a NVIDIA GeForce 8800 GT and one or two Matrox TripleHead2go/DoubleHead2Go units, instead of multiple graphics cards.

Good to hear that I’m probably on the right track with the rest of the plan. No doubt I will be back in the forums as I work along my OF learning curve. Looking forward to it.