Minority Report style mixing

[Asa]

Hi

I wonder if anyone may be able to help with a couple of tech questions.

I am working on an art installation in which i want to create a 'minority report'-esque mixing platform.

The idea is that as the viewer moves his arms, the images move and become mixed.

I am currently using MAX/MSP and SoftVNS as main tool.

Does anyone know of any way to mix more than 2 videos on Max/MSP? I found many patches which blend between 2 videos or sometimes 3 but i havent found a way to mixes between 5/6 videos.

Or is there a way of having a number of videos on, but being able to control which single video becomes displayed in main viewr? i.e. by having a number of buttons which controls which video gets viewed.


Any help would be very much appreciated.

Thank you

Asa

[dlublin]

Hey Asa,

Sounds like an interesting project, can you tell us more about it?

If you haven't already, you may want to post this question to the SoftVNS group. It's on Yahoo Groups Canada, contact David Rokeby for more info on how to sign up.

To answer your question: check out the object v.composite

Each inlet (video stream) has it's own blend / opacity level (0-255).

There are a bunch of mix modes available for each stream. Really depends on the aesthetic you are going for, they are all pretty easy to use.

In blend mode, the streams are layered, so if the first stream is set to a blend level of 255, you will not be able to see any video behind it. Sort of like Photoshop. The math / logic for getting a perfect blend of 5 or 6 streams gets a little hard when you have streams fading off and on.

To get you started, if you want to fade from one stream to another, you'd have the 2nd stream at a blend of 255 and just have the top stream go from 255 to 0 (or 0 to 255). To have an equal blend of two streams, you would set the top layer to 50% opacity and the bottom layer to 100%.

v.composite also lets you position each stream - you set the top left coordinate, use a v.crop or v.resize to change the size of the stream before it goes into v.composite. This makes it possible to actually move the video streams around in realtime, though it is very processor intensive and hard to make it look perfectly fluid because it doesn't have sub-pixel accuracy (though you could process everything at double resolution and then downsample).

I'm assuming you've got some idea of how you are going to get control data into the system, so now all you have to do is map that to the mix levels and positions.

- David