Hey Toorop

I think the patch you are using is a great starting point (especially if your projectors can do lensshift and you don´t have to handle the fine positioning within jitter). I just added a softblend-gradient to your patch. For that to be in the right place, one of your pictures needed to be flipped inside your jit.gl.pix Now you might need to re-flip the output of one of your projectors to overlap the images..

best, j

Thank you very much ! I like your way to dissolve borbers ! But your patch has a problem : it dissolves the picture too... : Take an image with 500 pixels verticaly, split it in two : 250px, with 20 pixels for the overlapping, and the image height is now 480 pix ! Am I clear ?

Yes, you are right. In the overlapping range pixels are lost. But there are enough horizontal pixels in your setup anyways, if i got that correctly. If using two 1024x768 projectors for your square projection you´ll end up with a screen of 1024x1024. Scale your content to fit the horizontal 1024 and move it to the bottom/top. Overlay and blend the duplicates.

Try this:

You're great ! Why gen is so poorly documented... I don't get your [* 2. 0.6] : I don't want any change of the scale. Can you explain this ? But the [- 0. 0.6] alone is perfect to lens shift the upper beamer !

Thanks again, I'm in the right way now...

Honestly, I don´t really get the gen-stuff as well. There definately could be a better documentation. But there are so many amazing patches and examples floating around, that can be investigated to learn - I just need to invest some more time, I guess.

From OSCII´s original patch I learned that [* ] and [+]/[- ] accept to numers as argument. In this case the first will affect the X- the second the Y-coordinates. The [* ] will scale the image, the [+ ]/[- ] results in offset. We need to scale down the image, since we can´t just split it in the middle if we wan´t some overlap. I can´t really explain the underlaying vector-maths, but eventually someone might drop in to bring some light into this...

I just fiddeled around with those numers to get the desired results. I kind of knew what I wanted, since I´m working on a similar project using two projectors for a floor-projection setup. But I´m currently talking a different approach using two gl.cameras and two gl.cornerpins. You´ll find patches demonstrating that technique in this here: https://cycling74.com/forums/sharing-triplehead2go-rendering-setup-updated-for-max6-and-jit-gl-node

good luck, j

Oh, and I just realized in my patch there was a missing connection to feed the gl.node´s output into the texture:

Is there maybe someone out there, who can hint me to a way how to implement the jit.gl.cornerpin´s functionality within jit.gl.pix? - So the use of multiple gl.cameras and the capture procedure using jit.gl.node (as it is done in DTR´s patch) would become obsolete, hopefully resulting in a severe performance boost. That would be the perfect addition to the patch above...

It feels like that should be easily possible. But I don´t really find a way how to set the corner´s coordinates individually..

Hi COBTOB, this is my flavour to your patch. It crops a webcam to a perfect square texture. It does overlapping in the second screen with a new mask. Thanks again. This forum rocks !