A half assed way to accomplish my goal (two images "morphing")
This is part of another simple patch for a friend that needs to crossfade between two static images for a performance.
The first part is going ok, thanks to the help of various people here. The second part is a stab at replicating what h was able to do using the application he was using before to do this kind of things, KeyWorks.
In his previous app he was able to "morph" between two static images and the result was the edged and BW’ed.
The effect created a movement of the "cells" of one image somewhat moving to the location of the second image.
I thought jit.repos was the way to do it and wrapped up something really silly to approach his results:it’s intended use is with static images and the top fader fades both the images and the spatial maps..
----------begin_max5_patcher---------- 1117.3ocyZtsjZiCDF9Z3oPkqbIKkZcvXsWM6yQpToLXMLNkOPrEYX2T4cek krIPFN3CxFtAWR9zu9T2sZ0leNel257CxROzei9LZ1reNe1LSWUcLqt8Luzv CaRBKMWlWl7870eyag8TJ4Ako6uEqVVluVlfdQ8Vgr7MDdoeyUksOMeuJQpL OBRcu6BUadKNa6WKjaTVIPoqVhWfHb+pC.lXNDrDi9R8MYeNp+cmzdGUu3ul FpJhO3s.44c7BiiL5Rq0+BDMB407LUVXp4d89mh3vjSOSY7+YNCf0u1iJONq Q3Piv0iOYlJTEmmcl38sh2LFv0+f9R0M8q4yq9Yg6gLzcHigJsw.lSgbfifb OvUYRbjr3xfnYRqPlnmv9g7rIxOBGtAJbCSHLt8vsPykgAoQLGGt21j50j7P k9YuaupuP3F1LuGmEk+NZeVdmMVHBCJXLis8J58sUVGls8pVI7GmQRprrLbq 7C.pPFFcaSmKYl.FeHfXnCsEtPW1Noi3.VxEbHX0kXB4NQm7MycBXvAmtFHK U465NHCNCj3dBRxzARFc7AYXgp6jjs5TRB8jjzIzjTHbDIuQruuqVll+iXo1 IEqCki6bDPFKvfRe6xkDSKBdfKWxbHluWdIXqWk0tcTrX6ULTZMXIvvhghEt zj8QGPjAmQk9FPD.7SKU5SzMZ.6Tr.8FKvCFK5L8zOitmzNwFqzjQJGe+XPl LJqB+XxG6xrnEqZpChUtILwbF7RQcuZFDmpGElN4cNdUOvlJe61DYmMaHA7e iMRKvVbl5Jvh54pwxUVs56oRUQt1kuuiRvN.sI5c6g4MLJXOXGjwbs7FPYQj qVKG.eWtZtaY1gWCij8.TB6hxlnt18AO7RD36lc+4XqpB4t7RzKo4QRDfdQG APVracrpDg6M2p2nrsZBCGb7mSvscMIYeZX2SotNDEXJpfuiJ.E6wUagw.RT aRNNFRzmRHYUMhfzIGss+6Tq1rZkwGDfUtAY3mRjM7hsyVQNod0NqNvvSIt5 2xfL62ifZqohiVFT7TFL2MtfMAsbqKXvyWUiM1SnOAcuLc1XTTisDyhI+NuO V+QHi.y6wKIN6O+fflQQU+myox78EaZTYSZJneORhjkp3LS0nN8hBN6hdKNJ RZNO93dKi1kqSBqVDWoNUsVS9sQS7IUSh1JIX5vTazzDOyQaClDSqlBZgllV iopzduOlfoUSPazDdZ0DoEZxeZm6Z0TGahm5ZU.S5DKp1PpIFTtWQbRUwTQf 86hvz4k2zZfZkLBZULRZkNBZk6ONZEZUPEq0K7j4s7mN52UUDAyvs5O9vIsF rdgwPuTA0NOu5T8V0Zv5EOF5kQVMR5EKFE9xGK81tL+l5EFayjtnaN5UUfgu n4CzeZqII24tNgyLg7ILvbPDbr0jjBqeuXKQuEat8CYzzZfp0mOBr02JV6eE lFzV0ZJDqnOZkBFyfp+XfMsFnVCbuTqMYabvp05GbvzM907+GPaQVlC -----------end_max5_patcher-----------
It’s stupid at best but maybe it is the right direction?
I’d like to implement this in GL domain if this is a direcion good enough..
Any input appreciated
Thanks a LOT
Try playing with one of the sort of convolution effects like jit.wake jit.slide or jit.glop
I’ve had good luck in the past with a sort of cheap morph type look with jit.wake if you tune the fb parameter just right.
Instead of crossfading, you could bang the first image through, have the feedback set pretty high and just bang the second image and they would sort of fade between the two
jit.repos is a good way to go too
for a sort of ‘mercury’ look I’ve combined wake/sobel and repos: http://vimeo.com/628531
send straight video into one input of repos, and then send straight video into jit.wake into jit.sobel into the second input of jit.repos (mode 1) and you can mess with interpbits to change your smoothness
Could be done with opengl, but getting pretty fades can be a little more difficult because of how the framebuffers work
a lot of food for thoughts!
I’v tried the MorphX app and is indeed very powerful.
It’s offline nature doesn’t fit the bill though..
The root of a "morph" effect is a basic xfade combined with a distortion of the 2 images. At the beginning of the morph, the 1st image is undistorted while the 2nd image is distorted so that defined points are positioned to match the corresponding points on the first image. As the morph progresses, the first image distorts to match the 2nd image points, and the 2nd image becomes less distorted. One way to fake a morph-like look might be to have some OpenGL shapes mapped with the images that get distorted in some way while crossfading the images.