This video appears to have lots of feedback and blurring going on. This could be done in Max using the jit.wake object and the jit.slide object. The slide object creates a trail effect in various amounts depending on the argument it is given. The video also increases and decreases in size which could be done using the jit.rota effect and sending it various zoom_x and zoom_y positions. The video also appears to increase in brightness and saturation and change colors at various points. These effects could easily be created by using the jit.brocsa and increasing the brightness and saturation. The color changes could be done using the jit.hue effect or changing the colors of the jit matrix. At certain points in the video there also appears to be two videos mixed together. This could be done using the jit.op object and experimenting with various arguments until the desired effect is achieved.

Autechre

Autechre Gantz Graf

I saw the music video for Gantz Graf around the time when I first started exploring 3D in Jitter. I have done a few tests exploring methods similar to this and I think I have a sense of some of the techniques. First, you would want to generate the basic set of forms (sphere, cylinder, plane, maybe a few others) either using the equations directly through jit.expr or by using preset shapes from jit.gridshape and enabling matrixoutput. The resulting matrixes would then be saved as still image files in float32 format. From there, you could either program real-time accoustic analysis to turn the audio into control information, or you could actually use analysis to generate a database of control values that could be used in a procedural, non-real-time fashion. Next you would want to look at how you would use that control data to control morphology through manipulation of the source matrixes. Some methods that appear to be in use in Grantz Graf are: matrix combining using srcdimstart/end dstdimstart/end combos. This would allow you to roll a section of one form through another form. Or a simple xfade between forms. Various jit.op settings would provide compelling manipulation possibilities, such as rescaling of individual points, rows, columns based on spectral energies or overall amplitude. At times changes appear to only affect one axis (x, y, or z independent of the others). This could be accomplished either by using a multi-planar jit.op or by using jit.unpack, manipulating a plane, then recombining with jit.pack. Because of the high resolution of the 3D forms as well as high resolution render output, the final output would almost certainly have to be generated in non-real-time. Thus a procedural frame-by-frame render scheduler would have to be created.

Cory