Using Adobe After Effects, the footage of the nights sky is sped up to two minutes; (the rough length of the experience, so it is digestible, see above) and adjusted using the threshold filter, which translates it into pure black and white pixels (see below). The particle plug in program Trapcode Particular changes the footage into a layer emitter.
The footage taken and edited in its threshold filtered state is set as the layer emitter. The brightness of the image is programmed to control the velocity of particle emission, so where there are white particles, there is a high rate of emission. To put this into a real context, picture a cardboard cutout stencil with paint passing through it. In the animation, cloud coverage and stars appear as white blocks, and so emit particles.
The amount of particles in the scene is the key controlling factor, and is dictated using the following methodology, which also provides the translation of the image to binary.
Emission is controlled on a per second basis, with film running at 25 frames per second. From the two minute film, one frame is taken every second, and saved as a jpeg image file. This results in 120 jpegs. The images are then loaded into ImageJ, a Java based image analysis program, developed at the National Institutes of Health, primarily used for three dimensional live cell shading and radiological image processing. By adjusting the image threshold, the program can analyze each individual pixel, placing the results in various groupings. In this case, the program has been used to count each pixel’s brightness level, on the RGB scale of 0-255, 0 being white, black being 255 (see below).
The results for each second is then logged, and divided by 25, (the number of frames per second) to generate the rate of emission within the final animation (See below).
With only two sets of numbers to work from, each pixel is effectively on or off, in binary 0 or 1. By copying the input data into Trapcode Particular, the particles behave in accordance with the original footage, emitting however many particles are determined by the data unique to each second for the 0’s. Assigning each particle a sprite graphic of the binary number 0, as particles are only emitting from the white, ‘on’ areas of the original footage, means that rather than particles, the image now has a sea of ‘0’s’ instead of small white dots (see above). This method disregards the black areas however, which have their own data, under the number 255, or 1 as it is assigned in binary, being ‘off’. To emit from the black areas, the original footage of the nights sky which is set as a layer emitter is duplicated, then inverted (see below).
This footage is then set as another layer emitter, with the rate of emission per second controlled by the data of the black pixels, or 1’s. a new sprite of the number ‘1’ is assigned to the particles, which are now emitting in the opposite manner to the ‘0’ layer of footage, so the whole image is covered with 1’s and 0’s instead of black and white pixels. It is worth mentioning at this point that both input layers of the original footage are hidden, and not viewable, it is only the particles with their attached sprites that are on screen.
The result of the two combined layers is a mass of thousands of ones and zeroes, behaving independently of each other, some staying fairly stationary, others tearing across the screen ferociously, dependent on whether or not they are conveying movement or steady light sources on the original footage.
No comments:
Post a Comment