The GLI.TC/H Festival Kickstarter Project, GLI.TC/H 0111!?▐▐▐▀▀▀▀▀▀▄▄▀▀▌▌▌▐▐▐DIT▀▀▀▀▀Do▄it▄2gather▀▀▀▀, appears to have the necessary funding and then some. Building on the success of GLI.TC/H 2010, GLI.TC/H 2011 will take place November 4, 5, and 6 in Chicago, November 11 and 12 in Amsterdam, and November 19 in Birmingham, UK. Check it out. Pledge: an over-the-top festival deserves to go well over the top.
Two large format digital prints by Paul Hertz will be shown in the juried art show at the annual Computational Aesthetics conference, held this year in Vancouver, Canada from August 5 through 7. The archival inkjet prints from the artist’s recent “Sampling Patterns” series, Ponente and Shimmer, were printed at Ignotus Editions.
Ponente and Shimmer are based on regular random distributions known as “blue noise.” Natural phenomena such as identically charged particles jostling for position within a limiting boundary or a flock of birds adjusting their mutual distances have similar distributions. Blue noise dot patterns have interesting visual and cognitive effects: Their regularity seems to imply an order just about to emerge, which their randomness negates. These and other works in my “Sampling Patterns” series are snapshots from interactive real-time animations where the geometric points of the distribution are used to sample functions that control color, scale, shape, and other visual attributes. The snapshots are further edited to produce prints.
In Ponente, blue noise grids determine the locations of distorted circular shapes in different scales and granularities. Low frequency wave functions control variations in scale and simple coloring rules distinguish different layers of shapes or populations within each layer. In Shimmer, a distribution is partitioned into three populations that are distinguished by algorithmically determined colors. Each population has its own shape-generation rule. A global rule for shape orientation (a wave function) creates swirling motions over the visual field.
You can reduce the noise in low light photography by taking multiple images and merging them to extract the statistical mean or median value. These statistical operations are available in Photoshop’s Layers > Smart Objects > Stack Mode menu. Here is a clear example of the effects of the mean and median operations on a stack of four similar images.
The leftmost image is one of four similar images that were stacked into a smart object. In the middle image, the mean value (average) of the four images is used as the value of at each pixel position. In the right image, the median value (midpoint of the range of values) is used. You can find the full median image here.
The noise reduction is pretty dramatic (you’ll have to click on the image and view it full size to see what I mean). I find the mean image somewhat smoother, visually, than the median image, but the median image has some advantages over the mean.
A man in a yellow rainjacket walked across the view while I was shooting. You can see four images of him in the mean image: logically, one fourth of the pixels in the stack at those points belonged to the moving man, so he has a ghostly presence. In the median image, he has practically disappeared: the influence of details or noise that appear in only one of the images is much less marked than in the mean image. The median operation is particularly useful for removing momentary details from a statistical composite.
Of course, the reason the original images were so noisy is that I was shooting hand-held at a high ISO (1600). If I had used a tripod and lower ISO with time exposure, I would not have had to resort to statistical operations for noise reduction. Knowing that the image would be noisy, I shot multiple images and used Photoshop’s File > Automate > Photomerge… command to position the images in layers. Details of how to do this can be found in an earlier post, Statistical Blending.
You can use statistical blending to render high dynamic range images, to reduce noise, or to create multiple exposure effects in Photoshop (CS4 extended edition). All these techniques require that you have multiple images to start with. For HDR images, you need different exposures of the same subject from a stationary viewpoint. The same is true for noise reduction, only the exposures should be identical. Multiple exposure effects can use any number of different images, all of the same dimensions. In each case, you start by stacking all the images into layers, selecting all layers, and converting them into a smart object. Then you use the Layer > Smart Objects > Stack Mode functions Mean or Median to create a statistical combination of all the images, which you can rasterize. Details after the break.