The Photo Mosaic Process
Our trained photo mosaic artists use cutting edge software to create breathtaking photo mosaics. We hope this explanation gives a bit of insight into our state of the art process.
The first step in the photo mosaic process is preparing the cell images. Our artists hand crop each picture to a standard 3:4 landscape size. After the cell images are prepared, the background image is prepared. Our artists crop the background image to the particular aspect ratio of your print size (for example, an 8x10 can be either landscape orientation or portrait orientation, but it must fit within an 8x10 rectangle). Our artists, along with the advanced facial recognition features of our software, identify the key elements of the source image that need to be prioritized. Faces, bodies, and other important elements are identified.
When our artists are finished working on the images being used in the photo mosaic, our advanced software begins analyzing the images. The source image is split into an array of smaller rectangular cells. Each of the cell images and each of the rectangular cells within the source image are examined for a number of data points. The cell images and rectangular cells are both broken down into smaller regions, and the smaller regions are viewed from two color spaces - RGB and HSV. Breaking the images down into regions allows our software to analyze patterns within the cell images. RGB and HSV are two methods to describe color within images. They're two tools that do the same job, but we find each to be useful in placing the cell images within the mosaic.
After the images have been analyzed, our software computes every possible combination of cell image and rectangular cell within the source image. A number of mathematical functions are put together and arranged in such a way that minimizing them produces an optimal photo mosaic. It's a complex process, and takes into account a number of factors. The important elements of the source image are optimized first - faces are given priority over bodies, bodies are given priority over the background, etc. The cells are compared using equations based on the HSV and RGB values gathered previously, given increased weight to certain color channels because the human eye does not weigh all color channels evenly. Finally, constraints set by our artists are considered within the algorithms, such as the maximum amount of times a cell photo can be repeated and the minimum distance between repeats.
After we calculate the optimal arrangement of cell photos within the source image, we then use cell colorization to give the image back details that were lost during the photo mosaic process. Cell colorization is where we put portions of the source image in the cell images to create a photo mosaic that more closely resembles the source image. The amount of cell colorization varies depending on the images used. For example, suppose a customer chose an image of a sunset and then selected cell images from their wedding. Because wedding images are almost always primarily black and white, more cell colorization is needed to create the bright sunset using the colors of the wedding images. Also, since smaller mosaics have fewer cell images, cell colorization is used more on our smaller photo mosaics than our larger photo mosaics.
After all of the cell images have been pieced together, the end result is a single image filled with hundreds of memories. The photo mosaic effect is amazing because the cell images are visible when viewed up close, but the background image is prominent when viewed from a distance.