Three framed prints of Commodore Sanford’s digital art, two of which were eStamps, were awarded First Prize at the Common Grounds Fair, validating the eStamp motif, and the eStamp-of-the-Day series as art. On the back of each frame was a descriptive page as below.
AI software uses the computer to take large volumes of data and identify patterns that are useful to its users. In data representing a single image, for example, a digital photograph, the software identifies features to change in visual characteristics of color, line, form, edges, etc.
Apps like PRISMA, have become popular for using neural networks to produce appealing distortions of a digital photo. Several apps are now available (within the last two weeks, I found four amazing new apps. They are so exciting, that many app creators are inspired to make them.
One such discovery, called ARTOMATON, will load video, which is exceptional. ARTISTO, another app, loads ten frames of video at a time. It took over an hour to load one of my visualization of music video 1-minute video. ARTOMATON loads the whole video, and lets you control these basic graphic functions:
In these images, the original is upper left, and the main window shows the calculated rendition, based on user-set oprions, like brushes, paper, lighting, frames and cross hatch direction and boldness. The result is a sketch-like picture, which surprises, and sometimes delights.
The first picture is the original photo, and the following two are processed by the apps shown. The picture is called contrejour because it is aimed directly at the sun, desiring the effect of shilouette while picking up highlights on the ice-coated branches on this very cold morning. Enjoy.
One particularly exciting facet of computer science is software that applies artificial intelligence (AI) to digital photography in order to generate a new, unique image that may be artful.
AI software has recently evolved to employe neural networks, emulating the human brain, to analyze two images: Content Image, the original photo, and a Style Image. The style image is analysed for its styleelements, such as edges, textures, colors. The content image is analyzed for its shapes, position in the frame, structure. The software transfers certain style qualities to the content image.
The result is a calculated synthesis that may please, sometimes even delight, but, always surprises.
The app Vinci sometimes produces painterly textures that enhance by softening focus and dispersing the light. The goal remember, is surprise and delight.
Why not download Vinci right now and put in your photos ?
You can get Vinci from the App Store or Google Play. It’s easy and free.
Encouraged by the portrait experimental trial, I decided to try a panoramic photo, in some neural network apps. Matissa compressed the image into square format, which was never-the-less interesting. Here is one rendition of the panoramic photo when squeezed into a square:
Speaking of squeezing, try pinching these images to expand, revealing details.
When this is the original after Snapseed:
But, let me show you some of the results of some apps that did input and run panoramic images.
Have many apps to try, so stay tuned, or subscribe.
I made portraits of my customers at farmers markets, and this one has been a favorite of mine. I wondered what a portrait might become in DeepDreamGenerator, the website app.
For the style image, I chose a flower still life.
The result did surprise and delight me.
This photo mistakenly included the whole sun; presenting an opportunity to see how the several neural network apps freely available, calculate variations of styles, shown here.
The wire in the upper right was also mistakenly included.
It is interesting the to see the variety of ways apps distorted the wire as well.
what I am learning is that the apps have broader latitude to calculate wrong with a whole sun in the frame, an area of intense brightness. The style image used will influence the network’s probability weighting, generating a surprising result.