The original crayon work looks like random drawing that anyone can make out of the boredom. In fact green and black scribbles are for the depth, the darker background, and red and lighter should be trunks, or the foreground. AI sometimes make mistakes in interpretation, so it does something that we humans do not. It messes the image so it loses 3D. But, even if it happens across the image and it still looks on average good, AI done its job.
Ordinary people will not notice anything unusual and think this is the abstract painting, but it is not. All this is digital.
The original scribble looks something any of us would draw on a bad frustrated day.
In fact, if I was more careful and detail oriented, the result would be far better, but I done this just because.
Just a lot of lines made with green, red, black and yellow and nothing else.
After deep dream AI modification it looks like this:
Now instead of the lines you got a moon (upper right corner), something that looks like a path under it, and a forest stream right under it, and the ghost forest on the left.
This is the template which I used to get the second image from the first. I always liked this painting for the trees.
This painting is made by Ivan Generalić, a Croatian painter in the naïve tradition.
Links:
The thing I did is I gave AI my very messy scheme and "told" it to recognize and superimpose parts of the second image (a painting) inside the first.
AI didn't make even close result to the painting masterpiece, but it created an original digital art with unique thematic.
If you paint or draw, the deep dream and neural networks can be a very good tools for imagination and inspiration.
It depends on programming and complexity what kind of result you will get, also how well you manipulate and what do you want to achieve.
Deep dream apps and programs are available for most platforms, web, mobile and desktop devices.
Have fun!
Woahh this is super great❤