I need to read up on this technology. I have so many questions. Is it using the internet for it’s memory or choice of images and life situations? Holy smoke…. It’s blowing my mind just thinking about it.
Thanks Loki!!! I know what I want for Christmas now!; an AI Art machine! haaaaa
Removing cannabis since I’ve been getting violations. I’ll make another attempt later. Here’s
spaceman snowboarding down mountain of tomatoes, digital art
What about machines enslaving/killing humans.
Maybe we shouldn’t give it ideas
So cool. I was hoping it would look like BOG alien head, lol. Damn. I guess it would also stimulate our own imaginations, just playing around with it.
SO COOL!
Narnia spilling from a glass decanter.
Oh, @Northern_Loki, I meant to say the key words that could have triggered the censorship, could have been NYMPHS and GROIN. ??
It utilizes trained neural convolutional networks. Simplified … what that means is, a digital neural ‘brain’ is fed lots of images that are tagged. The network consists of layers (of neurons) in which each layer may have been trained for a specific feature (might be selected based on the tags). For instance, one layer may consist of training by showing it a bunch of dog ears. The next layer may be trained to recognize a dogs head. It will utilize each layer depending on what is input, some layers utilized while others not. The whole of the output will be a probability that it matches what is being input. For a dog, it will be utilizing the ‘ears’ layer, the ‘head’ layer, and other dog-part layers to recognize a dog with a certain probability, for instance. The art generation starts with noise and interactively makes adjustments, feeds it into the recognition (I presume), until the probability output is relatively high.
There are papers on this subject / technology / model. Take a look at ‘deep-dream’ for some insight into this layer concept.
An Nvidia graphics card has the capability of creating such networks. The Dall-E is probably lots of those cards.
I’ve modified this to: Humans enslaved by machines, digital art
Here’s Humans enslaved by machines, photo:
Fascinating.
Just wow. Some of the images are remarkably outstanding while utilizing art fundamentals. Like the placement of things and color utilization…. composition I guess. I’m fascinated by the outcomes.
I was sort of into art and went back to college at 40 to try to get into it as well as photography. I was in the top of my classes. After a few years, life got all real on me and I never finished it.
I would never get anything done if I had total access to the hardware that makes it happen, lol.
The other side to it is the hit it will have on certain artists like illustrators/artists for books and corporate installations, etc. It looks to me like you could easily come up with an illustration that would work for you in mere minutes rather than all of the time it would take to find an artist to commission work that would be specific to a project.
You came up with more than a dozen awesome images that would work for somebody right away. I can only imagine it is not now, or may never be, affordable for most people.
Are there lots of systems already in place or is it still too new?
Amazing!
Yes, it’s crazy. Lighting, focal points, bokeh, style, impressionism. Any of those can be adjusted by tuning the input phrasing to get closer to what you’re imagining (if not distracted by what it creates).
This is relatively new. At least at the quality it is now producing and it’s ability to make inferences from natural language input. But, the pace is rapid (the scary part may be within our mortal time horizon).
A desert landscape in the style of dr seuss if possible @Northern_Loki
Way cool you should flip it and do machines enslaved by humans and see what it does these are so interesting
It hasn’t been fed any Cat and the Hat books, lol.
