Learning about Artificial Intelligence Illustration
Posted: Thu Oct 06, 2022 9:03 am
What is AI Illustration?
Learning about AI Illustration is like
Artificial Intelligence assisted Illustration. It's something that has become suddenly mainstream in a matter of months and it's partially reason for the rise of NFTs. In fact if you ever wanted to know what was responsible for the Bored Ape Yacht Club project? AI had a lot to do with it.
The technology is simple: you have artificial intelligence to be programmed to "learn" samples of data (illustrations of varying styles or of an array of styles) and you effectively have it sit until it's done doing its thing. From there, you have an interface of which you can command the AI program to generate new outputs of graphical fidelity based on what it has learned with every output being completely different from the last. This is a bit like saying using the function of Pi to generate a simple new, random hash number but it's graphics - and cool ones to boot!
But not all art that you see is solely originally from the AI as there are instances where it's part original work, AI processing, editing, more AI processing, and then a refined product that's be touched up with something like Photoshop. The key thing is having a mastery of the particular wording used (the prompt) that is fed to AI to then output the initial illustration - some people share their prompts and others keep them for themselves for their particular projects and commercial art enterprises like NFT collections.
Where to go and what to read?
There are three major brands of AI generating engines: Dall-E, Stable Diffusion and Midjourney. They all work basically the same (text-to-draw) and some allowing for particular refinements. The following are a sectioning of resources for each of them respectively but don't try to go over tutorials too closely - it's all about experimentation right now.
Dall-E
Stable Diffusion
Midjourney
Learning about AI Illustration is like
Artificial Intelligence assisted Illustration. It's something that has become suddenly mainstream in a matter of months and it's partially reason for the rise of NFTs. In fact if you ever wanted to know what was responsible for the Bored Ape Yacht Club project? AI had a lot to do with it.
The technology is simple: you have artificial intelligence to be programmed to "learn" samples of data (illustrations of varying styles or of an array of styles) and you effectively have it sit until it's done doing its thing. From there, you have an interface of which you can command the AI program to generate new outputs of graphical fidelity based on what it has learned with every output being completely different from the last. This is a bit like saying using the function of Pi to generate a simple new, random hash number but it's graphics - and cool ones to boot!
But not all art that you see is solely originally from the AI as there are instances where it's part original work, AI processing, editing, more AI processing, and then a refined product that's be touched up with something like Photoshop. The key thing is having a mastery of the particular wording used (the prompt) that is fed to AI to then output the initial illustration - some people share their prompts and others keep them for themselves for their particular projects and commercial art enterprises like NFT collections.
Where to go and what to read?
There are three major brands of AI generating engines: Dall-E, Stable Diffusion and Midjourney. They all work basically the same (text-to-draw) and some allowing for particular refinements. The following are a sectioning of resources for each of them respectively but don't try to go over tutorials too closely - it's all about experimentation right now.
Dall-E
Stable Diffusion
Midjourney