mNo edit summary |
mNo edit summary |
||
Line 4: | Line 4: | ||
A new type of AI called diffusion models can generate images. They learn concepts by observing billions of images of ''all'' the things, then recreating them. While you can have the models generate images entirely for you with text prompting, there are many other means of tightly controlling their output. Under the hood, they're the same type of model as large language generators such as ChatGPT. | A new type of AI called diffusion models can generate images. They learn concepts by observing billions of images of ''all'' the things, then recreating them. While you can have the models generate images entirely for you with text prompting, there are many other means of tightly controlling their output. Under the hood, they're the same type of model as large language generators such as ChatGPT. | ||
=== '''Is AI art theft as some say?''' === | === '''Is AI art theft as some say?''' === | ||
No, not metaphorically, not literally. The difference is sharp and easy to see when the distinction is drawn between the reality and the misinformation: They not only don't deprive the original creator of their works, they don't store any portion of those works in any way. They can't be used to recreate works they learned from. | No, not metaphorically, not literally. The difference is sharp and easy to see when the distinction is drawn between the reality and the misinformation: They not only don't deprive the original creator of their works, they don't store any portion of those works in any way. They can't be used to recreate works they learned from, outside of maliciously crafting them to do so, or works that are short and heavily used in thousands of times over, like the phrase "in God we trust" or the lyrics of Happy Birthday. | ||
It should be noted that there is a ''massive'' amount of toxic misinformation directed at imagery making use of AI tools. More on this below. | It should be noted that there is a ''massive'' amount of toxic misinformation directed at imagery making use of AI tools. More on this below. |
Revision as of 20:52, 10 December 2024
Frequently asked questions
What are AI generated images?
A new type of AI called diffusion models can generate images. They learn concepts by observing billions of images of all the things, then recreating them. While you can have the models generate images entirely for you with text prompting, there are many other means of tightly controlling their output. Under the hood, they're the same type of model as large language generators such as ChatGPT.
Is AI art theft as some say?
No, not metaphorically, not literally. The difference is sharp and easy to see when the distinction is drawn between the reality and the misinformation: They not only don't deprive the original creator of their works, they don't store any portion of those works in any way. They can't be used to recreate works they learned from, outside of maliciously crafting them to do so, or works that are short and heavily used in thousands of times over, like the phrase "in God we trust" or the lyrics of Happy Birthday.
It should be noted that there is a massive amount of toxic misinformation directed at imagery making use of AI tools. More on this below.
For the curious, here's a good introductory video by respected mathematics educator Grant Sanderson of 3blue1brown going over exactly how they work. Lengthy deep-dives into the technology are also available on his channel.
Why is there such backlash against it?
The companies that first created them had the gall to suggest replacing artists with them. Given that they can only produce things of value in the hands of artists, and are tools for artists rather than replacements, marketing diffusion models as artist replacements is about the shittiest kind of marketing they could have conceived of. Unfortunately, anger about that gets misdirected at people making legitimate use of these tools as part of an actual creative process, and they feel justified in using any tactics to fight against it, including flat out fabricating misinformation and acting as toxic as they feel justified in acting.
If these companies had marketed diffusion models as tools for artists to begin with, things would be somewhat different, but scummy companies will do scummy things.
Do you support these companies?
No, none of them get a red cent from me. I use a free model run locally on a PC with a beefy GPU. I fine-tuned it myself, developing a style based on an aesthetic I wanted that no one was doing how I envisioned it. I'm not releasing any of my own tunings, nor techniques I developed to bring things to life.
Is it artistic?
There are various use cases, creating a spectrum. Text prompting is obviously less artistic than manually editing an image afterward, or manually creating a 3D scene in Blender and using a diffusion model for the final render. My personal workflow involves the latter.
When are you going to finish this FAQ?
Probably in the next few days.
Would you like to know more?