Running locally (especially if you choose ComfyUI, the most advanced and flexible of the interfaces) provides some advantages compared to creating images and videos with an online service:
On the other hand, the online services can use models that are too big and complex to run on local hardware, online services don't require that you own any special graphics card, and the learning curve is generally easier with online services. If you're not sure, you could always start out by picking an online image generation service and then come back and explore local generation options when you want to learn more skills.
If you've already got a PC with a suitable graphics card, then choosing an open-source interface to download is completely free. If your computer couldn't handle local generations, then the other free option is to use an online image generator. Most of the leading online services let you generate images for free, and what they offer for free is easily good enough for most casual users.
Some online generation options do filter your prompts to avoid generating NSFW (Not Safe For Work) images. But others don't filter your prompts at all, so the content you create is up to you. Here's a ranked list of uncensored image generation options.
No, you don't need to be a programmer to use Stable Diffusion. For many non-programmers, you may find that installing Stable Diffusion locally on your own computer is the first time that you have to pull files from GitHub, or the first time that you've had to type a few commands in a shell window on your PC. But even if the installation isn't a simple one-click process for some of the choices, there is plenty of help available. Reviews of Stable Diffusion interfaces include not just a download link, but also a link to a tutorial video where available that will walk you through the installation.
If an AI does all of the work in creating an image, then under US law you can't claim copyright over the image that the AI made. Only work authored by a human is eligible for copyright. Here's an article about it.
If you're doing a substantial amount of work yourself, then even if AI generation contributes to your project, your work is still yours. Many people use AI generation to extend a background, change some things here or there, or create a few images that appear within a larger work, but they are still the authors of the larger work. (The exact amount of work you have to do here is a legal gray area that could use more clarity.)
Some leading generative AI models, like OpenAI's DALL-E 3 and Adobe Firefly, are trained on image libraries legally licensed from stock photography companies. Contributing photographers were paid for the use of their work in training the AI. This is in combination with using public domain art and photography, which anyone is legally allowed to use.
For some other models, there's a legal gray area around their training, with a lawsuit still moving through the court system.
Copyright © 2024-2025 by Jeremy Birn
Welcome to the Internet! Websites use cookies to analyze traffic and optimize performance. By accepting cookies, data will be aggregated with all other user data.