This is a discussion we have been having under the hood for over a year and it is about time to start writing articles on it. I had the privilege of following all the advances in sensorial AI and generative AI art in particular for the last year and half. Was actually part of one of the first AI communities https://twitter.com/neuralismai
As things go though most of the experienced folks ended up working with stability or midjourney. But enough of that, lets talk about what has been happening.
There has been an explosion of AI art, videos, chatbots. Now suddenly there is 100s of 100s of apps and subscriptions (most of which I donot recommend, are just using stable diffusion or openai under the hood) and lot of paranoia, let us decode and break it down for the community.
Here is my strategy after lot of research :
1) Identify new cutting edge apps which are free to try (usually as discord bot, plugin or website)
2)They all get paid eventually and start milking your addiction, so do not wait for that, you have friends who can help you figure out the underlying code and tools and you can run them locally or in google colab.
3) Running things locally and streamlining easy ways to find relevant info in the increasingly insane maze of AI research and innovation. With that said keep a few paid apps handy for when there are commercial projects to be done.
SO what are the most useful AI tools available for a non-coder. We'll cover free as well as paid ones.
Free (for now)
Opera One browser (cannot be recommended enough, fully integrated with AI and web3, very modular workflow)
Pika Labs (text/image to video) https://www.pika.art
Leia Pix (image to 3d depth) https://convert.leiapix.com
Leonardo AI (stable diffusion api, paid and free versions, arguably one of the best. Can make images as well as 3d textures or models, train your own models) https://app.leonardo.ai
ChatGPT (still a great one, paid one only recommended for commercial projects, these subscriptions tend to add up from experience) https://chat.openai.com
Perplexity AI www.perplexity.ai (chatbot specialized for non-coding needs, works better in those situations than chatGPT)
Stability AI (the discord server has free usage of the AI bots from time to time, last 2 weeks for example I'm making cutting edge stuff with SDXL the new stable diffusion model)
Skybox (Blockade labs) : (text to 360 images/skyboxes. There is also a Unity plugin. As of this week, they introduced premium tier. I made 100s of 360 images last 2 months but now you can only make 15 per month. I'll break down the tech involved in another article) https://www.blockadelabs.com
Also worth mentioning here are OpenAI and RunwayML, for language models and video editing respectively, but their free features are rather limited.
Midjourney : Much as I donot like their closed source model, there is a certain uniqueness to midjourney (been there from free private beta times and it has really become quite powerful), worth learning and best used for making assets to be further used with other apps/utilities. Can quickly drain you, with private mode it is a hefty 60$ per month, only recommended for commercial usage. The free usage is too limited to be worth mentioning for me. midjourney.com
GitHub Copilot : Amazing for coders, they started charging subscription last year, however very useful for developers. Fully integrates with Visual Studio and VS code. https://github.com/features/copilot
Kaiber AI : the best one for making code free AI video compositions and VJ loops. kaiber.ai/create
(I'm sticking to these for now as there is simply too many out there, if there is anything particular you want mentioned do add in comments.)
This is it for now, I'm writing a second part to this article about the underlying codes and useful repos, how to run them locally or on a rented server. In the long run, as a community of professionals with lot of content and data we need to take this tech seriously, understanding all the ethics and safeguard ourselves by building our own models (more on that in future).