AI Art Generation Without a GPU
From Deadeagle63
Somethings of Note
It is worth noting most of these platforms operate on a credit system where each day you get a certain amount of credits you can use to generate images, or upscale your previously generated images and the links are in the TLDR section :)
NightCafe
Most likely the most known platform when it comes to generating "free" AI art pieces. It works on a credit system and you get 5 free credits per day.
To get started simply choose a model to start creating something special, in this example I will choose Stable.
Now you can play with settings how you want but for the most part, K_LMS at the moment has better output until PLMS/DDIM is added.
For the image, I decided to do a landscape in the desert with the prompt
a fantasy landscape in a desert environment featuring magical plants, cinematic, cinematic lighting, illustration, digital painting, highly detailed, hyper realistic
StarryAI
StarryAI is basically the exact same as NightCafe where you have daily credits to spend on creating image(s).
For the prompt I chose argo method and the prompt
a fantasy landscape in a desert environment featuring magical plants, cinematic, cinematic lighting, illustration, digital painting, highly detailed, hyper realistic
# ArtBreeder
ArtBreeder is an interesting one, you can generate as many creations as you want - but if you want to save them to your account you only get a certain number of things you can save per month. But artbreeder is an amazing tool.
To keep with the landscape theme I will generate a landscape in artbreeder by choosing landscapes which share fantasy themes.
Images used:
Dream by Wombo
Dream doesn't have any limits for generating the art to my knowledge and even allows you mint or print the generation if you give Wombo recognition as per their T&C's
The prompt
a fantasy landscape in a desert environment featuring magical plants, cinematic, cinematic lighting, illustration, digital painting, highly detailed, hyper realistic
FAQ?
What is AI Art?
Well simply put, it is when a user interacts with a data-model to produce an image, typically through a text prompt or image prompt
What is a Data-Model?
It is a massive collection of pre-trained models to help guide the AI towards a more suitable goal
What do you mean by pre-trained?
AI isn't magic that everyone thinks it is, for the most part it works like a child does with a Gi-Go (garbage-in garbage-out) principle. So people a whole lot smarter than me spend their time scraping for images on the internet and teach the AI what things are and how to construct it
How is AI taught?
By reinforcement, typically AI is graded on a 0.00 to a 1.00 scale, where 0.00 means it did poorly and 1.00 means it did a super fantastic job - so when it produces an image (guessess) the program rates its output to get better results
That sounds interesting, is there somewhere I can read about this?
Why yes ofcourse, if you fancy I read this is good article, and if you like watching someone smarter than yourself solve things Code Bullet has a fantastic channel dedicated to solving problems with AI
TLDR
Here are the links you beautiful person: NightCafe, StarryAI, ArtBreeder, Dream by Wombo
Deadeagle63
Congratulations @deadeagle63! You have completed the following achievement on the Hive blockchain and have been rewarded with new badge(s):
Your next target is to reach 300 upvotes.
You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word
STOPSupport the HiveBuzz project. Vote for our proposal!
Lovely post! I'll add Disco Diffusion to the list, you're using Tesla GPU on a google server, which means you don't need a powerful dedicated machine to create amazing generated art, or animation!