• We have updated the guidelines regarding posting political content: please see the stickied thread on Website Issues.

C.O.T.

Devoted Cultist
Joined
Dec 29, 2020
Messages
157
Location
Spain
Hi, bring here some that sure that will make this wekend a lot more fun. It's about the new AI image generation software that has been released this last months. It consist on a different approaches of neural networks that have been trained to draw new images. They do it after the "brain" has seen millions of images from the web, so the system learns what is more likely to apear on a image.
The images are original and , as i think are owned by the person who press the generation button. So you can post or use them freely even with professional or comercial purposes. Anyway is a lots of fun and quite addictive.
The way it have to be used is by writing what you want to apear on the image, and the style you want it to look. That is called "the prompt" is very important on the style part to use names of well known artist or illustrators, to add names of websites that hold images of that artist helps a lot too. You can specify more the look by ki adding photography terminology, as close up, general view, macro, etc.
Nowadays it needs to have a very powerful computer in home, and some technical skills to build your system at home. But a company that have released it for free, has build a web service in which you can try it online for free, even from any tablet or phone. The page is called DREAM STUDIO. Uses the open source Stable Diffusion software.
Feel free to ask for advice on using it, or to post your works here. Here is the link:
https://beta.dreamstudio.ai/dream
Some works made from me:
Zombiesque Skatereske
j3ItAVHq_2x.jpg
 
AI can certainly make stuff look pretty authentic. It's 'spooky' art.
This will just mean that fine artists will stick to traditional media in order for art to have a real value in the art markets.
Art collectors may decide not to purchase art produced by electronic means.
 
There is a lot of talk about the future of art employments, the panorama isn't still clear enough to take positions. My experience using it it's that it needs a lot of human especialized work to get it finished. Maybe this is as the time when calculator machines arrived. Now we have more accountants, not less.
 
Someone who have tried the link could post it's work? it would be interesting.
 
Someone who have tried the link could post it's work? it would be interesting.

The DreamStudio link requires a registration and login to access Stability AI.
 
Registration is free, only give your mail and it will give you around 100 images to render, the software is free, but need a quite big hardware to run, the webpage service rents that big hardware so you can try it from any device, don't matter how weak it could be ( a phone for example ) cause the calculations are send to a big graphic cards and the answer is returned to your device.
To run it locally in your own computer, you would need to have a graphic card of at least 12 GB of video ram. ( 12 only in the graphic card, other more in the computer itself)
 
Other example, quite extrange...i used the prompt: " invocation of Gradius at the main square" ... Who is Gradius?...i dont know.
It apeared this image.
143508982_Invocation_of_gradius_at_the_central_square__by_Raphael_Lacoste__Dan_Kozachkov__matt...png

I'm sure that a similar scene has happened, but not in this planet and not in this time :D
 
Mods, please feel free to move/remove as you see fit.
I thought I'd have a go at employing A.I to make me some interesting images using DALL.E : https://labs.openai.com/
You are limited to a set amount of 'interactions' per month. I used my allowance up pretty quickly last night.
I'd urge you to have a try and post up your interesting results along with the key words and phrases you fed in to get there.

Here are a couple of my favourites :

Cubist rod Stewart playing a kazoo whilst rollerskating​

DALL·E 2023-03-26 14.39.54 - Cubist rod Stewart playing a kazoo whilst rollerskating .png


Photograph of Gibby Haynes brushing his teeth​

DALL·E 2023-03-26 14.23.23.png

Butthole Surfers frontman Gibby looking suitably odd here, true to the real life form.
Notice the brush goes through his neck and the reflection doesn't match the image before it. Excellent
 
That Gibby picture needs to have blood everywhere. I mean, it looks like he's gone through the artery.
 
'Gab' have their own AI now, called 'gabby', which does pretty much the same thing.
You feed it instructions and it churns out some images. Sometimes excellent. Often weird.
Let me try it with the Rod Stewart one you used above and see what it gives us.

Hmmm - the first one doesn't seem to get close, apart from that is deffo Rod, and the second one is better but it lacks cubism, and the hands are still odd.

1679840384982.png
1679840426734.png
 
'Gab' have their own AI now, called 'gabby', which does pretty much the same thing.
You feed it instructions and it churns out some images. Sometimes excellent. Often weird.
Let me try it with the Rod Stewart one you used above and see what it gives us.

Hmmm - the first one doesn't seem to get close, apart from that is deffo Rod, and the second one is better but it lacks cubism, and the hands are still odd.

View attachment 64721
Why is Rod holding a pregnancy testing kit/covid test?

And why does the other picture show Howard Jones?
 
This was done yesterday.

"Will Smith Eating Spaghetti"

Done by a Reddit user named chaindrop, using Modelscope Text2Video generator.
A machine learning model that turns text inputs into short video clips.


 
Midjourney gives better quick results, but is a payed app, StableDifussion is free and opensource, but you need to invest some bucks on a good graphic card. The opensource nature of Stable Difussion has allowed people to hack the program and modify it. One of the main tendences has been the ability to train the neural network with some pics of someone. For example yourself, or the very posted ones of fashion Pope Francis and the detention of Donald Trump.
 
I'm a fan of philosophical horror writer Thomas Ligotti.
His story "The strange designs of Master Rignolo" is one of my favorites.
I've tried to illustrate the story using OpenAI. Some of the pictures are spot on, others less so. I've discarded tons of pictures.
I guess this cost me 5 dollars in total. I'm satisfied with the result and may try other stories.

This is a series of ten blog posts:
You can read, and see, the whole story here:
Part 1 Part 2 Part 3 Part 4 Part 5, Part 6, Part 7, Part 8, Part 9, Part 10

1685347634268.png
1685347668310.png
 
Maybe would be interesting if members of the forum would do the same with the cases they write about, the forum would became a collection of illustrated narrations about forteana and mistery, would by really fun :D
 
I’m not kidding.
We have GPT at work and after the whole Snow White thing, I asked it to do a synopsis of Snow White set In space.
It started with Nova White escaping from her luxury planet ruled by her evil queen stepmother and crashing on a mining planet where she met a bunch of mining robots whose sole purpose in life was to endlessly mine jewels to satisfy Evil Queen‘s passing interest in shiny objects of which she held no value.
EQueen gets jealous and on the advice of her hologram ‘other’ sends a bounty hunting assassination droid on a mission to kill her. Nova White bests the assassination droid and persuades it to give up the chase.
It all ends up with visits to 7 different planets where each of her Seven droids find a way to convince each of the ruling councils of each planet to overthrow the wicked queen and at the end, a massive space battle ensues.
Nova White’s ship is critically damaged and they all flee into an escape pod that crashes into EQueen’s palace where the droids and Nova, each using their unique abilities make their way to the throne room where White defeats EQueen.

I could have asked it to flesh out for dialogue but I thought that was enough.
 
Last edited:
New York Times sues Microsoft and OpenAI for 'billions'
US news organisation the New York Times is suing ChatGPT-owner OpenAI over claims its copyright was infringed to train the system. The lawsuit, which also names Microsoft as a defendant, says the firms should be held responsible for "billions of dollars" in damages.
ChatGPT and other large language models (LLMs) "learn" by analysing a massive amount of data often sourced online.

The lawsuit claims "millions" of articles published by the New York Times were used without its permission to make ChatGPT smarter, and claims the tool is now competing with the newspaper as a trustworthy information source.

It alleges that when asked about current events, ChatGPT will sometimes generate "verbatim excerpts" from New York Times articles, which cannot be accessed without paying for a subscription.
https://www.bbc.co.uk/news/technology-67826601.amp
 
I asked for a Victorian ghost. She doesn’t look very ghost like. Not sure what the thing between her and the mirror. She doesn’t seem to have a reflection so it may have confused vampire with ghost.
IMG_3795.jpeg
 
Back
Top