I’m really very impressed, and surprised, at how easy it is to install GPT models on a local machine with GPT4All, plus the bonus of Stable Diffusion
After not having any success with setting up a GPT chat bot on a Raspberry PI 3, and even found I couldn’t do it on an old Lenovo T510, I decided to try on my Linux box, and was blown away by the options of models available, the only set back is the size of the more advanced models … will probably clear a SSD fo a rarely used Linux OS to make room to play more ::grin::
Sure I’ll post more on these models in the future, as I understand them more, but today I’m playing with Stable Diffusion:
Stable Diffusion is an open-source text-to-image model that allows users to generate images based on a given prompt or description. It uses a type of AI called diffusion models, which are designed to learn and mimic the patterns found in data.
In this case, Stable Diffusion has been trained on a massive dataset of images from various sources (e.g., internet archives) and can be fine-tuned for specific tasks like generating artwork based on text prompts. This means that users can input a description or prompt into the model, and it will generate an image that attempts to match their request.
Stable Diffusion has gained popularity among artists, designers, writers, and hobbyists due to its ability to produce unique images quickly and efficiently. However, like any AI tool, it’s not perfect and may require some tweaking or post-processing to achieve the desired results.
[as described by GPT4All Llama 3.1]
So today I’m asking Llama 3.1 to describe William Gibson Sprawl book covers, then with that text ask Stable Diffusion to create images… all on my local machine!
First off my favourite, and the iconic original
Neuromancer
The original 1984 UK edition cover for William Gibson’s novel “Neuromancer”, as well as some other early editions, features an image of a woman with her face obscured by a computer-generated grid or matrix. This iconic artwork is actually the work of Michael Whelan.
In this depiction, the female figure appears to be surrounded by lines and patterns that evoke a sense of digital code or circuitry. The cover art has become synonymous with Gibson’s novel and its exploration of virtual reality, artificial intelligence, and the intersection of technology and humanity in 1980s cyberpunk culture.
Count Zero
The original 1986 UK edition cover art for William Gibson’s novel “Count Zero” features a striking and somewhat unsettling illustration by Michael Whelan. On this iconic cover, there is an image of a figure with their face distorted or obscured in some way. The main focus of the artwork appears to be on the character’s hands, which are shown as being heavily tattooed or scarred, possibly indicating a connection to the novel’s themes of virtual reality and cybernetic enhancements. The overall design has a dark, gritty feel that is characteristic of much 1980s science fiction art.
Mona Lisa Overdrive
The original 1988 UK edition cover art for William Gibson’s novel “Mona Lisa Overdrive” features a striking image by Michael Whelan.
On this iconic cover, there is an illustration of Molly Millions (the main character) with her face partially obscured or distorted in some way. Her eyes are visible but the rest of her face appears to be reflected and fragmented across multiple surfaces, possibly indicating a connection to the novel’s themes of virtual reality and identity fragmentation.
The overall design has a dark, gritty feel that is characteristic of much 1980s science fiction art.
As someone with zero artistic ability I find these images amazing, this is going to open the floodgates of my imagination, but raises the question of the huge volumes of AI generated content that will be flooding the internet … & I understand the irony of that statement ;p
