Enquiring Minds
AI Galleries

AI Galleries

AI Image Galleries

Over the last couple of years, I’ve found myself increasingly interested in AI image generation. Partly, I suspect, because of the Aphantasia, and partly because (perhaps as a result) I cannot draw at all, this democratisation of art is pretty amazing to me, and thanks to the magic of open source software, I’m able to run a local instance of Stable Diffusion to create anything that springs to mind.

One of my favourite things to do is so-called “raw prompting” in which no instructions are provided regarding what the image should contain, or what form the style should take etc. Only the raw text of the prompt and the settings are specified, and then you let the model make its own “decisions” about everything else.

All these galleries contain images created in this way. Images are cherry-picked, but not edited at all, and no in-painting (except Adetailer) or any kind of post-processing has been done. The prompt can be seen in the gallery intro text, the full parameters of each image (model, sampling method, etc. etc.*) should be able to be read from any image, if you know how to do it.

(Also, I realise some people may find some of the images disturbing, (such is the nature of art, including AI art), (hell, I find some of them disturbing), so fair warning etc. etc. I’ve tried to select only images which (in my opinion) speak to the nature of the source material, but some of that can be pretty dark, and I make no apologies for my taste in poetry, music, etc. )


*Boring Technical Stuff

Most of these are 1024×576, 50 steps, CFG 5, random seed and random full strength variation seed, no negatives, no Adetailer prompt, and a mix of SD1.5 and SDXL. I typically ran these as batches of 5 against 20 or 30 models each. (Thanks Agent Scheduler.) (No LoRa’s etc.) Model, seed, etc. can be extracted with PNGInfo if you want it. All rights to images only are reversed, (if you like it enough to take it, credit would be nice, but whatever :D ). Thanks to the creators of the various models used. Apologies to the authors of all prompt texts, thanks, I love your work. ;)