New: 1.1 LoRA that adds additional material and the "startling stories" trigger.
This model uses a trigger: sd-pulp, along with other optional triggers described below.
Pulp is a high quality model trained on Public Domain, CC0 licensed images from pulp magazine covers of the 30s and 40s with titles like Startling Mystery, Dime Detective, Horror Stories, and Uncanny Tales. It excels at producing images with multiple characters, shocked expressions and action but can be coaxed into applying its visual styles to a wide range of subjects. You can swap or combine various magazine titles in your prompts to adjust color palettes and tone; the full set of titles for use in prompts (all lowercase):
dime mystery
horror stories
startling mystery
terror tales
thrilling mystery
uncanny tales
vice squad detective
spicy adventure
dime detective
detective short stories
startling stories (v1.1 Lora only)
The source material sometimes lacks detail, but this can usually be corrected with negative prompts.
Useful negative prompts: words, text, titles, signature, watermark, wordmark, magazine cover
I post about Stable Diffusion and model making at diffusionary.substack.com
Description
This is a LoRA with some new added material and a new trigger, "startling stories." In my testing it seems to work very well with numerous models.
FAQ
Comments (9)
just wanted to throw out that I much prefer working with checkpoints over loras and appreciate you uploading both. Feel like there's always the "this should be a LORA" comment, and I don't want those people to run off everyone from uploading these as well.
No worries, thank you for the feedback! I think both are useful for different types of creative scenarios and I'm happy to provide both options.
Thank you. Great one!
This really makes me wish SD had any idea what it was doing when it comes to firearms. What kind of self-respecting noir detective leaves home without a snub nosed revolver?
Totally agree; I'm working on a new model right now and especially for anything illustrated the guns are very weird
There are thousands of models of guns in 3d you can render in Blender free and paste into the "hands" of your image. If you want your snub nose or whatever else, you can do it with minimal effort, just not at the click of one button. Prove those that are upset about AI art wrong and that it is not all low effort "art" and take the time to make your own art creation instead of relying on the algorithms to do all of the work. I mean this comment in the best way possible.
Hey, nice model! I've become so inspired by it that I want to train my own model based on public domain "funny animals" comics, and I might actually do it someday. Keep up the good work! :D
Looks like the LoRA is incorrectly tagged as a CheckPoint Merge?
Fixed thanks; LoRA's weren't an option when I uploaded it.