monster-cars-sdxl-lora

LoRA for SDXL 1.0 Base for generating Monster + Car images. The LoRA is in a safetensors format for use in diffusers or in UIs such as A1111.

How to use

In A1111, specify the LoRA in the prompt along with a weight <lora:M0nstrcrz_SDXL_v1_32:1>, then use the trigger keyword. Further example images with A1111 prompts at (https://civitai.com/models/152956/monster-cars-sdxl)

Example diffusers prompt which you can run in the inference to the right: 'm0nstrcrz, driving, yellow car, Frankenstein, zombies, bats, red moon, fog, purple clouds (cfg = 8, seed = 3972587558)'

Recommended Weight:

1.0 (lowering the LoRA weight in A1111 may produce less monstery results)

Trigger:

m0nstrcrz

Helper:

In general you can throw in characters and cars in different scenes.

Notes:

The LoRA seems to work well with other base SDXL models, but I didn't spend much time playing with this. If producing images at 512x512, this model is known as Abstractio 3000, i.e., the images are so awful that it's hilarious.

Methodology:

This model was trained on onlyimages at 1024x1024. No regularization images were used. 30 epochs with 14580 overall steps.