LoRA for SDXL 1.0 Base for for generating images of the widely recognized Div0 tank... known as Div0 because $/quality results in a divide by zero error. People have generally given up on the tank, and are trying to find new uses for it. Hipsters generally avoid being seen in such a non-trendy ride. Common uses are demonstrated in the sample images. Can you discover more? The LoRA is in a safetensors format for use in diffusers or in UIs such as A1111.

In A1111, specify the LoRA in the prompt along with a weight <lora:vtnktnk_SDXL_v1_32-000015:1>, then use the trigger keyword. Further example images with A1111 prompts at (https://civitai.com/models/157883/div0-xl-tank-model)

Example diffusers prompt which you can run in the inference to the right: 'a (miniature) vtnktnk tank being used as a flower pot, dslr, 8k, beautiful, hyper realistic (cfg = 7.5, seed = 158618480)'

1.0 (lowering the LoRA weight in A1111 will produce other effects)

vtnktnk

In general you can generate a wide variety of tanks, characters, and scenes.

The LoRA seems to work well with other base SDXL models.

This model was trained on only images at 1024x1024. No regularization images were used. 15 epochs with 5880 overall steps.