File size: 2,159 Bytes
2542dc3 3ecd160 aab15e2 8565c86 3ecd160 8565c86 3ecd160 6bd5c19 3ecd160 6bd5c19 2542dc3 3ecd160 8942e95 3ecd160 e0087f3 3ecd160 76dc433 38f3a80 3ecd160 0ba7b3b 3ecd160 8942e95 3ecd160 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 |
---
license: apache-2.0
base_model: stabilityai/stable-diffusion-xl-base-1.0
tags:
- stable-diffusion
- stable-diffusion-xl
- stable-diffusion-xl-diffusers
- stable-diffusion-diffusers
- text-to-image
- diffusers
- lora
widget:
- text: 4runner1, (quartering in view), (red:1.9) body, lifted, modified, topper off, ((chrome stock spoke wheels):1.9), blue sky, driving on the (beach:0.8), (ocean0.8), (sunset:0.8),8k, uhd, dslr, perfect focus (cfg = 4, seed = 410282563)
inference: true
language:
- en
---
# toyota-4runner-1st-gen-sdxl-lora
LoRA for SDXL 1.0 Base for generating 1st generation 4runners (1984-1989). 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:4runner1_SDXL_v1:0.8\>, then use the trigger keyword. Further example images with A1111 prompts at (https://civitai.com/models/147916/toyota-4runner-1st-gen-sdxl)
Example diffusers prompt which you can run in the inference to the right: '4runner1, (quartering in view), (red:1.9) body, lifted, modified, topper off, ((chrome stock spoke wheels):1.9), blue sky, driving on the (beach:0.8), (ocean0.8), (sunset:0.8),8k, uhd, dslr, perfect focus (cfg = 4, seed = 410282563)'
## Recommended Weight:
0.8 (lowering the LoRA weight in A1111 will make it hybridize with newer 4runners)
## Trigger:
4runner1
## Helper:
In general you can vary the color of components (body, topper, wheels, grill, bumper) with some fiddling. Also, lifted vehicles can be generated with lifted, modified. The following view tags were used in training although it might be hard to generate images with these specific views: 'quartering away view', 'quartering in view', 'front view', 'rear view', 'side view'.
## Notes:
The LoRA seems to work well with other base SDXL models, but I didn't spend much time playing with this.
## Methodology:
This LoRA was trained on only 1984-1985 images at 1024x1024. 92 images of vehicles and one modified vehicle. Two epochs with 1656 steps each. No regularization images were used.
|