It does pictures of rally style Toyota Celica GT-Four ST205 in classic Castrol livery.
The trigger trained word toyota_celica_gt-four_1205 isn't a typo, 1205 avoids 205 which is a different car in epicRealismV5.
The training was made in 1024x768 resolution, so expect to perform better at max 1024 horizontal. 2 types of training: Adafactor and Prodigy. You can try the following orientations, but it lacks consistency
toyota_celica_gt-four_1205 car from front
toyota_celica_gt-four_1205 car from front left view
toyota_celica_gt-four_1205 car from front right view
toyota_celica_gt-four_1205 car from left rear view
toyota_celica_gt-four_1205 car from left side view
toyota_celica_gt-four_1205 car from rear
toyota_celica_gt-four_1205 car from right rear view
toyota_celica_gt-four_1205 car from right side top view
toyota_celica_gt-four_1205 car from right side view
toyota_celica_gt-four_1205 car three-quarter left front view
toyota_celica_gt-four_1205 car three-quarter left rear view
toyota_celica_gt-four_1205 car three-quarter right front view
toyota_celica_gt-four_1205 car three-quarter right rear viewWorkflow:
Old V1:
A lora aimed to understand the classic castrol rally painting scheme used in rally, Trained on civitai using "Concept" option.
List of possible text that I would like to work on any surface to be painted as if its main inspiration is the same scheme from this rally car.
Castrol paint scheme
Castrol themed liveries
Castrol sponsor
Castrol Rally scheme
classic Castrol Rally Scheme
Castrol Racing Design
Castrol Motorsport Livery
Castrol Branding
Castrol Racing Colors
Castrol Racing Graphics
Castrol Racing Aesthetic
Description
Epoch 3 of 3
Files: 113
Steps: 4.011
Model: epicRealism v5
{
"lr": 0.0005,
"engine": "ai-toolkit",
"epochs": 3,
"ecosystem": "sd1",
"keepTokens": 0,
"networkDim": 32,
"numRepeats": 71,
"resolution": 1024,
"lrScheduler": "cosine",
"minSnrGamma": 5,
"noiseOffset": 0.1,
"networkAlpha": 32,
"optimizerType": "adafactor",
"shuffleTokens": false,
"textEncoderLr": 0.00005,
"flipAugmentation": false,
"trainTextEncoder": true
}