
I f****d it up. Completely. Instead of cropping the training pics to 512 x 512, I threw this mess into the training device, here on CivitAI. That's why there are a some nice bars at the edge of some pictures.
What's this LoRa good for at all ?
It will generate cars in combination with realistic Checkpoint models, that can be generated with tags like
fun_car , vw_t1 fun_car, 19XX fun_car( time period of your choice ),
and you can also try tags like
belvedere65, fordg13, fordp2, fordp3, fordp7a, fordp7b, fordtc, fordcortina,
mercedes_w124, opelkap54, opelkappl, opelkad64, toycrown69, toycrown71
It's not impossible that you'll get something generated, that looks a little bit like an Opel Kapitän, an old Ford or an old Toyota Crown....if you pray hard.
It's useful to type cool_cars within the prompt.
I've uploaded some example pics, which show the prompts, settings and checkpoint models i've tested so far.
描述:
训练词语:
名称: Cool_Cars.safetensors
大小 (KB): 36978
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success