Lada Granta FL (Sedan) — Experimental Beta

Made this LoRA mostly out of boredom and as an experiment to learn how to train models on static objects and vehicles instead of just character faces. Decided to test the workflow on the facelifted Lada Granta FL sedan.

This is an early beta release. I used this project to figure out dataset cleanup, body geometry preservation, and proper trigger handling. The result turned out surprisingly solid: the car is instantly recognizable, the signature front "X-face" trim is intact, and the rear tail lights along with body proportions hold up well. It handles color changes, different camera angles, and weather conditions smoothly.

Interior details might not be 100% accurate, but it works great for street photography, highway shots, rain, gas stations, and grounded realistic scenes.

Generation Settings:- Trigger word: It works, as practice has shown, both without a trigger and with lore for the characters.

ДЛЯ Qwen 2.1 РЕКОМЕНДУЮ ИСПОЛЬЗОВАТЬ МОДЕЛЬ https://civarchive.com/models/452459/fineview-or-story2?modelVersionId=3370753

- LoRA Weight: 1.0

- Tips: For best results, specify the car color (white color, black paint) and camera angle (front three-quarter view, side profile).

Example Prompts:

1. Clean Daylight / Front 3/4:

granta_fl sedan, white color, front three-quarter view, parked in an open concrete parking lot, bright daylight, clear sky, realistic reflections on car hood, factory silver alloy wheels, clear headlights, sharp details, candid shot, 35mm photography

2. Moody Wet Street / Night:

granta_fl sedan, glossy black paint, low angle front three-quarter view, parked on a wet city street at night, neon reflections on wet asphalt and car hood, rain droplets, soft cinematic lighting, sharp focus on front bumper and grille, photorealistic

Feel free to post your generations in the gallery, curious to see what you guys create with it!