metadata
license: apache-2.0
language:
- en
tags:
- text-to-image
- lora
- aircraft
- aviation
datasets:
- nyuuzyou/aircraft-images
base_model:
- black-forest-labs/FLUX.1-schnell
library_name: diffusers
pipeline_tag: text-to-image
widget:
- text: >-
2015 American Airlines Airbus A320-214, registration EC-LVC, landing
against a clear blue sky
output:
url: samples/1.jpg
- text: >-
2022 Lufthansa Airbus A321-271NX, registration D-AIEQ, is airborne over a
forested area
output:
url: samples/2.jpg
- text: >-
2018 Azimuth Airlines Sukhoi Superjet 100-95LR, registration RA-89094,
taxiing on the runway near some greenery and airport infrastructure
output:
url: samples/3.jpg
- text: >-
2008 Lufthansa Regional Bombardier CRJ-900LR (CL-600-2D24), registration
D-ACKA, airborne against a clear blue sky
output:
url: samples/4.jpg
- text: >-
2017 Air India Boeing 787-8 Dreamliner, registration VT-ANW, airborne
against a cloudy sky
output:
url: samples/5.jpg
A LoRA trained on 165K high-resolution aircraft images and their structured captions, fine-tuning FLUX.1 [schnell] for enhanced aircraft image generation. This adaptation specializes in:
- Accurate generation of commercial aircraft with correct proportions and details
- Precise rendering of specific aircraft models (Boeing, Airbus, etc.)
- Realistic airline liveries and sometimes registration markings
Best used for generating photorealistic aircraft imagery from detailed prompts. Maintains FLUX.1's general image generation capabilities while significantly improving aviation-specific output quality and accuracy.
Training data: 165K aircraft images with structured captions
Recommended prompt: Follow caption format:
[year] [airline] [manufacturer] [type], registration [number], [state] [surrounding objects]
Example: 2023 Lufthansa Airbus A321-271NX, registration D-AIEQ, taking off against a clear sky