XonarLabs/OLOID_Framework

OLOID Framework

Prompt

xonar_ext_SUV_rough, A professional studio photography of a SUV, silver metallic gloss paint, 8k resolution, cinematic lighting, dark grey studio background.

Prompt

xonar_int_concept, Photograph of a futuristic car interior with a sleek, minimalist design. The dashboard features a red and black color scheme with a white steering wheel and a central console. The seats are upholstered in light gray fabric with red accents. The background shows a sandy, desert-like terrain.

Prompt

xonar_ext_odd, A professional studio photography of a concept car, silver metallic gloss paint, 8k resolution, cinematic lighting, dark grey studio background.

English | 中文

About OLOID

OLOID Framework is an open-source LoRA framework for automotive design generation.

Built on Flux.1 dev, it is designed for agent retrieval and invocation, aiming to hand over the tedious parameter-tuning process in design gacha (LoRA stacking + seed sampling) workflows to agents and free up designers’ time.
It also supports manual combination and use by designers.

Framework Structure

The current release includes 49 LoRAs, organized across four dimensions and designed to be combined based on different design goals.
For example: PRD (define volume) + CRT (add form language) + INS (expand creative possibilities) + VIS (control visual output)

Dimension	Code	Role
Production	PRD	Production-oriented volume, proportion, and structural foundation
Creative	CRT	Advances form language and design variation
Inspiration	INS	Introduces cross-domain design language and additional stimulus
Visuals	VIS	Controls image style, material feel, and overall visual impression

A full structural overview is shown below: Figure: Full structure across PRD / CRT / INS / VIS and EXT / INT / PART.

Index

For each LoRA’s trigger words, reference prompts, and recommended weights, please refer to the index:

• Huggingface
• Feishu Doc

Download

• Download them in the Files & versions tab.

Project Status

The current phase of OLOID on Flux.1 has been completed. We have already moved on to exploring next-generation model training and a more complete agent-oriented workflow, so this generation will no longer be updated.
Feedback and testing are welcome.

For discussions around AI generation systems for automotive design, please contact OkeyDesign: hello@okeydesign.com
For custom training, agent-related and other technical topics, please contact XonarLabs: info@xonarlabs.com

License

This model is built on FLUX.1 dev.
Usage may be subject to the base model's license restrictions.

Credits

OkeyDesign & XonarLabs