Model usage

This model must not be used at full strength but at approximately 70%. E.g. in Auto1111 and Forge... < lora:DroidDiffusionXL:0.7 > .

Example output

The images below were generated with this prompt:

{A Zerg| Victorian| Transformer| Swordpunk| Steampunk Cyborg| Warhammer 40000| Starcraft| Robo Recall| Protoss| Post-Structuralist| Parametric| Overwatch| Organic| Oceanpunk| Neo Viking| medieval| Neo Tokyo| Neo Noir| Hexatron| Gothic| Futuristic| Borderlands| Biomechanical| Art Deco| Celtic| Art Nouveau| Neo Noir| Post Apocalyptic| Baroque}, {A Zerg| Victorian| Transformer| Swordpunk| Steampunk Cyborg| Warhammer 40000| Starcraft| Robo Recall| Protoss| Post-Structuralist| Parametric| Overwatch| Organic| Oceanpunk| Neo Viking| medieval| Neo Tokyo| Neo Noir| Hexatron| Gothic| Futuristic| Borderlands| Biomechanical| Art Deco| Celtic| Art Nouveau| Neo Noir| Post Apocalyptic| Baroque} {A Zerg| Victorian| Transformer| Swordpunk| Steampunk Cyborg| Warhammer 40000| Starcraft| Robo Recall| Protoss| Post-Structuralist| Parametric| Overwatch| Organic| Oceanpunk| Neo Viking| medieval| Neo Tokyo| Neo Noir| Hexatron| Gothic| Futuristic| Borderlands| De Stijl| Biomechanical| Art Deco| Celtic| Art Nouveau| Neo Noir| Post Apocalyptic| Baroque} Robot < lora:DroidDiffusionXL:0.7 >

The main keyword for this model is:

• Robot

• e.g. A Protoss Art Deco Robot < lora:droiddiffusionxl:0.7 >

Model Details

• Model Name: DroidDiffusionXL

• Version: 1.0

• Model Type: Image Generative LORA Model based on SDXL Base

• Developers: Fiacre

• Release Date: May 26, 2024

• Model Repository: Hugging Face Models Hub

Overview

DroidDiffusion is a LORA (Latent Optimization with Representational Adjustment) based on the SDXL (Stable Diffusion XL) architecture. It is specially designed for generating high-quality, stylised photo of robots. It is really good at making robots that stay within the frame of the image. This was intentional as the image set was severly selected to include only in frame images.

Training Data

DroidDiffusionXL was trained on a high-quality synthetic dataset curated to include a wide variety of robotic forms and styles.

Key Configuration and Settings

• Learning Rate: 0.0009.

• Rank: 256

Limitations

• Limited styles.

Licensing and Usage

license: openrail

Future Work

Future updates will include more styles. Community suggestions are appreciated.