Cathy is a photorealistic character LoRA trained for FLUX.1, designed to reproduce a consistent female character identity across a wide range of compositions, expressions, hairstyles, clothing, environments, lighting conditions, and camera angles.
The training dataset was curated from native-resolution photographic-style images with varied framing and visual conditions. The LoRA is intended primarily for character consistency rather than a specific artistic style.
Trigger word: cathy001
Recommended usage: Start with a LoRA strength around 0.7–1.0 and adjust according to the desired balance between identity strength and prompt flexibility.
For best results: Use cathy001 in the prompt and describe the desired pose, camera angle, expression, clothing, environment, hairstyle, and lighting separately. The character is intended to work across portraits, upper-body compositions, three-quarter shots, and full-body scenes.
Base model: FLUX.1
Training Details
Base Model: FLUX.1
LoRA Type: Character-focused LoRA
Dataset Size: 300+ images
Training Steps: 2,400
Dataset Resolution: Native-resolution images
Training Approach: Bucketed training with aspect-ratio preservation
Primary Focus: Character identity and consistency
Intended Use
This model is intended for creative image generation, character development, fashion/editorial photography, cinematic scenes, and storytelling.
Description
Initial release of Cathy v1.0. Character-focused FLUX.1 LoRA trained for consistent identity across varied poses, compositions, hairstyles, clothing, environments, lighting, and camera angles. Trained for 2,400 steps using a varied 300+ image dataset.