Description:
This is a LoRA trained on images of Peter Dutton, an Australian politician, designed for use with the FLUX.1-dev base model. It aims to generate realistic depictions of his likeness in various contexts.
Usage Notes:
I recommend using a de-distilled derivative base model, though this should work just fine on any FLUX base. Showcase images were generated using the excellent Colossus Project Flux (v5.0_AIO) model,
I have tested with CFG 2.5 and weight 1.0 and selected the best checkpoint with these settings.
Can be used in various lighting and artistic styles.
Ethical Considerations:
This model is intended for artistic and satirical purposes. Please use responsibly and in accordance with platform guidelines regarding public figures.
Description
Was trained using 28 images, trigger word 'pdut' along with detailed captions.
28 epochs, 1960 steps.