Norman Lewis

Custom FLUX.1-dev LoRA, fine-tuned on Norman Lewis's own canvas work.

A LoRA (low-rank adaptation) is a compact set of trainable weights that sits on top of a much larger frozen base model. Instead of retraining FLUX.1-dev, the open-source 12-billion-parameter text-to-image transformer from Black Forest Labs, the LoRA only adjusts ~50 million new parameters using a small artist-specific dataset. The result keeps FLUX's general image-generation ability intact while teaching it one artist's visual language.

Pieces in this gallery are produced by a private FLUX.1-dev LoRA trained on 14 hand-curated reproductions from WikiArt catalog, drawn from his mature abstract period (1945-1979) - calligraphic processions of small marks, glyphs, and crowds dissolving across deep atmospheric grounds. Images were auto-captioned by the trainer with the trigger word prepended, binding it to the artist's visual language rather than a generic style label.

Trigger word norman_lewis_style is prepended to every weather-derived prompt before inference. Trained via ostris/flux-dev-lora-trainer on Replicate.