<h2 id="french-manicure-lora"><strong>French Manicure LoRA</strong></h2><p><strong>Note: </strong><a rel="ugc" href="https://civarchive.com/models/2407120?modelVersionId=2706503"><strong>v2</strong></a><strong> is trained with a cleaner dataset and performs slightly better.</strong></p><p>A LoRA fine-tune of Flux.1 D trained on a curated dataset of french manicure images.</p><h2 id="trigger-word"><strong>Trigger Word</strong></h2><p>Use frnchmani in your prompt to activate the LoRA style.</p><p>frnchmani, your prompt here</p><h2 id="usage"><strong>Usage</strong></h2><p>Load this LoRA in ComfyUI using LoraLoaderModelOnly (trained on unet only) with strength 0.8–1.0.</p><h2 id="training-details"><strong>Training Details</strong></h2><p><strong>ParameterValue</strong>Base modelFlux.1 DNetwork rank (dim)16Network alpha1.0Precisionbf16Steps1500Learning rate0.0004OptimizerCAMELR schedulerconstantTraining resolutions512, 768, 1024pxBucket trainingNo — see dataset configWeighting schemelogit_normalTimestep samplingshift</p><h2 id="compatibility"><strong>Compatibility</strong></h2><ul><li><p>Base model: Flux.1 D</p></li><li><p>Node: LoraLoaderModelOnly (unet-only — no text encoder weights)</p></li><li><p>Recommended strength: 0.8 – 1.0</p></li><li><p>Sampler: Euler + Simple scheduler, 20 steps</p></li><li><p>Guidance: FluxGuidance 3.8-4.8 (for Flux)</p></li></ul><p></p>

French Manicure LoRA

<p>Trained 1500 steps on Flux.1 D. Rank 16, alpha 1.0, bf16 precision. Optimizer: CAME, LR: 0.0004. Trigger: <code>frnchmani</code>.</p>

Files with SHA256: 328b38747154500d74edf32cdd281325a64819b9e740184ffd207fa074ef08f5