Please Donate Buzz for FLUX Lora Training !
FLUX v1.0 :
For this version I ask for some buzz, since training FLUX costs more than 2000 Buzz per model.
Trained on FLUX.1 [dev] with 160 photos of Miranda Kerr with detailed GPT-4 captions. Tested on FLUX 1.D (full) and FLUX fp8 and FLUX nf4 ! Use around strength 0.8-1.2. No keywords needed! Distilled CFG around 1-4 and CFG 1.0 (without negative prompt). Clipskip 1. Can be used for example as follows:
Positive : {Artstyle, Character and scene description in usual FLUX fashion}, <lora:Miranda_Kerr_FLUX_v1-000040:1>
v3.0 :
Trained on Juggernaut X with 160 photos of Miranda Kerr . Tested on Juggernaut X, Juggernaut v7, RealismEngine 2, RealVisXL3, AlbedoBaseV2! Use with keyword : "mirxkerr" . Use around strength 1.0-1.1. CFG 5.0-7.0. Clipskip 1. Can be used for example as follows:
Positive : {Artstyle}, {Character and scene description},mirxkerr, <lora:mirkerr_juggerX_xl_1_wocap_merger_33_126_01_09-mirxkerr:1>
Negative : ugly, deformed, airbrushed, photoshop, rendered, (multiple people), child
v2.0:
Lora of Australian celebrity and model Miranda Kerr trained with 160 images of her on SDXL 1.0 base model and captioned with GPT-4 Vision. The Lora model is also improved to allow more flexibility while preserving likeness. Recommended weight is 0.8-1.1.
v1.0 and v1.1:
Lora of Australian celebrity and model Miranda Kerr trained with 190 images of her on SDXL 1.0 base model. Recommended weight for ComfyUi is 0.9-1.1.
