Trained on FLUX.1 [dev] with 120 photos of Barbara Palvin in 2024 with detailed GPT-4 captions. Tested on FLUX 1.dev (full) and FLUX fp8 and FLUX nf4 ! Use around strength 0.8-1.0. 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:barpalvin_local_merger_29_54_04_06:1>

FLUX v1.0 Palvin (2019-2022) :

Trained on FLUX.1 [dev] with 80 photos of Barbara Palvin (2019-2022) with detailed GPT-4 captions. Tested on FLUX 1.dev (full) and FLUX fp8 and FLUX nf4 ! Use around strength 0.8-1.0. 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:Barbara_Palvin_FLUX_v1-000061:1>

FLUX v3.0 (2017-2019):

Trained on FLUX.1 [dev] with 80 photos of Barbara Palvin (2017-2019) with detailed GPT-4 captions. Tested on FLUX 1.dev (full) and FLUX fp8 and FLUX nf4 ! Use around strength 0.8-1.0. 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:Barbara_Palvin_2017_2019_FLUX_v2-000037:1.0>

SDXL v3.0 :

Trained on Juggernaut X with 210 photos of mdoel Barbara Palvin. Tested on Juggernaut X, Juggernaut v7, RealismEngine 2, RealVisXL3 and AlbedoBaseV2 ! Use with keyword : "brxpalvin" . Use around strength 1.0 or less. Can be used for example as follows:

Positive : {Artstyle}, {Character and scene description}, brxpalvin, <lora:barpalvin_juggerX_xl_1_wocap_merger_84_91_06_04-brxpalvin:1>

Negative : ugly, deformed, airbrushed, photoshop, rendered, (multiple people), child, mouth open

SDXL v2.0 160mb :

Trained on 183 photos of Barbara Palvin on SDXL 1.0 base with better Lora presets. Less than half the size of the previous version. Use with keyword : "barxpalvin"! Lora strength with prompt should be around 1.0-1.1. Following prompt structure on Juggernaut v7, realismEngine 2.0 or Juggernautv9 lightning works well:

positive : {art style}, {character descrioption and action},barxpalvin, <lora:barpalvin_32_small_xl_4_standard_wo_cap-merger-11_29_69_04_036_024-barxpalvin:1.05>

negative : photoshop, rendered, (multiple people), blurry, child, braids

SDXL v1.0:

Lora Model trained on 250 images of Barbara Palvin on the SDXL 1.0 Base model. The Lora weight should be 1.0-1.1.

Download model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.

Use it with the 🧨 diffusers library

from diffusers import AutoPipelineForText2Image
import torch

device = "cuda" if torch.cuda.is_available() else "cpu"

pipeline = AutoPipelineForText2Image.from_pretrained('black-forest-labs/FLUX.1-dev', torch_dtype=torch.bfloat16).to(device)
pipeline.load_lora_weights('Keltezaa/barbara-palvin-sdxl-flux', weight_name='barpalvin_local_merger_29_54_04_06.safetensors')
image = pipeline('High quality photo in side view of a woman looking at the viewer standing on top of a hill in the windy scottish highlands wearing a tight full sleeve figure hugging turtle neck shirt  and white dress pants that are also form fitted,  , ').images[0]

For more details, including weighting, merging and fusing LoRAs, check the documentation on loading LoRAs in diffusers