Flux-Leonetto-Cappiello-LoKr

This is a LyCORIS adapter derived from black-forest-labs/FLUX.1-dev.

No validation prompt was used during training.

None

Validation settings

CFG: 3.0
CFG Rescale: 0.0
Steps: 20
Sampler: None
Seed: 42
Resolution: 1024x1024

Note: The validation settings are not necessarily the same as the training settings.

You can find some example images in the following gallery:

Prompt
unconditional (blank prompt)

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A woman in a flowing crimson dress becomes one with a bottle of wine, her form spiraling upward in a liquid dance. Her dress transforms into wine splashing against a pitch-black background, with 'BORDEAUX MYSTIQUE' in elegant gold lettering.

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A joyful acrobat in a yellow costume leaps through a giant swirling lollipop, trailing rainbow sugar crystals. Set against a deep violet background, with 'BONBONS CIRCUS' in circus-style typography.

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A barista in vibrant green becomes a living coffee tree, their arms transforming into branches bearing glowing coffee beans. Steam forms mysterious symbols against the black background, with 'CAFÉ MYSTÈRE' in art nouveau text.

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A figure in a glowing blue bodysuit phases through multiple dimensions while holding an impossible geometric shape. Digital artifacts and quantum patterns spiral around them against a void-black background. Text reads 'QUANTUM BROWSE' in fractal-based letters.

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A digital shaman in a scarlet robe dances with floating holographic memes, their form partially dissolving into viral symbols and trending hashtags. Against a deep cyber-blue background, with 'VIRAL ELIXIR' in glitch-art typography.

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A figure in an electric purple suit merges with a giant neural network, their body becoming one with glowing synapses and data streams. Set against a midnight background, text proclaims 'BRAIN BOOST PRIME' in circuit-board lettering.

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A social media influencer in a prismatic dress becomes a living feed, their form unraveling into an infinite scroll of content and eldritch emoji. Against a deep space background, with 'INFINITE FEED' in impossible geometry text.

Negative Prompt
blurry, cropped, ugly

Prompt
In the style of c4pp13lll0 vintage advertising poster, A digital deity in orange and gold robes emerges from a smartphone, trailing augmented reality artifacts and virtual blessings. Their form shifts between human and divine avatar against a dark technological void. Text reads 'METAMORPH NOW' in transcendent typography.

Negative Prompt
blurry, cropped, ugly

The text encoder was not trained. You may reuse the base model text encoder for inference.

Training settings

Training epochs: 5
Training steps: 6000
Learning rate: 0.0008
Max grad norm: 2.0
Effective batch size: 2
- Micro-batch size: 2
- Gradient accumulation steps: 1
- Number of GPUs: 1
Prediction type: flow-matching (flux parameters=['flux_guidance_value=1.0'])
Rescaled betas zero SNR: False
Optimizer: adamw_bf16
Precision: Pure BF16
Quantised: Yes: int8-quanto
Xformers: Not used
LyCORIS Config:

{
    "algo": "lokr",
    "multiplier": 1.0,
    "linear_dim": 10000,
    "linear_alpha": 1,
    "factor": 16,
    "apply_preset": {
        "target_module": [
            "Attention",
            "FeedForward"
        ],
        "module_algo_map": {
            "Attention": {
                "factor": 16
            },
            "FeedForward": {
                "factor": 8
            }
        }
    }
}

Datasets

leonetto-cappiello-512

Repeats: 12
Total number of images: 62
Total number of aspect buckets: 4
Resolution: 0.262144 megapixels
Cropped: False
Crop style: None
Crop aspect: None
Used for regularisation data: No

leonetto-cappiello-1024

Repeats: 4
Total number of images: 62
Total number of aspect buckets: 4
Resolution: 1.048576 megapixels
Cropped: False
Crop style: None
Crop aspect: None
Used for regularisation data: No

leonetto-cappiello-512-crop

Repeats: 12
Total number of images: 62
Total number of aspect buckets: 1
Resolution: 0.262144 megapixels
Cropped: True
Crop style: random
Crop aspect: square
Used for regularisation data: No

leonetto-cappiello-1024-crop

Repeats: 4
Total number of images: 62
Total number of aspect buckets: 1
Resolution: 1.048576 megapixels
Cropped: True
Crop style: random
Crop aspect: square
Used for regularisation data: No

Inference

import torch
from diffusers import DiffusionPipeline
from lycoris import create_lycoris_from_weights

model_id = 'black-forest-labs/FLUX.1-dev'
adapter_id = 'pytorch_lora_weights.safetensors' # you will have to download this manually
lora_scale = 1.0
wrapper, _ = create_lycoris_from_weights(lora_scale, adapter_id, pipeline.transformer)
wrapper.merge_to()

prompt = "An astronaut is riding a horse through the jungles of Thailand."

pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu')
image = pipeline(
    prompt=prompt,
    num_inference_steps=20,
    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(1641421826),
    width=1024,
    height=1024,
    guidance_scale=3.0,
).images[0]
image.save("output.png", format="PNG")

davidrd123
/

Flux-Leonetto-Cappiello-LoKr