Cezanne-Phase1-Log-SNR-Try2

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: FlowMatchEulerDiscreteScheduler
• Seed: 42
• Resolution: 968x1280
• Skip-layer guidance:

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 a Paul Cezanne oil painting, A landscape with trees, a body of water reflecting greenery, and a building with a tower in the background. The scene is dominated by natural elements, with the tall trees and textured foliage surrounding a calm water surface.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Paul Cezanne oil painting, A man with curly hair and a beard, wearing a suit and a blue bow tie, is depicted at an angle with his head tilted to the side. The background is a plain light color.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Paul Cezanne oil painting, Three nude figures outdoors near water, surrounded by trees and greenery. One stands, another sits with their back turned, and a third kneels by the water's edge. The environment includes green foliage and a partly cloudy sky.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Paul Cezanne oil painting, A man is seated at a table, leaning forward and holding playing cards. He wears a brown suit with a white shirt and a brown hat. The background is plain with a hint of a curtain or drapery on the right side.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Paul Cezanne oil painting, A collection of modern technology devices arranged on a wooden table - a smartphone, laptop, and wireless earbuds - alongside traditional apples and a wine bottle. The objects cast subtle shadows on a white tablecloth, with a window visible in the background.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Paul Cezanne oil painting, A contemporary city street with geometric buildings and trees, showing the interplay of natural and architectural forms. Electric vehicles and pedestrians move through the scene, while maintaining characteristic treatment of perspective and form.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Paul Cezanne oil painting, Inside a modern coffee shop, baristas work at elaborate espresso machines while customers sit at wooden tables with their laptops. Steam rises from cups, creating atmospheric effects similar to the treatment of natural elements.

Negative Prompt

blurry, cropped, ugly

Prompt

In the style of a Paul Cezanne oil painting, A factory complex nestled in a traditional Provençal landscape, where industrial structures echo the geometric forms of Mont Sainte-Victoire. Solar panels glint on rooftops while cypress trees frame the scene.

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: 7
• Training steps: 6000
• Learning rate: 0.0004
  • Learning rate schedule: polynomial
  • Warmup steps: 100
• Max grad norm: 0.1
• Effective batch size: 3
  • Micro-batch size: 3
  • Gradient accumulation steps: 1
  • Number of GPUs: 1
• Gradient checkpointing: True
• Prediction type: flow-matching (extra parameters=['shift=3', 'flux_guidance_mode=constant', 'flux_guidance_value=1.0', 'flow_matching_loss=compatible'])
• Optimizer: adamw_bf16
• Trainable parameter precision: Pure BF16
• Caption dropout probability: 10.0%

LyCORIS Config:

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

Datasets

cezanne-512

• Repeats: 11
• Total number of images: 33
• Total number of aspect buckets: 2
• Resolution: 0.262144 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

cezanne-768

• Repeats: 11
• Total number of images: 33
• Total number of aspect buckets: 9
• Resolution: 0.589824 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

cezanne-1024

• Repeats: 5
• Total number of images: 33
• Total number of aspect buckets: 2
• Resolution: 1.048576 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

cezanne-1536

• Repeats: 2
• Total number of images: 33
• Total number of aspect buckets: 14
• Resolution: 2.359296 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

cezanne-crops-512

• Repeats: 11
• Total number of images: 33
• Total number of aspect buckets: 1
• Resolution: 0.262144 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square
• Used for regularisation data: No

cezanne-crops-768

• Repeats: 11
• Total number of images: 33
• Total number of aspect buckets: 1
• Resolution: 0.589824 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square
• Used for regularisation data: No

cezanne-crops-1024

• Repeats: 5
• Total number of images: 33
• 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


def download_adapter(repo_id: str):
    import os
    from huggingface_hub import hf_hub_download
    adapter_filename = "pytorch_lora_weights.safetensors"
    cache_dir = os.environ.get('HF_PATH', os.path.expanduser('~/.cache/huggingface/hub/models'))
    cleaned_adapter_path = repo_id.replace("/", "_").replace("\\", "_").replace(":", "_")
    path_to_adapter = os.path.join(cache_dir, cleaned_adapter_path)
    path_to_adapter_file = os.path.join(path_to_adapter, adapter_filename)
    os.makedirs(path_to_adapter, exist_ok=True)
    hf_hub_download(
        repo_id=repo_id, filename=adapter_filename, local_dir=path_to_adapter
    )

    return path_to_adapter_file
    
model_id = 'black-forest-labs/FLUX.1-dev'
adapter_repo_id = 'davidrd123/Cezanne-Phase1-Log-SNR-Try2'
adapter_filename = 'pytorch_lora_weights.safetensors'
adapter_file_path = download_adapter(repo_id=adapter_repo_id)
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
lora_scale = 1.0
wrapper, _ = create_lycoris_from_weights(lora_scale, adapter_file_path, pipeline.transformer)
wrapper.merge_to()

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


## Optional: quantise the model to save on vram.
## Note: The model was quantised during training, and so it is recommended to do the same during inference time.
from optimum.quanto import quantize, freeze, qint8
quantize(pipeline.transformer, weights=qint8)
freeze(pipeline.transformer)
    
pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
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(42),
    width=968,
    height=1280,
    guidance_scale=3.0,
).images[0]
image.save("output.png", format="PNG")