albert-bierstadt-sdxl-lokr

This is a LyCORIS adapter derived from stabilityai/stable-diffusion-xl-base-1.0.

The main validation prompt used during training was:

brst_style, hamster

Validation settings

CFG: 4.2
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
brst_style painting of a hipster making a chair

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style painting of a hamster

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A winding dirt road leads through a grassy area with rocks and flowers. Two people rest beside the road, one horse nearby. Tall trees frame the road, opening to a vast landscape with green fields, a lake, and towering mountains in the background. A small village with a church is visible by the lake.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A large waterfall cascades down a cliff. Trees surround the scene with some positioned on top of the cliff and others at the base. Rocks and a stream occupy the foreground. A mountain range appears in the background.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A seated woman with long dark hair is depicted in a front-facing view. She is wearing a dress with a white collar and appears to be in her thirties. Her hands are on her lap. Green leaves and flowers surround her.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, Two mountain robots stand on rocky outcrops. with skyscrapers scattered across the land. One robot is on a higher rock, looking slightly left, while the other is on a lower rock, facing right. Snow-capped mountains and green valleys stretch into the distance.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A panoramic view of Yosemite Valley with El Capitan and Half Dome visible. Lush forests in the foreground, a winding river, and dramatic clouds in the sky.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A futuristic space station orbiting Earth, with the planet's curvature and some continents visible. The space station has solar panels and docking bays.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A busy urban street with tall buildings, cars, and pedestrians. A mix of modern and classical architecture, with trees lining the sidewalks.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A small oasis in a vast desert landscape. Palm trees surround a small pool of water, with sand dunes stretching to the horizon. A camel caravan approaches in the distance.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, An underwater coral reef teeming with colorful fish, sea turtles, and various marine life. Sunlight filters through the water's surface.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, Albert Bierstadt taking a selfie with a smartphone in front of one of his landscape paintings. He's wearing his 19th-century attire but holding a modern device.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, A nighttime cityscape with skyscrapers illuminated against a starry sky. The moon is large and prominent, casting a glow over the urban landscape.

Negative Prompt
blurry, cropped, ugly

Prompt
brst_style, hamster

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: 28
Training steps: 10000
Learning rate: 0.0001
Effective batch size: 16
- Micro-batch size: 8
- Gradient accumulation steps: 1
- Number of GPUs: 2
Prediction type: epsilon
Rescaled betas zero SNR: False
Optimizer: adamw_bf16
Precision: Pure BF16
Quantised: No
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

albert-bierstadt-sdxl-512

Repeats: 10
Total number of images: ~82
Total number of aspect buckets: 6
Resolution: 0.262144 megapixels
Cropped: False
Crop style: None
Crop aspect: None

albert-bierstadt-sdxl-1024

Repeats: 10
Total number of images: ~84
Total number of aspect buckets: 12
Resolution: 1.048576 megapixels
Cropped: False
Crop style: None
Crop aspect: None

albert-bierstadt-sdxl-512-crop

Repeats: 10
Total number of images: ~78
Total number of aspect buckets: 1
Resolution: 0.262144 megapixels
Cropped: True
Crop style: random
Crop aspect: square

albert-bierstadt-sdxl-1024-crop

Repeats: 10
Total number of images: ~78
Total number of aspect buckets: 1
Resolution: 1.048576 megapixels
Cropped: True
Crop style: random
Crop aspect: square

Inference

import torch
from diffusers import DiffusionPipeline
from lycoris import create_lycoris_from_weights

model_id = 'stabilityai/stable-diffusion-xl-base-1.0'
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 = "brst_style, hamster"
negative_prompt = 'blurry, cropped, ugly'
pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu')
image = pipeline(
    prompt=prompt,
    negative_prompt=negative_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=4.2,
    guidance_rescale=0.0,
).images[0]
image.save("output.png", format="PNG")

davidrd123
/

albert-bierstadt-sdxl-lokr