john-singer-sargent-sdxl-lokr-02

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

The main validation prompt used during training was:

jss_style, A young boy holds a small black dog in his arms. He wears a red bow tie and stands in front of a textured backdrop. His red socks and shoes are notable.

Validation settings

• CFG: 4.2
• CFG Rescale: 0.0
• Steps: 30
• 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

jss_style, A woman sits slouched on a sofa beneath a mantelpiece, wrapped in blankets. Her eyes are closed, and her head rests on the back.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, A man wearing a lavish robe holds a large, decorated sword. A young boy stands behind him, holding a cushion with a crown. They are in an ornate hall.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, Three young girls in white aprons. One stands alone, two stand together by a large vase. A girl sits on the floor holding a doll. Large vases flank the scene.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, hamster

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, hipster making a chair

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, A elegant woman in a long, black evening gown stands in a grand, dimly lit room. She has a confident pose, with one hand on her hip. A ornate gold frame mirror is visible in the background.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, An opulent interior with a grand piano in the foreground. A woman in a white dress sits at the piano, her back to the viewer. Sunlight streams through tall windows.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, A portrait of a modern tech CEO in a casual outfit, standing in front of a wall of computer screens

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, A group portrait of a modern diverse family in a living room. Capture individual personalities while maintaining group cohesion.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, A lively scene in a 1920s dance hall. Couples dancing, musicians playing, ambient lighting. Capture the movement and atmosphere in Sargent's distinctive style.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, People dining at tables outside a Parisian café.

Negative Prompt

blurry, cropped, ugly

Prompt

jss_style, A young boy holds a small black dog in his arms. He wears a red bow tie and stands in front of a textured backdrop. His red socks and shoes are notable.

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: 8
• Training steps: 8500
• Learning rate: 0.0001
• Effective batch size: 4
  • Micro-batch size: 4
  • Gradient accumulation steps: 1
  • Number of GPUs: 1
• Prediction type: epsilon
• Rescaled betas zero SNR: False
• Optimizer: optimi-lion
• 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

jss-sdxl-512

• Repeats: 10
• Total number of images: 81
• Total number of aspect buckets: 7
• Resolution: 0.262144 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None

jss-sdxl-1024

• Repeats: 10
• Total number of images: 81
• Total number of aspect buckets: 14
• Resolution: 1.048576 megapixels
• Cropped: False
• Crop style: None
• Crop aspect: None

jss-sdxl-512-crop

• Repeats: 10
• Total number of images: 81
• Total number of aspect buckets: 1
• Resolution: 0.262144 megapixels
• Cropped: True
• Crop style: random
• Crop aspect: square

jss-sdxl-1024-crop

• Repeats: 10
• Total number of images: 81
• 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 = "jss_style, A young boy holds a small black dog in his arms. He wears a red bow tie and stands in front of a textured backdrop. His red socks and shoes are notable."
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=30,
    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")