--- license: other base_model: "terminusresearch/FluxBooru-v0.3" tags: - flux - flux-diffusers - text-to-image - diffusers - simpletuner - not-for-all-audiences - lora - template:sd-lora - lycoris inference: true widget: - text: 'unconditional (blank prompt)' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_0_0.png - text: 'a photo of a q1s woman in a suit standing on the streets of Tokyo' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_1_0.png - text: 'a anime of a q1s woman in school uniform standing on a bridge' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_2_0.png - text: 'a photo of a q1s woman in leather jacket on the beach' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_3_0.png - text: 'A stunning portrait of a q1s woman adorned with a luxurious wide-brimmed red hat that elegantly covers part of her face, illustration, vibrant, sophisticated, portrait photography' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_4_0.png - text: 'A grainy, sepia-toned photograph taken with a Kodak Portra 400 film camera in the late 1990s of a q1s woman with captivating gaze poses confidently. She wears a vintage, black MinMin T-shirt, slightly oversized and slightly worn, with the band''s iconic logo prominently displayed. The background is blurred, suggesting a dimly lit, underground music venue or a friend''s basement.' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_5_0.png - text: 'A captivating minimalist masterpiece illustration of a q1s woman exuding elegance and tranquility. The beautiful subject is dressed in a linen dress, featuring pastel beige and light grey hues. The background consists of a harmonious composition of rectangles and squares, subtly outlined in black pencil. This vibrant and conceptual artwork elicits a cinematic and fashionable illustration.' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_6_0.png - text: 'a digital painting of a q1s woman in a green hoodie and purple skirt standing in front of an office building' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_7_0.png - text: 'A whimsical miniature 3D alcohol ink portrait of a q1s woman mermaid nestled in a giant seashell. She has a shimmering fish tail adorned with intricate scales in various shades of blue and turquoise. She wears a top made of tiny seashells and pearls. Surrounding her are small, bioluminescent jellyfish, adding a magical glow to the scene. The lighting creates a dreamy, underwater atmosphere that captures the essence of oceanic wonder and enchantment.' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_8_0.png - text: 'a photo of a q1s woman as a historical marble statue in a grand bathroom' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_9_0.png - text: 'A captivating watercolor portrayal of a q1s woman. Her eyes are the focal point. The vibrant color palette features bright red, deep green, and pastel beige and white, creating a striking contrast. The fluid, free-flowing brushstrokes imbue the artwork with movement and emotion. The background is an abstract tapestry of color, with splashes of the same hues, adding depth and dimension to the portrait. Painting, illustration.' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_10_0.png - text: 'an oil painting of q1s in a hoordie and skirt walking on the beach' parameters: negative_prompt: 'blurry, cropped, ugly' output: url: ./assets/image_11_0.png --- # flux-test-q1sV22 This is a LyCORIS adapter derived from [terminusresearch/FluxBooru-v0.3](https://huggingface.co/terminusresearch/FluxBooru-v0.3). The main validation prompt used during training was: ``` an oil painting of q1s in a hoordie and skirt walking on the beach ``` ## Validation settings - CFG: `3.5` - 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](#training-settings). You can find some example images in the following gallery: The text encoder **was not** trained. You may reuse the base model text encoder for inference. ## Training settings - Training epochs: 6 - Training steps: 3250 - Learning rate: 0.001 - Max grad norm: 1.0 - Effective batch size: 3 - Micro-batch size: 1 - Gradient accumulation steps: 3 - Number of GPUs: 1 - Prediction type: flow-matching (extra parameters=['shift=1.0', 'flux_guidance_value=3.5', 'flux_attention_masked_training']) - Optimizer: adamw_bf16weight_decay=1e-3 - Trainable parameter precision: Pure BF16 - Quantised base model: Yes (int8-quanto) - Xformers: Not used - LyCORIS Config: ```json { "algo": "lokr", "multiplier": 1.0, "linear_dim": 10000, "linear_alpha": 1, "factor": 12, "apply_preset": { "target_module": [ "Attention", "FeedForward" ], "module_algo_map": { "Attention": { "factor": 12 }, "FeedForward": { "factor": 6 } } } } ``` ## Datasets ### q1sV22-1024-crop - Repeats: 30 - Total number of images: 15 - Total number of aspect buckets: 1 - Resolution: 1.048576 megapixels - Cropped: True - Crop style: center - Crop aspect: square - Used for regularisation data: No ### q1sV22-512-crop - Repeats: 30 - Total number of images: 19 - Total number of aspect buckets: 1 - Resolution: 0.262144 megapixels - Cropped: True - Crop style: center - Crop aspect: square - Used for regularisation data: No ### gen-and-inpaint-q1s-1024-reg - Repeats: 0 - Total number of images: 530 - Total number of aspect buckets: 1 - Resolution: 1.048576 megapixels - Cropped: True - Crop style: center - Crop aspect: square - Used for regularisation data: Yes ## Inference ```python 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 = 'terminusresearch/FluxBooru-v0.3' adapter_repo_id = 'DragonQuix/flux-test-q1sV22' 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 oil painting of q1s in a hoordie and skirt walking on the beach" ## 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(1641421826), width=1024, height=1024, guidance_scale=3.5, ).images[0] image.save("output.png", format="PNG") ```