kadirnar/diffusersplus

This project is under development.

Diffusers++: A User-Friendly and Diffusers-Based Library.

Installation

pip install diffusersplus

Usage

To use the diffusersplus library, follow the steps below for different tasks:

Stable Diffusion Text2Image Generate:

from diffusersplus import diffusion_pipeline

model = diffusion_pipeline(
    task_id="stable-txt2img", 
    stable_model_id="dreamlike-art/dreamlike-anime-1.0", 
    scheduler_name="DDIM"
)

output = model(
    prompt="A photo of an anime character",
    negative_prompt="bad",
    num_images_per_prompt=1,
    num_inference_steps=30,
    guidance_scale=7.0,
    guidance_rescale=0.0,
    generator_seed=0,
    height=512,
    width=512,
)

Stable Diffusion Image2Image Generate:

from diffusersplus import diffusion_pipeline

model = diffusion_pipeline(
    task_id="stable-img2img",
    stable_model_id="dreamlike-art/dreamlike-anime-1.0",
    scheduler_name="DDIM"
)

output = model(
    image_path="../data/image.png",
    prompt="A photo of a cat.",
    negative_prompt="bad",
    num_images_per_prompt=1,
    num_inference_steps=50,
    guidance_scale=7.0,
    strength=0.5,
    generator_seed=0,
    resize_type="center_crop_and_resize",
    crop_size=512,
    height=512,
    width=512,
)

Stable Diffusion Upscale:

from diffusersplus import diffusion_pipeline

model = diffusion_pipeline(
    task_id="stable-upscale",
    stable_model_id="stabilityai/stable-diffusion-x4-upscaler",
    scheduler_name="DDIM"
)

output = model(
    image_path="../data/image.png",
    prompt="A photo of a anime character.",
    negative_prompt="bad",
    resize_type="center_crop_and_resize",
    noise_level=20,
    num_images_per_prompt=1,
    num_inference_steps=20,
    guidance_scale=7.0,
    generator_seed=0,
)

Controlnet:

from diffusersplus import diffusion_pipeline

model = diffusion_pipeline(
    task_id="controlnet",
    stable_model_id="dreamlike-art/dreamlike-anime-1.0",
    controlnet_model_id="lllyasviel/sd-controlnet-canny",
    scheduler_name="DDIM",
)
output = model(
    image_path="../data/image.png",
    prompt="A photo of cat.",
    negative_prompt="bad",
    height=512,
    width=512,
    preprocess_type="Canny",
    resize_type="center_crop_and_resize",
    guess_mode=False,
    num_images_per_prompt=1,
    num_inference_steps=50,
    guidance_scale=7.0,
    controlnet_conditioning_scale=0.2,
    generator_seed=0,
)

Controlnet Inpaint:

from diffusersplus import diffusion_pipeline

model = diffusion_pipeline(
    task_id="controlnet-inpaint",
    stable_model_id="dreamlike-art/dreamlike-anime-1.0",
    controlnet_model_id="lllyasviel/sd-controlnet-canny",
    scheduler_name="DDIM",
)
output = model(
    image_path="../data/image.png",
    mask_path="../data/mask_image.png",
    prompt="A photo of a cat.",
    negative_prompt="bad",
    height=512,
    width=512,
    preprocess_type="Canny",
    resize_type="center_crop_and_resize",
    strength=0.5,
    guess_mode=False,
    num_images_per_prompt=1,
    num_inference_steps=50,
    guidance_scale=7.0,
    controlnet_conditioning_scale=1.0,
    generator_seed=0,
)

Controlnet Image2Image:

from diffusersplus import diffusion_pipeline

model = diffusion_pipeline(
    task_id="controlnet-img2img",
    stable_model_id="dreamlike-art/dreamlike-anime-1.0",
    controlnet_model_id="lllyasviel/sd-controlnet-canny",
    scheduler_name="DDIM",
)
output = model(
    image_path="../data/image.png",
    prompt="A photo of a cat.",
    negative_prompt="bad",
    height=512,
    width=512,
    preprocess_type="Canny",
    resize_type="center_crop_and_resize",
    guess_mode=False,
    num_images_per_prompt=1,
    num_inference_steps=20,
    guidance_scale=7.0,
    controlnet_conditioning_scale=1.0,
    strength=0.5,
    generator_seed=0,
)