webuiapi 0.0.1

Python API client for AUTOMATIC1111/stable-diffusion-webui

sdwebuiapi

API client for AUTOMATIC1111/stable-diffusion-webui

Supports txt2img, img2img, extra-single-image, extra-batch-images API calls.

API support have to be enabled from webui. Add --api when running webui. It's explained here.

You can use --api-auth user1:pass1,user2:pass2 option to enable authentication for api access. (Since it's basic http authentication the password is transmitted in cleartext)

API calls are (almost) direct translation from http://127.0.0.1:7860/docs as of 2022/11/21.

Usage

webuiapi_demo.ipynb contains example code with original images. Images are compressed as jpeg in this document.

create API client

import webuiapi

# create API client
api = webuiapi.WebUIApi()

# create API client with custom host, port
api = webuiapi.WebUIApi(host='127.0.0.1', port=7860)

# create API client with default sampler, steps.
api = webuiapi.WebUIApi(sampler='Euler a', steps=20)

# optionally set username, password when --api-auth is set on webui.
api.set_auth('username', 'password')

txt2img

result1 = api.txt2img(prompt="cute squirrel",
                    negative_prompt="ugly, out of frame",
                    seed=1003,
                    styles=["anime"],
                    cfg_scale=7,
#                      sampler_index='DDIM',
#                      steps=30,
                    )
# images contains the returned images (PIL images)
result1.images

# image is shorthand for images[0]
result1.image

# info contains text info about the api call
result1.info

# info contains paramteres of the api call
result1.parameters

result1.image

img2img

result2 = api.img2img(images=[result1.image], prompt="cute cat", seed=5555, cfg_scale=6.5, denoising_strength=0.6)
result2.image

img2img inpainting

from PIL import Image, ImageDraw

mask = Image.new('RGB', result2.image.size, color = 'black')
# mask = result2.image.copy()
draw = ImageDraw.Draw(mask)
draw.ellipse((210,150,310,250), fill='white')
draw.ellipse((80,120,160,120+80), fill='white')

mask

inpainting_result = api.img2img(images=[result2.image],
                                mask_image=mask,
                                inpainting_fill=1,
                                prompt="cute cat",
                                seed=104,
                                cfg_scale=5.0,
                                denoising_strength=0.7)
inpainting_result.image

extra-single-image

result3 = api.extra_single_image(image=result2.image,
                                 upscaler_1=webuiapi.Upscaler.ESRGAN_4x,
                                 upscaling_resize=1.5)
print(result3.image.size)
result3.image

(768, 768)

extra-batch-images

result4 = api.extra_batch_images(images=[result1.image, inpainting_result.image],
                                 upscaler_1=webuiapi.Upscaler.ESRGAN_4x,
                                 upscaling_resize=1.5)
result4.images[0]

result4.images[1]

Configuration APIs

# return map of current options
options = api.get_options()

# change sd model
options = {}
options['sd_model_checkpoint'] = 'model.ckpt [7460a6fa]'
api.set_options(options)

# when calling set_options, do not pass all options returned by get_options().
# it makes webui unusable (2022/11/21).

# get available sd models
api.get_sd_models()

# misc get apis
api.get_samplers()
api.get_cmd_flags()      
api.get_hypernetworks()
api.get_face_restorers()
api.get_realesrgan_models()
api.get_prompt_styles()
api.get_artist_categories()
api.get_artists()