edenartlab/sdxl-pipelines – Run with an API on Replicate

edenartlab / sdxl-pipelines

Cold

Public
554.9K runs
A100 (80GB)

Run with an API

Playground API Examples README Versions

Input

mode

string

Mode

Default: "create"

stream

boolean

yield individual results if True

Default: false

stream_every

integer

(minimum: 1, maximum: 25)

for mode create, how many steps per update to stream (steam must be set to True)

Default: 1

width

integer

(minimum: 512, maximum: 2048)

Width

Default: 1024

height

integer

(minimum: 512, maximum: 2048)

Height

Default: 1024

checkpoint

string

Which Stable Diffusion checkpoint to use

Default: "juggernaut_XL2"

lora

string

Shift + Return to add a new line

(optional) URL of Lora finetuning

lora_scale

number

(minimum: 0, maximum: 1.5)

Lora scale (how much of the Lora finetuning to apply)

Default: 0.7

sampler

string

Which sampler to use

Default: "euler"

steps

integer

(minimum: 10, maximum: 70)

Diffusion steps

Default: 35

guidance_scale

number

(minimum: 0, maximum: 20)

Strength of text conditioning guidance

Default: 7.5

upscale_f

number

(minimum: 1, maximum: 2)

Upscaling resolution

Default: 1

init_image

string

Shift + Return to add a new line

Load initial image from file, url, or base64 string

init_image_strength

number

(minimum: 0, maximum: 1)

Strength of initial image

Default: 0

adopt_aspect_from_init_img

boolean

Adopt aspect ratio from init image

Default: true

controlnet_type

string

Controlnet type

Default: "off"

control_image

string

Shift + Return to add a new line

image for controlnet guidance

control_image_strength

number

(minimum: 0, maximum: 1.5)

Strength of control image

Default: 0

ip_image

string

Shift + Return to add a new line

Load ip_adapter image from file, url, or base64 string

ip_image_strength

number

(minimum: 0, maximum: 1.25)

Strength of image conditioning from ip_adapter (vs txt conditioning from clip-interrogator or prompt) (used in remix, upscale, blend and real2real)

Default: 0.65

text_input

string

Shift + Return to add a new line

Text input

text_inputs_to_interpolate

string

Shift + Return to add a new line

Text inputs to interpolate, separated by |

text_inputs_to_interpolate_weights

string

Shift + Return to add a new line

Text input weights to interpolate, separated by |

uc_text

string

Shift + Return to add a new line

nude, naked, text, watermark, low-quality, signature, padding, margins, white borders, padded border, moiré pattern, downsampling, aliasing, distorted, blurry, blur, jpeg artifacts, compression artifacts, poorly drawn, low-resolution, bad, grainy, error, bad-contrastnude, naked, text, watermark, low-quality, signature, padding, margins, white borders, padded border, moiré pattern, downsampling, aliasing, distorted, blurry, blur, jpeg artifacts, compression artifacts, poorly drawn, low-resolution, bad, grainy, error, bad-contrast

Negative text input (mode==all)

Default: "nude, naked, text, watermark, low-quality, signature, padding, margins, white borders, padded border, moiré pattern, downsampling, aliasing, distorted, blurry, blur, jpeg artifacts, compression artifacts, poorly drawn, low-resolution, bad, grainy, error, bad-contrast"

seed

integer

(minimum: 0, maximum: 10000000000)

random seed

Default: 13

n_samples

integer

(minimum: 1, maximum: 4)

batch size

Default: 1

n_frames

integer

(minimum: 3, maximum: 1000)

Total number of frames for video modes

Default: 40

interpolation_texts

string

Shift + Return to add a new line

Interpolation texts for video modes

interpolation_seeds

string

Shift + Return to add a new line

Seeds for interpolated texts for video modes

interpolation_init_images

string

Shift + Return to add a new line

Interpolation init images, file paths or urls for video modes

interpolation_init_images_power

number

(minimum: 0.5, maximum: 5)

Power for interpolation_init_images prompts for video modes

Default: 2.5

interpolation_init_images_min_strength

number

(minimum: 0, maximum: 1)

Minimum init image strength for interpolation_init_images prompts for video modes

Default: 0.05

interpolation_init_images_max_strength

number

(minimum: 0, maximum: 1)

Maximum init image strength for interpolation_init_images prompts for video modes

Default: 0.95

audio_file

string

Shift + Return to add a new line

An audio file to use for real2real_audio

loop

boolean

Loops (mode==interpolate & real2real)

Default: true

smooth

boolean

Smooth (mode==interpolate & real2real)

Default: true

latent_blending_skip_f

string

Shift + Return to add a new line

What fraction of the denoising trajectory to skip at the start and end of each interpolation phase, two floats, separated by a pipe (|)

Default: "0.05|0.6"

n_anchor_imgs

integer

(minimum: 3, maximum: 6)

Number of anchor frames to render (including keyframes) before activating latent blending

Default: 3

n_film

integer

(minimum: 0, maximum: 3)

Number of times to smooth final frames with FILM (default is 0) (mode==interpolate)

Default: 1

fps

integer

(minimum: 1, maximum: 30)

Frames per second (mode==interpolate & real2real)

Default: 12

use_lcm

boolean

Smooth (mode==interpolate & real2real)

Default: false

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run edenartlab/sdxl-pipelines using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "edenartlab/sdxl-pipelines:a536abda9b4633adb3b76b6eddcdf8680546c573a820fac706ff4c994a5b09ce",
    input={
        "fps": 12,
        "loop": True,
        "mode": "create",
        "seed": 13,
        "steps": 35,
        "width": 1024,
        "height": 1024,
        "n_film": 1,
        "smooth": True,
        "stream": False,
        "sampler": "euler",
        "uc_text": "nude, naked, text, watermark, low-quality, signature, padding, margins, white borders, padded border, moiré pattern, downsampling, aliasing, distorted, blurry, blur, jpeg artifacts, compression artifacts, poorly drawn, low-resolution, bad, grainy, error, bad-contrast",
        "use_lcm": False,
        "n_frames": 40,
        "n_samples": 1,
        "upscale_f": 1,
        "checkpoint": "juggernaut_XL2",
        "lora_scale": 0.7,
        "stream_every": 1,
        "n_anchor_imgs": 3,
        "guidance_scale": 7.5,
        "controlnet_type": "off",
        "ip_image_strength": 0.65,
        "init_image_strength": 0,
        "control_image_strength": 0,
        "latent_blending_skip_f": "0.05|0.6",
        "adopt_aspect_from_init_img": True,
        "interpolation_init_images_power": 2.5,
        "interpolation_init_images_max_strength": 0.95,
        "interpolation_init_images_min_strength": 0.05
    }
)

# The edenartlab/sdxl-pipelines model can stream output as it's running.
# The predict method returns an iterator, and you can iterate over that output.
for item in output:
    # https://replicate.com/edenartlab/sdxl-pipelines/api#output-schema
    print(item)

To learn more, take a look at the guide on getting started with Python.

Output

No output yet! Press "Submit" to start a prediction.

Run time and cost

This model costs approximately $0.040 to run on Replicate, or 25 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia A100 (80GB) GPU hardware. Predictions typically complete within 29 seconds. The predict time for this model varies significantly based on the inputs.

Readme

This model doesn't have a readme.