lucataco / style-aligned

GoogleAI: Style Aligned Image Generation via Shared Attention

  • Public
  • 1.2K runs
  • A100 (80GB)
  • GitHub
  • Paper
  • License
Run with an API
Playground API Examples README Versions

Run lucataco/style-aligned with an API

Input schema

seedinteger

Random seed. Leave blank to randomize the seed

imageuri

Input image for style transfer

widthinteger

Output image width. Will be rounded to nearest 64

Default
768
Minimum
256
Maximum
2048
heightinteger

Output image height. Will be rounded to nearest 64

Default
768
Minimum
256
Maximum
2048
promptstring

Input prompts, separated by newlines

style_promptstring

Input style prompt. If using an image for style transfer, this should describe the style of that image

image_subjectstring

Subject (not style) of image used for style transfer. Ignored if image not set.

Default
"None"
guidance_scalenumber

Guidance scale

Default
7
Minimum
1.1
Maximum
20
negative_promptstring

Input negative prompt

Default
"low-resolution"
shared_score_scalenumber

Shared score scale. Higher value induces higher, set 1 for no rescale

Default
1
Maximum
1
shared_score_shiftnumber

Shared score shift (Actual value will take log of this). Higher value induces higher fidelity, set 1 for no shift

Default
2
Minimum
0.01
Maximum
10
num_inference_stepsinteger

Number of inference steps

Default
50
Minimum
10
Maximum
150

Output schema

Type
uri[]