stability-ai/sdxl
SDXL consists of an ensemble of experts pipeline for latent diffusion: In a first step, the base model is used to generate (noisy) latents, which are then further processed with a refinement model (available here: https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/) specialized for the final denoising steps. Note that the base model can be used as a standalone module.
SDXL consists of an ensemble of experts pipeline for latent diffusion: In a first step, the base model is used to generate (noisy) latents, which are then further processed with a refinement model (available here: https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/) specialized for the final denoising steps. Note that the base model can be used as a standalone module.
HTTP/cURL API
Input fields
schedulerstring
scheduler
Default value: DDIM
Allowed values: DDIMDPMSolverMultistepHeunDiscreteKarrasDPMK_EULER_ANCESTRALK_EULERPNDM
num_inference_stepsinteger
Number of denoising steps
Default value: 50
Range: 1 ≤ num_inference_steps ≤ 500
guidance_scalenumber
Scale for classifier-free guidance
Default value: 7.5
Range: 1 ≤ guidance_scale ≤ 50
prompt_strengthnumber
Prompt strength when using img2img / inpaint. 1.0 corresponds to full destruction of information in image
Default value: 0.8
Range: 0 ≤ prompt_strength ≤ 1
refinestring
Which refine style to use
Default value: no_refiner
Allowed values: no_refinerexpert_ensemble_refinerbase_image_refiner
high_noise_fracnumber
for expert_ensemble_refiner, the fraction of noise to use
Default value: 0.8
Range: 0 ≤ high_noise_frac ≤ 1
refine_stepsinteger
for base_image_refiner, the number of steps to refine, defaults to num_inference_steps
apply_watermarkboolean
Applies a watermark to enable determining if an image is generated in downstream applications. If you have other provisions for generating or deploying images safely, you can use this to disable watermarking.
Default value: true
