Schedulers
Discover the various schedulers available for image generation.
What schedulers do
Schedulers are algorithms that control the progression of the denoising process in generative models. When generating an image, the model starts with random noise and progressively refines this noise into a coherent image. The scheduler dictates how the noise reduction happens over each iteration, impacting the quality, speed, and style of the generated image.
Although the scheduler parameter is optional (the model's own scheduler is used by default), you can specify a different scheduler from the following tables.
Supported schedulers for FLUX
| Scheduler | Friendly name | Description |
|---|---|---|
| Euler | Euler | Basic Euler method implementation. |
| FlowMatchEulerDiscreteScheduler | FlowMatch Euler | Flow matching techniques combined with Euler's method for precision. |
| DPM++ | DPM++ | Basic implementation of the DPM++ sampling method. |
| DPM++ SDE | DPM++ SDE | Stochastic differential equations-based scheduler for high-quality images. |
| DPM++ 2M | DPM++ 2M | Enhanced DPM++ with a second-order multistep method. |
| DPM++ 2M SDE | DPM++ 2M SDE | Second-order scheduler with SDE for robust image generation. |
| DPM++ 3M | DPM++ 3M | Advanced DPM++ with a third-order multistep method. |
| Euler Beta | Euler Beta | Euler method with Beta noise scheduling. |
| Euler Exponential | Euler Exponential | Euler method with an exponential noise scheduling strategy. |
| Euler Karras | Euler Karras | Euler method with Karras noise scheduling. |
| DPM++ Beta | DPM++ Beta | DPM++ scheduler with Beta noise scheduling. |
| DPM++ Exponential | DPM++ Exponential | DPM++ scheduler with an exponential noise scheduling method. |
| DPM++ Karras | DPM++ Karras | DPM++ scheduler with Karras noise scheduling. |
| DPM++ SDE Beta | DPM++ SDE Beta | Stochastic differential equations-based scheduler with Beta noise scheduling. |
| DPM++ SDE Exponential | DPM++ SDE Exponential | Stochastic differential equations-based scheduler with exponential noise scheduling. |
| DPM++ SDE Karras | DPM++ SDE Karras | Stochastic differential equations-based scheduler with Karras noise scheduling. |
| DPM++ 2M Beta | DPM++ 2M Beta | Second-order multistep DPM++ scheduler with Beta noise scheduling. |
| DPM++ 2M Exponential | DPM++ 2M Exponential | Second-order multistep DPM++ scheduler with exponential noise scheduling. |
| DPM++ 2M Karras | DPM++ 2M Karras | Second-order multistep DPM++ scheduler with Karras noise scheduling. |
| DPM++ 2M SDE Beta | DPM++ 2M SDE Beta | Second-order multistep SDE scheduler with Beta noise scheduling. |
| DPM++ 2M SDE Exponential | DPM++ 2M SDE Exponential | Second-order multistep SDE scheduler with exponential noise scheduling. |
| DPM++ 2M SDE Karras | DPM++ 2M SDE Karras | Second-order multistep SDE scheduler with Karras noise scheduling. |
| DPM++ 3M Beta | DPM++ 3M Beta | Third-order multistep DPM++ scheduler with Beta noise scheduling. |
| DPM++ 3M Exponential | DPM++ 3M Exponential | Third-order multistep DPM++ scheduler with exponential noise scheduling. |
| DPM++ 3M Karras | DPM++ 3M Karras | Third-order multistep DPM++ scheduler with Karras noise scheduling. |
Supported schedulers for Stable Diffusion
| Scheduler | Friendly name | Description |
|---|---|---|
| Default | Depends on model | The default scheduler used by the Stable Diffusion model. |
| DDIM | DDIM | Direct implementation of DDIM sampling method. |
| DDIMScheduler | DDIM | Accelerates denoising with fewer steps, maintaining high quality. |
| DDPMScheduler | DDPM | Standard scheduler following the original diffusion process. |
| DEISMultistepScheduler | DEIS Multistep | Efficient denoising with multiple steps for improved quality. |
| DPMSolverSinglestepScheduler | DPM-Solver Single-step | Single-step solver for direct denoising efficiency. |
| DPMSolverMultistepScheduler | DPM-Solver Multi-step | Multi-step solver enhancing denoising accuracy and stability. |
| DPMSolverMultistepInverse | DPM-Solver Multi-step Inverse | Inverse multi-step solver for reverse processes. |
| DPM++ | DPM++ | Basic implementation of DPM++ sampling method. |
| DPM++ Karras | DPM++ Karras | DPM++ scheduler with Karras noise scheduling. |
| DPM++ 2M | DPM++ 2M | Enhanced DPM++ with 2nd-order multistep method. |
| DPM++ 2M Karras | DPM++ 2M Karras | Advanced Karras scheduler with second-order method. |
| DPM++ 2M SDE Karras | DPM++ 2M SDE Karras | Combines second-order and SDE techniques for superior denoising. |
| DPM++ 2M SDE | DPM++ 2M SDE | Second-order scheduler with SDE for robust image generation. |
| DPM++ 3M | DPM++ 3M | Advanced DPM++ with 3rd-order multistep method. |
| DPM++ 3M Karras | DPM++ 3M Karras | Third-order DPM++ with Karras noise scheduling. |
| DPM++ SDE Karras | DPM++ SDE Karras | Karras method integrated with SDE for optimal denoising. |
| DPM++ SDE | DPM++ SDE | Stochastic differential equations-based scheduler for high-quality images. |
| EDMEulerScheduler | EDM Euler | Enhanced denoising with modifications to the Euler method. |
| EDMDPMSolverMultistepScheduler | EDM DPM-Solver Multi-step | EDM-enhanced multi-step solver for precision and quality. |
| Euler | Euler | Basic Euler method implementation. |
| EulerDiscreteScheduler | Euler | Utilizes Euler's method for simple and effective denoising. |
| Euler Karras | Euler Karras | Euler method with Karras noise scheduling. |
| Euler a | Euler Ancestral | Euler method with ancestral sampling variant. |
| EulerAncestralDiscreteScheduler | Euler Ancestral | Euler's method with ancestral sampling for better quality. |
| FlowMatchEulerDiscreteScheduler | FlowMatch Euler | Flow matching techniques combined with Euler's method for precision. |
| Heun | Heun | Basic implementation of Heun's method. |
| HeunDiscreteScheduler | Heun | Uses Heun's method for precise and efficient denoising. |
| Heun Karras | Heun Karras | Heun's method with Karras noise scheduling. |
| IPNDMScheduler | IPNDM | Improved pseudo numerical methods for efficient diffusion. |
| KDPM2DiscreteScheduler | KDPM2 | Second-order scheduler for refined noise reduction. |
| KDPM2AncestralDiscreteScheduler | KDPM2 Ancestral | Second-order scheduler incorporating ancestral sampling. |
| LCM | LCM | Basic implementation of Low-Complexity Method. |
| LCMScheduler | LCM | Low-complexity scheduler designed for efficient denoising with minimal computational resources. |
| LMS | LMS | Basic Linear Multi-Step method implementation. |
| LMSDiscreteScheduler | LMS | Linear multistep method for balanced speed and quality. |
| LMS Karras | LMS Karras | Linear Multi-Step method with Karras noise scheduling. |
| PNDMScheduler | PNDM | Combines numerical methods with diffusion for effective denoising. |
| TCDScheduler | TCD | Time-continuous diffusion scheduler for smooth denoising processes. |
| UniPC | UniPC | Basic Universal PC scheduler implementation. |
| UniPCMultistepScheduler | UniPC Multistep | Universal multi-step scheduler for diverse applications. |
| UniPC Karras | UniPC Karras | Universal PC scheduler with Karras noise scheduling. |
| UniPC 2M | UniPC 2M | Second-order Universal PC scheduler. |
| UniPC 2M Karras | UniPC 2M Karras | Second-order Universal PC with Karras scheduling. |
| UniPC 3M | UniPC 3M | Third-order Universal PC scheduler. |
| UniPC 3M Karras | UniPC 3M Karras | Third-order Universal PC with Karras scheduling. |
Extended DPM++, IPNDM, and UniPC schedulers
Recent updates introduced new "Uniform" variants for the DPM++, IPNDM, and UniPC families. These schedulers use different noise weighting strategies (such as Uniform, Beta, Exponential, and Karras) to fine-tune denoising behavior and control image sharpness, contrast, and smoothness.
| Scheduler | Friendly name | Description |
|---|---|---|
| DPM++ Uniform | DPM++ Uniform | Standard DPM++ scheduler with uniform noise weighting. |
| DPM++ Uniform Beta | DPM++ Uniform Beta | Applies a Beta noise distribution for smoother tonal gradients. |
| DPM++ Uniform Exponential | DPM++ Uniform Exponential | Uses an exponential noise curve for faster denoising convergence. |
| DPM++ Uniform Karras | DPM++ Uniform Karras | Karras-style noise scaling optimized for high-frequency details. |
| DPM++ 2M Uniform | DPM++ 2M Uniform | Second-order multistep variant with uniform noise weighting. |
| DPM++ 2M SDE Uniform | DPM++ 2M SDE Uniform | Second-order SDE version using uniform noise weighting. |
| DPM++ 3M Uniform | DPM++ 3M Uniform | Third-order multistep version with uniform noise weighting. |
| DPM++ 3M SDE Uniform | DPM++ 3M SDE Uniform | Third-order SDE scheduler with uniform noise weighting. |
| IPNDM Uniform | IPNDM Uniform | Improved pseudo numerical diffusion method with uniform noise weighting. |
| IPNDM Uniform Beta | IPNDM Uniform Beta | Uses a Beta-based noise profile for stable, balanced generation. |
| IPNDM Uniform Exponential | IPNDM Uniform Exponential | Applies an exponential decay to noise scheduling. |
| IPNDM Uniform Karras | IPNDM Uniform Karras | Karras noise scaling with IPNDM’s efficient denoising core. |
| UniPC Uniform | UniPC Uniform | Universal predictor-corrector scheduler using uniform noise weighting. |
| UniPC 2M Uniform | UniPC 2M Uniform | Second-order variant for faster convergence and smooth results. |
| UniPC 3M Uniform | UniPC 3M Uniform | Third-order variant with uniform weighting for high-quality images. |
| UniPC Uniform Beta | UniPC Uniform Beta | Applies Beta weighting to balance detail and noise control. |
| UniPC Uniform Exponential | UniPC Uniform Exponential | Emphasizes early denoising with exponential noise decay. |
| UniPC Uniform Karras | UniPC Uniform Karras | Employs Karras noise scaling for crisp edges and fine textures. |