Best Scheduler + Sampler Combos

Sampler and scheduler choices change how the image is denoised. Most combinations can run, but only a smaller set is consistently useful for quality, speed, or specific model families. Use these combinations as reliable starting points. If an image looks noisy, flat, or over-smoothed, try one of the pairings below before changing the prompt.

Quick Picks

Goal	Sampler	Scheduler	Notes
Sharp general output	`euler`	`normal`	Fast, predictable, and a good baseline.
Balanced high quality	`dpmpp_2m`	`karras`	Strong default for SDXL-style work.
Realistic depth and shading	`dpmpp_2m_sde`	`karras`	Good for portraits, materials, and lighting.
Complex SDXL scenes	`dpmpp_3m_sde`	`linear_quadratic`	Better for rich scenes and gradients.
Fast low-step generation	`lcm`	`sgm_uniform`	Best when speed matters more than maximum detail.
Clean progressive output	`deis`	`simple`	Useful for stable text-to-image results.
Controlled diffusion steps	`ipndm`	`ddim_uniform`	Good compromise when you want tighter noise control.
Sequential or turbo workflows	`res_multistep`	`karras`	Useful for Z-Image style or sequential inference workflows.
Anime or 3D-looking renders	`er_sde`	`exponential`	Can produce smooth depth and stylized volume.
High-resolution adaptive output	`uni_pc`	`kl_optimal`	Best when the model supports it well.

Model Starting Points

Model family	Recommended starter
Stable Diffusion 1.5	`ddim` + `normal`, or `euler` + `normal` for sharper output
Flux1 Dev	`ddim` + `sgm_uniform`
Flux2 Klein	`euler` + `flux2`; try `euler` + `beta` for an alternate scheduler
Anima	`er_sde` + `simple`; try `er_sde` + `exponential` for more dimensional renders
Qwen Image	`euler` + `simple`
Z-Image	`res_multistep` + `simple`; try `res_multistep` + `karras` for a different motion/detail balance
SDXL	`dpmpp_2m` + `karras`, or `dpmpp_3m_sde` + `linear_quadratic` for complex scenes
Pony Diffusion	`dpmpp_2m` + `karras`, or `euler` + `normal` for a simpler baseline

Detailed Pairing List

Sampler	Best scheduler	When to use it
`euler`	`normal`	Fast, sharp, reliable baseline.
`euler_cfg_pp`	`karras`	CFG-heavy workflows that need better detail retention.
`euler_ancestral`	`exponential`	Softer images, dreamy lighting, gradual transitions.
`dpmpp_2m`	`karras`	Balanced quality and stability.
`dpmpp_2m_sde`	`karras`	Realism, depth, and detailed lighting.
`dpmpp_3m_sde`	`linear_quadratic`	Complex SDXL scenes and smooth gradients.
`heunpp2`	`karras`	Intricate prompts and cleaner token-weight transitions.
`lcm`	`sgm_uniform`	Fast low-step generation.
`uni_pc`	`kl_optimal`	Adaptive high-resolution workflows.
`deis`	`simple`	Clean, progressive text-to-image output.
`ipndm`	`ddim_uniform`	Noise-controlled diffusion steps.
`res_multistep`	`karras`	Sequential inference and animation-like workflows.
`er_sde`	`exponential`	Smooth depth, stylized renders, and SDXL variants.

Avoid These Unless You Are Testing

Avoid lcm with exponential, kl_optimal, or linear_quadratic; it is designed for fast, low-step generation.
Avoid uni_pc with simple or ddim_uniform if the result looks flat.
Avoid GPU/SDE samplers with high-noise schedulers unless you intentionally want unstable or experimental output.
Avoid CFG-specific samplers unless your workflow is built for that conditioning style.

​Best Scheduler + Sampler Combos

​Quick Picks

​Model Starting Points

​Detailed Pairing List

​Avoid These Unless You Are Testing

​References

Best Scheduler + Sampler Combos

Quick Picks

Model Starting Points

Detailed Pairing List

Avoid These Unless You Are Testing

References