improvements

See:
AUTOMATIC1111#10649 (comment)

modules/generation_parameters_copypaste.py

@@ -306,6 +306,18 @@ def parse_generation_parameters(x: str):
     if "RNG" not in res:
         res["RNG"] = "GPU"
 
+    if "KDiff Sched Type" not in res:
+        res["KDiff Sched Type"] = "Automatic"
+
+    if "KDiff Sched max sigma" not in res:
+        res["KDiff Sched max sigma"] = 14.6
+
+    if "KDiff Sched min sigma" not in res:
+        res["KDiff Sched min sigma"] = 0.3
+
+    if "KDiff Sched rho" not in res:
+        res["KDiff Sched rho"] = 7.0
+
     return res
 
 
@@ -318,10 +330,10 @@ def parse_generation_parameters(x: str):
     ('Conditional mask weight', 'inpainting_mask_weight'),
     ('Model hash', 'sd_model_checkpoint'),
     ('ENSD', 'eta_noise_seed_delta'),
-    ('KDiffusion Scheduler Type', 'k_sched_type'),
-    ('KDiffusion Scheduler sigma_max', 'sigma_max'),
-    ('KDiffusion Scheduler sigma_min', 'sigma_min'),
-    ('KDiffusion Scheduler rho', 'rho'),
+    ('KDiff Sched Type', 'k_sched_type'),
+    ('KDiff Sched max sigma', 'sigma_max'),
+    ('KDiff Sched min sigma', 'sigma_min'),
+    ('KDiff Sched rho', 'rho'),
     ('Noise multiplier', 'initial_noise_multiplier'),
     ('Eta', 'eta_ancestral'),
     ('Eta DDIM', 'eta_ddim'),

modules/sd_samplers_kdiffusion.py

@@ -296,12 +296,6 @@ def initialize(self, p):
 
         k_diffusion.sampling.torch = TorchHijack(self.sampler_noises if self.sampler_noises is not None else [])
 
-        if opts.k_sched_type != "Automatic":
-            p.extra_generation_params["KDiffusion Scheduler Type"] = opts.k_sched_type
-            p.extra_generation_params["KDiffusion Scheduler sigma_max"] = opts.sigma_max
-            p.extra_generation_params["KDiffusion Scheduler sigma_min"] = opts.sigma_min
-            p.extra_generation_params["KDiffusion Scheduler rho"] = opts.rho
-
         extra_params_kwargs = {}
         for param_name in self.extra_params:
             if hasattr(p, param_name) and param_name in inspect.signature(self.func).parameters:
@@ -326,14 +320,27 @@ def get_sigmas(self, p, steps):
         if p.sampler_noise_scheduler_override:
             sigmas = p.sampler_noise_scheduler_override(steps)
         elif opts.k_sched_type != "Automatic":
-            sigma_min, sigma_max = (0.1, 10) if opts.use_old_karras_scheduler_sigmas else (self.model_wrap.sigmas[0].item(), self.model_wrap.sigmas[-1].item())
-            sigmas_func = k_diffusion_scheduler[opts.k_sched_type]
+            m_sigma_min, m_sigma_max = (self.model_wrap.sigmas[0].item(), self.model_wrap.sigmas[-1].item())
+            sigma_min, sigma_max = (0.1, 10)
             sigmas_kwargs = {
-                'sigma_min': opts.sigma_min or sigma_min,
-                'sigma_max': opts.sigma_max or sigma_max
+                'sigma_min': sigma_min if opts.use_old_karras_scheduler_sigmas else m_sigma_min,
+                'sigma_max': sigma_max if opts.use_old_karras_scheduler_sigmas else m_sigma_max
             }
+
+            sigmas_func = k_diffusion_scheduler[opts.k_sched_type]
+            p.extra_generation_params["KDiff Sched Type"] = opts.k_sched_type
+
+            if opts.sigma_min != 0.3:
+                # take 0.0 as model default
+                sigmas_kwargs['sigma_min'] = opts.sigma_min or m_sigma_min
+                p.extra_generation_params["KDiff Sched min sigma"] = opts.sigma_min
+            if opts.sigma_max != 14.6:
+                sigmas_kwargs['sigma_max'] = opts.sigma_max or m_sigma_max
+                p.extra_generation_params["KDiff Sched max sigma"] = opts.sigma_max
             if opts.k_sched_type != 'exponential':
                 sigmas_kwargs['rho'] = opts.rho
+                p.extra_generation_params["KDiff Sched rho"] = opts.rho
+
             sigmas = sigmas_func(n=steps, **sigmas_kwargs, device=shared.device)
         elif self.config is not None and self.config.options.get('scheduler', None) == 'karras':
             sigma_min, sigma_max = (0.1, 10) if opts.use_old_karras_scheduler_sigmas else (self.model_wrap.sigmas[0].item(), self.model_wrap.sigmas[-1].item())

modules/shared.py

@@ -523,8 +523,8 @@ def list_samplers():
     's_tmin':  OptionInfo(0.0, "sigma tmin",  gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}),
     's_noise': OptionInfo(1.0, "sigma noise", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}),
     'k_sched_type':  OptionInfo("Automatic", "scheduler type", gr.Dropdown, {"choices": ["Automatic", "karras", "exponential", "polyexponential"]}),
-    'sigma_max': OptionInfo(0.0, "sigma max", gr.Number).info("the maximum noise strength for the scheduler. Set to 0 to use the same value which 'xxx karras' samplers use."),
-    'sigma_min': OptionInfo(0.0, "sigma min", gr.Number).info("the minimum noise strength for the scheduler. Set to 0 to use the same value which 'xxx karras' samplers use."),
+    'sigma_max': OptionInfo(14.6, "sigma max", gr.Number).info("the maximum noise strength for the scheduler. Set to 0 to use the same value which 'xxx karras' samplers use."),
+    'sigma_min': OptionInfo(0.3, "sigma min", gr.Number).info("the minimum noise strength for the scheduler. Set to 0 to use the same value which 'xxx karras' samplers use."),
     'rho':  OptionInfo(7.0, "rho", gr.Number).info("higher will make a more steep noise scheduler (decrease faster). default for karras is 7.0, for polyexponential is 1.0"),
     'eta_noise_seed_delta': OptionInfo(0, "Eta noise seed delta", gr.Number, {"precision": 0}).info("ENSD; does not improve anything, just produces different results for ancestral samplers - only useful for reproducing images"),
     'always_discard_next_to_last_sigma': OptionInfo(False, "Always discard next-to-last sigma").link("PR", "https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/6044"),

scripts/xyz_grid.py

@@ -220,10 +220,10 @@ def __init__(self, *args, **kwargs):
     AxisOption("Sigma min", float, apply_field("s_tmin")),
     AxisOption("Sigma max", float, apply_field("s_tmax")),
     AxisOption("Sigma noise", float, apply_field("s_noise")),
-    AxisOption("KDiffusion Scheduler Type", str, apply_override("k_sched_type"), choices=lambda: list(sd_samplers_kdiffusion.k_diffusion_scheduler)),
-    AxisOption("KDiffusion Scheduler Sigma Min", float, apply_override("sigma_min")),
-    AxisOption("KDiffusion Scheduler Sigma Max", float, apply_override("sigma_max")),
-    AxisOption("KDiffusion Scheduler rho", float, apply_override("rho")),
+    AxisOption("KDiff Sched Type", str, apply_override("k_sched_type"), choices=lambda: list(sd_samplers_kdiffusion.k_diffusion_scheduler)),
+    AxisOption("KDiff Sched min sigma", float, apply_override("sigma_min")),
+    AxisOption("KDiff Sched max sigma", float, apply_override("sigma_max")),
+    AxisOption("KDiff Sched rho", float, apply_override("rho")),
     AxisOption("Eta", float, apply_field("eta")),
     AxisOption("Clip skip", int, apply_clip_skip),
     AxisOption("Denoising", float, apply_field("denoising_strength")),