fix(samplers): honor fallback setting in remaining sampler fallback paths

create_sampler restored the model default scheduler on a prediction-type
mismatch, on an unknown sampler config, and on any scheduler-constructor
exception regardless of schedulers_fallback; only SD_SAMPLER_DEBUG could turn
the prediction mismatch into an error. Raise like the other capability gates
when the fallback setting is disabled.

An unresolved sampler name substituted UniPC before any of those gates could
run; pass the requested name through instead, so it falls back to the model
default (or raises when fallback is disabled) and the infotext records
Default rather than the unresolved name. find_sampler now also resolves an
unspecified sampler to Default instead of UniPC, matching the platform
default used everywhere else.

modules/processing_helpers.py

@@ -584,9 +584,9 @@ def update_sampler(p, sd_model, second_pass=False):
         if sampler_selection == 'None':
             return
         sampler = sd_samplers.find_sampler(sampler_selection)
-        if sampler is None:
-            log.warning(f'Sampler: "{sampler_selection}" not found')
-            sampler = sd_samplers.all_samplers_map.get("UniPC")
+        resolved = sampler is not None
+        if not resolved:
+            log.warning(f'Sampler: name="{sampler_selection}" not found')
         sched_override_keys = [
             'schedulers_prediction_type', 'schedulers_beta_schedule', 'schedulers_timesteps',
             'schedulers_sigma', 'schedulers_use_thresholding', 'schedulers_use_loworder',
@@ -596,8 +596,8 @@ def update_sampler(p, sd_model, second_pass=False):
             'schedulers_timestep_spacing', 'schedulers_timesteps_range',
         ]
         scheduler_overrides = {k: getattr(p, k) for k in sched_override_keys if getattr(p, k, None) is not None}
-        sampler = sd_samplers.create_sampler(sampler.name, sd_model, scheduler_overrides=scheduler_overrides)
-        if sampler is None or sampler_selection == 'Default':
+        sampler = sd_samplers.create_sampler(sampler.name if resolved else sampler_selection, sd_model, scheduler_overrides=scheduler_overrides)
+        if sampler is None or not resolved or sampler_selection == 'Default':
             if second_pass:
                 p.hr_sampler = 'Default'
             else:

modules/sd_samplers.py

@@ -15,7 +15,7 @@ loaded_config = None
 
 def find_sampler(name:str):
     if name is None or name == 'None':
-        return all_samplers_map.get("UniPC", None)
+        return all_samplers_map.get("Default", None)
     for sampler in all_samplers:
         if sampler.name.lower() == name.lower() or name in sampler.aliases:
             return sampler
@@ -107,6 +107,8 @@ def create_sampler(name, model, scheduler_overrides=None):
         config = find_sampler_config(name)
 
     if config is None or config.constructor is None:
+        if debug or not shared.opts.schedulers_fallback:
+            raise ValueError(f'Sampler: name="{name}" unknown')
         return restore_default(model, name)
 
     from modules import sd_samplers_diffusers
@@ -126,16 +128,16 @@ def create_sampler(name, model, scheduler_overrides=None):
         pass
     elif (model is not None) and (is_flow and not requires_flow):
         log.error(f'Sampler: "{sampler.name}" cls={sampler.sampler.__class__.__name__} pipe={model.__class__.__name__} type={pred_type} model requires sampler with discrete prediction')
-        if not debug:
-            return restore_default(model, name)
-        else:
+        if debug or not shared.opts.schedulers_fallback:
             raise ValueError(f'Sampler: name="{sampler.name}" cls={sampler.sampler.__class__.__name__} type={pred_type} model requires sampler with discrete prediction')
+        else:
+            return restore_default(model, name)
     elif (model is not None) and (not is_flow and requires_flow):
         log.error(f'Sampler: "{sampler.name}" cls={sampler.sampler.__class__.__name__} pipe={model.__class__.__name__} type={pred_type} model requires sampler with flow prediction')
-        if not debug:
-            return restore_default(model, name)
-        else:
+        if debug or not shared.opts.schedulers_fallback:
             raise ValueError(f'Sampler: name="{sampler.name}" cls={sampler.sampler.__class__.__name__} type={pred_type} model requires sampler with flow prediction')
+        else:
+            return restore_default(model, name)
 
     # assign sampler
     if model is not None:

modules/sd_samplers_diffusers.py

@@ -519,6 +519,8 @@ class DiffusionSampler:
             log.error(f'Sampler: "{name}" {e}')
             if debug:
                 errors.display(e, 'Samplers')
+            if debug or not shared.opts.schedulers_fallback:
+                raise
             self.sampler = None
             return