Change the default SDXL sampler to Euler A + Linear and set Guidence scale value to 4

configs/sdxl/scheduler/scheduler_config.json

@@ -1,18 +1,19 @@
 {
-  "_class_name": "EulerDiscreteScheduler",
-  "_diffusers_version": "0.19.0.dev0",
+  "_class_name": "EulerAncestralDiscreteScheduler",
+  "_diffusers_version": "0.35.1",
   "beta_end": 0.012,
-  "beta_schedule": "scaled_linear",
+  "beta_schedule": "linear",
   "beta_start": 0.00085,
   "clip_sample": false,
   "interpolation_type": "linear",
   "num_train_timesteps": 1000,
   "prediction_type": "epsilon",
+  "rescale_betas_zero_snr": false,
   "sample_max_value": 1.0,
   "set_alpha_to_one": false,
   "skip_prk_steps": true,
   "steps_offset": 1,
-  "timestep_spacing": "leading",
+  "timestep_spacing": "trailing",
   "trained_betas": null,
   "use_karras_sigmas": false
 }

installer.py

@@ -1287,7 +1287,7 @@ def install_optional():
     install('hf_xet', ignore=True, quiet=True)
     install('nvidia-ml-py', ignore=True, quiet=True)
     install('optimum-quanto==0.2.7', ignore=True, quiet=True)
-    install('pillow-jxl-plugin==1.3.4', ignore=True, quiet=True)
+    install('pillow-jxl-plugin==1.3.5', ignore=True, quiet=True)
     install('torchao==0.10.0', ignore=True, quiet=True)
     install('ultralytics==8.3.40', ignore=True, quiet=True)
     try:

modules/sd_models.py

@@ -567,18 +567,30 @@ def load_sdnq_model(checkpoint_info, pipeline, diffusers_load_config, op):
     return sd_model
 
 
-def set_overrides(sd_model, checkpoint_info):
+def set_overrides(sd_model, checkpoint_info, model_type):
     checkpoint_info_name = checkpoint_info.name.lower()
+    if "Kandinsky" in sd_model.__class__.__name__:
+        sd_model.scheduler.name = 'DDIM'
+    elif (
+        checkpoint_info.path.lower().endswith('.safetensors')
+        and model_type.startswith("Stable Diffusion") and model_type != "Stable Diffusion 3"
+    ): # SDXL and SD 1.5
+        scheduler_config = sd_model.scheduler.config
+        scheduler_config['beta_schedule'] = 'linear'
+        scheduler_config['timestep_spacing'] = 'trailing'
+        sd_model.scheduler = diffusers.EulerAncestralDiscreteScheduler.from_config(scheduler_config)
+
     if 'bigaspv25' in checkpoint_info_name or ('flow' in checkpoint_info_name and 'flower' not in checkpoint_info_name):
         scheduler_config = sd_model.scheduler.config
         scheduler_config['prediction_type'] = 'flow_prediction'
-        scheduler_config['use_flow_sigmas'] = True
         scheduler_config['beta_schedule'] = 'linear'
+        scheduler_config['use_flow_sigmas'] = True
         sd_model.scheduler = diffusers.UniPCMultistepScheduler.from_config(scheduler_config)
         shared.log.info(f'Setting override: model="{checkpoint_info.name}" component=scheduler prediction="flow-prediction"')
     elif 'vpred' in checkpoint_info_name or 'v-pred' in checkpoint_info_name or 'v_pred' in checkpoint_info_name:
         scheduler_config = sd_model.scheduler.config
         scheduler_config['prediction_type'] = 'v_prediction'
+        scheduler_config['beta_schedule'] = 'scaled_linear'
         scheduler_config['rescale_betas_zero_snr'] = True
         sd_model.scheduler = diffusers.EulerDiscreteScheduler.from_config(scheduler_config)
         shared.log.info(f'Setting override: model="{checkpoint_info.name}" component=scheduler prediction="v-prediction" rescale=True')
@@ -590,6 +602,7 @@ def set_overrides(sd_model, checkpoint_info):
             if 'v_pred' in keys: # NoobAI VPred models added empty v_pred and ztsnr keys
                 scheduler_config = sd_model.scheduler.config
                 scheduler_config['prediction_type'] = 'v_prediction'
+                scheduler_config['beta_schedule'] = 'scaled_linear'
                 if 'ztsnr' in keys:
                     scheduler_config['rescale_betas_zero_snr'] = True
                 sd_model.scheduler = diffusers.EulerDiscreteScheduler.from_config(scheduler_config)
@@ -713,12 +726,9 @@ def load_diffuser(checkpoint_info=None, op='model', revision=None): # pylint: di
             shared.log.error(f'Load {op}: name="{checkpoint_info.name if checkpoint_info is not None else None}" not loaded')
             return
 
-        set_overrides(sd_model, checkpoint_info)
+        set_overrides(sd_model, checkpoint_info, model_type)
         set_defaults(sd_model, checkpoint_info)
 
-        if "Kandinsky" in sd_model.__class__.__name__: # need a special case
-            sd_model.scheduler.name = 'DDIM'
-
         if hasattr(sd_model, "unet") and model_type not in ['Stable Cascade']: # others calls load_diffuser again
             sd_unet.load_unet(sd_model, checkpoint_info.path)
 

modules/ui_guidance.py

@@ -15,7 +15,7 @@ def create_guidance_inputs(tab):
                 guidance_btn = ui_components.ToolButton(value=ui_symbols.book, elem_id=f"{tab}_guider_docs")
                 guidance_btn.click(fn=None, _js='getGuidanceDocs', inputs=[guidance_name], outputs=[])
             with gr.Row(visible=shared.opts.model_modular_enable):
-                guidance_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.1, label='_Guidance scale', value=6.0, elem_id=f"{tab}_guidance_scale")
+                guidance_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.1, label='_Guidance scale', value=4.0, elem_id=f"{tab}_guidance_scale")
                 guidance_rescale = gr.Slider(minimum=0.0, maximum=1.0, step=0.05, label='_Guidance rescale', value=0.0, elem_id=f"{tab}_guidance_rescale")
             with gr.Row(visible=shared.opts.model_modular_enable):
                 guidance_start = gr.Slider(minimum=0.0, maximum=1.0, step=0.05, label='_Guidance start', value=0.0, elem_id=f"{tab}_guidance_start")