make it possible to use checkpoints of different types (SD1, SDXL) in…

… first and second pass of hires fix

modules/processing.py

@@ -1060,16 +1060,21 @@ def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subs
         if not self.enable_hr:
             return samples
 
+        if self.latent_scale_mode is None:
+            decoded_samples = decode_first_stage(self.sd_model, samples)
+        else:
+            decoded_samples = None
+
         current = shared.sd_model.sd_checkpoint_info
         try:
             if self.hr_checkpoint_info is not None:
                 sd_models.reload_model_weights(info=self.hr_checkpoint_info)
 
-            return self.sample_hr_pass(samples, seeds, subseeds, subseed_strength, prompts)
+            return self.sample_hr_pass(samples, decoded_samples, seeds, subseeds, subseed_strength, prompts)
         finally:
             sd_models.reload_model_weights(info=current)
 
-    def sample_hr_pass(self, samples, seeds, subseeds, subseed_strength, prompts):
+    def sample_hr_pass(self, samples, decoded_samples, seeds, subseeds, subseed_strength, prompts):
         self.is_hr_pass = True
 
         target_width = self.hr_upscale_to_x
@@ -1100,7 +1105,6 @@ def save_intermediate(image, index):
             else:
                 image_conditioning = self.txt2img_image_conditioning(samples)
         else:
-            decoded_samples = decode_first_stage(self.sd_model, samples)
             lowres_samples = torch.clamp((decoded_samples + 1.0) / 2.0, min=0.0, max=1.0)
 
             batch_images = []