Merge branch 'development' into dev/diffusers

Author: keturn

ldm/invoke/args.py

@@ -250,8 +250,6 @@ def dream_prompt_str(self,**kwargs):
             switches.append('--seamless')
         if a['hires_fix']:
             switches.append('--hires_fix')
-            if a['strength'] and a['strength']>0:
-                switches.append(f'-f {a["strength"]}')
 
         # img2img generations have parameters relevant only to them and have special handling
         if a['init_img'] and len(a['init_img'])>0:

ldm/invoke/generator/txt2img2img.py

@@ -6,8 +6,6 @@
 
 import torch
 from PIL import Image
-from ldm.invoke.devices import choose_autocast
-from ldm.invoke.image_util import InitImageResizer
 
 from ldm.invoke.generator.base import Generator
 from ldm.invoke.generator.omnibus import Omnibus
@@ -47,13 +45,16 @@ def make_image(x_T):
                     ddim_num_steps=steps, ddim_eta=ddim_eta, verbose=False
             )
 
+            #x = self.get_noise(init_width, init_height)
+            x = x_T
+
             if self.free_gpu_mem and self.model.model.device != self.model.device:
                 self.model.model.to(self.model.device)
 
             samples, _ = sampler.sample(
                 batch_size                   = 1,
                 S                            = steps,
-                x_T                          = x_T,
+                x_T                          = x,
                 conditioning                 = c,
                 shape                        = shape,
                 verbose                      = False,
@@ -69,21 +70,11 @@ def make_image(x_T):
                  )
 
             # resizing
-
-            image = self.sample_to_image(samples)
-            image = InitImageResizer(image).resize(width, height)
-
-            image = np.array(image).astype(np.float32) / 255.0
-            image = image[None].transpose(0, 3, 1, 2)
-            image = torch.from_numpy(image)
-            image = 2.0 * image - 1.0
-            image = image.to(self.model.device)
-
-            scope = choose_autocast(self.precision)
-            with scope(self.model.device.type):
-                samples = self.model.get_first_stage_encoding(
-                    self.model.encode_first_stage(image)
-                ) # move back to latent space
+            samples = torch.nn.functional.interpolate(
+                samples,
+                size=(height // self.downsampling_factor, width // self.downsampling_factor),
+                mode="bilinear"
+            )
 
             t_enc = int(strength * steps)
             ddim_sampler = DDIMSampler(self.model, device=self.model.device)