Fix txt2img + control adapters

nod-ai · Dec 7, 2023 · 3b892bd · 3b892bd
1 parent 500636d
commit 3b892bd
Showing 1 changed file with 49 additions and 13 deletions.
diff --git a/apps/stable_diffusion/src/pipelines/pipeline_shark_stable_diffusion_stencil.py b/apps/stable_diffusion/src/pipelines/pipeline_shark_stable_diffusion_stencil.py
@@ -125,6 +125,31 @@ def unload_controlnet_512(self, index):
         self.controlnet_512_id[index] = None
         self.controlnet_512[index] = None
 
+    def prepare_latents(
+        self,
+        batch_size,
+        height,
+        width,
+        generator,
+        num_inference_steps,
+        dtype,
+    ):
+        latents = torch.randn(
+            (
+                batch_size,
+                4,
+                height // 8,
+                width // 8,
+            ),
+            generator=generator,
+            dtype=torch.float32,
+        ).to(dtype)
+
+        self.scheduler.set_timesteps(num_inference_steps)
+        self.scheduler.is_scale_input_called = True
+        latents = latents * self.scheduler.init_noise_sigma
+        return latents
+
     def prepare_image_latents(
         self,
         image,
@@ -519,19 +544,30 @@ def generate_images(
 
         # guidance scale as a float32 tensor.
         guidance_scale = torch.tensor(guidance_scale).to(torch.float32)
-
-        # Prepare input image latent
-        init_latents, final_timesteps = self.prepare_image_latents(
-            image=image,
-            batch_size=batch_size,
-            height=height,
-            width=width,
-            generator=generator,
-            num_inference_steps=num_inference_steps,
-            strength=strength,
-            dtype=dtype,
-            resample_type=resample_type,
-        )
+        if image is not None:
+            # Prepare input image latent
+            init_latents, final_timesteps = self.prepare_image_latents(
+                image=image,
+                batch_size=batch_size,
+                height=height,
+                width=width,
+                generator=generator,
+                num_inference_steps=num_inference_steps,
+                strength=strength,
+                dtype=dtype,
+                resample_type=resample_type,
+            )
+        else:
+            # Prepare initial latent.
+            init_latents = self.prepare_latents(
+                batch_size=batch_size,
+                height=height,
+                width=width,
+                generator=generator,
+                num_inference_steps=num_inference_steps,
+                dtype=dtype,
+            )
+            final_timesteps = self.scheduler.timesteps
 
         # Get Image latents
         latents = self.produce_stencil_latents(