Clarify purpose and mark as deprecated

Fix inversion prompt broadcasting

Author: clarencechen

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_pix2pix_zero.py

@@ -36,6 +36,7 @@
 from ...utils import (
     PIL_INTERPOLATION,
     BaseOutput,
+    deprecate,
     is_accelerate_available,
     is_accelerate_version,
     logging,
@@ -721,23 +722,31 @@ def prepare_image_latents(self, image, batch_size, dtype, device, generator=None
             )
 
         if isinstance(generator, list):
-            init_latents = [
-                self.vae.encode(image[i : i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size)
-            ]
-            init_latents = torch.cat(init_latents, dim=0)
+            latents = [self.vae.encode(image[i : i + 1]).latent_dist.sample(generator[i]) for i in range(batch_size)]
+            latents = torch.cat(latents, dim=0)
         else:
-            init_latents = self.vae.encode(image).latent_dist.sample(generator)
-
-        init_latents = self.vae.config.scaling_factor * init_latents
-
-        if batch_size > init_latents.shape[0] and batch_size % init_latents.shape[0] != 0:
-            raise ValueError(
-                f"Cannot duplicate `image` of batch size {init_latents.shape[0]} to {batch_size} text prompts."
-            )
+            latents = self.vae.encode(image).latent_dist.sample(generator)
+
+        latents = self.vae.config.scaling_factor * latents
+
+        if batch_size != latents.shape[0]:
+            if batch_size % latents.shape[0] == 0:
+                # expand image_latents for batch_size
+                deprecation_message = (
+                    f"You have passed {batch_size} text prompts (`prompt`), but only {latents.shape[0]} initial"
+                    " images (`image`). Initial images are now duplicating to match the number of text prompts. Note"
+                    " that this behavior is deprecated and will be removed in a version 1.0.0. Please make sure to update"
+                    " your script to pass as many initial images as text prompts to suppress this warning."
+                )
+                deprecate("len(prompt) != len(image)", "1.0.0", deprecation_message, standard_warn=False)
+                additional_latents_per_image = batch_size // latents.shape[0]
+                latents = torch.cat([latents] * additional_latents_per_image, dim=0)
+            else:
+                raise ValueError(
+                    f"Cannot duplicate `image` of batch size {latents.shape[0]} to {batch_size} text prompts."
+                )
         else:
-            init_latents = torch.cat([init_latents], dim=0)
-
-        latents = init_latents
+            latents = torch.cat([latents], dim=0)
 
         return latents