Update app.py

app.py

@@ -132,7 +132,7 @@ def generate(segment, image, prompt, size, guidance_scale, num_inference_steps,
         image_latents = pipe.image_latents(
                 image, 1, size, size, device, torch.float32, num_channels_latents, num_latent_frames
         )
-        image_latents = image_latents.to(pipe.transformer.dtype)
+        image_latents = image_latents.to("cuda", pipe.transformer.dtype)
         pipe.vae.to("cpu")
         torch.cuda.empty_cache()
         guidance = torch.tensor([guidance_scale] * latents.shape[0], dtype=transformer_dtype, device=device) * 1000.0