Added Support of Xlabs controlnet to FluxControlNetInpaintPipeline (#9770)

* added xlabs support

src/diffusers/pipelines/flux/pipeline_flux_controlnet_inpainting.py

@@ -932,6 +932,9 @@ class FluxControlNetInpaintPipeline(DiffusionPipeline, FluxLoraLoaderMixin, From
             )
             height, width = control_image.shape[-2:]
 
+            # xlab controlnet has a input_hint_block and instantx controlnet does not
+            controlnet_blocks_repeat = False if self.controlnet.input_hint_block is None else True
+            if self.controlnet.input_hint_block is None:
                 # vae encode
                 control_image = self.vae.encode(control_image).latent_dist.sample()
                 control_image = (control_image - self.vae.config.shift_factor) * self.vae.config.scaling_factor
@@ -954,7 +957,9 @@ class FluxControlNetInpaintPipeline(DiffusionPipeline, FluxLoraLoaderMixin, From
         elif isinstance(self.controlnet, FluxMultiControlNetModel):
             control_images = []
 
-            for control_image_ in control_image:
+            # xlab controlnet has a input_hint_block and instantx controlnet does not
+            controlnet_blocks_repeat = False if self.controlnet.nets[0].input_hint_block is None else True
+            for i, control_image_ in enumerate(control_image):
                 control_image_ = self.prepare_image(
                     image=control_image_,
                     width=width,
@@ -966,6 +971,7 @@ class FluxControlNetInpaintPipeline(DiffusionPipeline, FluxLoraLoaderMixin, From
                 )
                 height, width = control_image_.shape[-2:]
 
+                if self.controlnet.nets[0].input_hint_block is None:
                     # vae encode
                     control_image_ = self.vae.encode(control_image_).latent_dist.sample()
                     control_image_ = (control_image_ - self.vae.config.shift_factor) * self.vae.config.scaling_factor
@@ -1129,6 +1135,7 @@ class FluxControlNetInpaintPipeline(DiffusionPipeline, FluxLoraLoaderMixin, From
                     img_ids=latent_image_ids,
                     joint_attention_kwargs=self.joint_attention_kwargs,
                     return_dict=False,
+                    controlnet_blocks_repeat=controlnet_blocks_repeat,
                 )[0]
 
                 # compute the previous noisy sample x_t -> x_t-1