updated app.py

app.py

@@ -1,14 +1,68 @@
 import gradio as gr
-​
-# inference function takes prompt, negative prompt and image
-def infer(image, prompt):
-    # implement your inference function here
-    output_image = image
-    return output_image
-​
+import jax
+import jax.numpy as jnp
+import numpy as np
+from flax.jax_utils import replicate
+from flax.training.common_utils import shard
+from PIL import Image
+from diffusers import FlaxStableDiffusionControlNetPipeline, FlaxControlNetModel
+import cv2
+
 title = "ControlNet for Cartoon-ifying"
 description = "This is a demo on ControlNet for changing images of people into cartoons of different styles."
 examples = [["./simpsons_human_1.jpg", "turn into a simpsons character", "./simpsons_animated_1.jpg"]]
-​
-gr.Interface(fn = infer, inputs = ["image", "text"], outputs = "image",
-            title = title, description = description, examples = examples, theme='gradio/soft').launch()
+
+
+
+# Constants
+low_threshold = 100
+high_threshold = 200
+
+base_model_path = "runwayml/stable-diffusion-v1-5"
+controlnet_path = "lmattingly/controlnet-uncanny-simpsons"
+#controlnet_path = "JFoz/dog-cat-pose"
+
+# Models
+controlnet, controlnet_params = FlaxControlNetModel.from_pretrained(
+    controlnet_path, dtype=jnp.bfloat16
+)
+pipe, params = FlaxStableDiffusionControlNetPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5", controlnet=controlnet, revision="flax", dtype=jnp.bfloat16
+)
+
+def create_key(seed=0):
+    return jax.random.PRNGKey(seed)
+
+def infer(prompts, image):
+    params["controlnet"] = controlnet_params
+    
+    num_samples = 1 #jax.device_count()
+    rng = create_key(0)
+    rng = jax.random.split(rng, jax.device_count())
+    im = image
+    image = Image.fromarray(im)
+    
+    prompt_ids = pipe.prepare_text_inputs([prompts] * num_samples)
+    processed_image = pipe.prepare_image_inputs([image] * num_samples)
+    
+    p_params = replicate(params)
+    prompt_ids = shard(prompt_ids)
+    processed_image = shard(processed_image)
+    
+    output = pipe(
+        prompt_ids=prompt_ids,
+        image=processed_image,
+        params=p_params,
+        prng_seed=rng,
+        num_inference_steps=50,
+        jit=True,
+    ).images
+    
+    output_images = pipe.numpy_to_pil(np.asarray(output.reshape((num_samples,) + output.shape[-3:])))
+    return output_images
+
+
+gr.Interface(fn = infer, inputs = ["text", "image"], outputs = "image",
+             title = title, description = description, theme='gradio/soft',
+             examples=[["a simpsons cartoon character", "simpsons_human_1.jpg"]]
+).launch()