Create lorify.py

lorify.py

@@ -0,0 +1,105 @@
+import os
+import random
+import numpy as np
+from huggingface_hub import AsyncInferenceClient
+from translatepy import Translator
+from gradio_client import Client, handle_file
+from PIL import Image
+
+# Constants
+MAX_SEED = np.iinfo(np.int32).max
+HF_TOKEN = os.getenv('HF_TOKEN')  # Set the environment variable for HF_TOKEN
+HF_TOKEN_UPSCALER = os.getenv('HF_TOKEN_UPSCALER')  # Set the environment variable for HF_TOKEN_UPSCALER
+
+
+class Lorify:
+    def __init__(self, hf_token=None, hf_token_upscaler=None):
+        # Optionally load tokens from environment if not passed
+        self.hf_token = hf_token or HF_TOKEN
+        self.hf_token_upscaler = hf_token_upscaler or HF_TOKEN_UPSCALER
+
+        # Initialize clients
+        self.qwen_client = Client("K00B404/HugChatWrap", hf_token=self.hf_token)
+        self.client = AsyncInferenceClient()
+
+        # List of available LoRAs (replace with your LoRA repo names or paths)
+        self.loaded_loras = []
+        self.loras = [
+            "Shakker-Labs/FLUX.1-dev-LoRA-add-details",
+            "XLabs-AI/flux-RealismLora",
+            "enhanceaiteam/Flux-uncensored"
+        ]
+        self.loaded_loras.extend(self.loras)
+
+    # Enable or disable LoRA
+    def enable_lora(self, lora_add, basemodel):
+        return basemodel if not lora_add else lora_add
+
+    # Generate image function
+    async def generate_image(self, prompt, model, lora_word, width, height, scales, steps, seed):
+        try:
+            if seed == -1:
+                seed = random.randint(0, MAX_SEED)
+            seed = int(seed)
+
+            # Translate prompt
+            text = str(Translator().translate(prompt, 'English')) + "," + lora_word
+
+            # Generate image
+            image = await self.client.text_to_image(
+                prompt=text,
+                height=height,
+                width=width,
+                guidance_scale=scales,
+                num_inference_steps=steps,
+                model=model
+            )
+
+            return image, seed
+        except Exception as e:
+            print(f"Error generating image: {e}")
+            return None, None
+
+    # Upscale image function
+    def upscale_image(self, prompt, img_path, upscale_factor):
+        try:
+            # Initialize the upscale client
+            upscale_client = Client("finegrain/finegrain-image-enhancer", hf_token=self.hf_token_upscaler)
+            result = upscale_client.predict(
+                input_image=handle_file(img_path),
+                prompt=prompt,
+                negative_prompt="worst quality, low quality, normal quality",
+                upscale_factor=upscale_factor,
+                controlnet_scale=0.6,
+                controlnet_decay=1,
+                condition_scale=6,
+                denoise_strength=0.35,
+                num_inference_steps=18,
+                solver="DDIM",
+                api_name="/process"
+            )
+            return result[1]  # Return upscale image path
+        except Exception as e:
+            print(f"Error scaling image: {e}")
+            return None
+
+    # Main method to generate and optionally upscale image
+    async def gen_image(self, prompt, basemodel, width, height, scales, steps, seed, upscale_factor, process_upscale, lora_model, process_lora):
+        model = self.enable_lora(lora_model, basemodel) if process_lora else basemodel
+
+        image, seed = await self.generate_image(prompt, model, "", width, height, scales, steps, seed)
+
+        if image is None:
+            print("Image generation failed.")
+            return []
+
+        image_path = "temp_image.jpg"
+        image.save(image_path, format="JPEG")
+
+        upscale_image_path = None
+        if process_upscale:
+            upscale_image_path = self.upscale_image(prompt, image_path, upscale_factor)
+            if upscale_image_path and os.path.exists(upscale_image_path):
+                return [image_path, upscale_image_path]
+
+        return [image_path]