“Update”

app.py

@@ -1,5 +1,5 @@
 # app.py 
-# Import gradio_interface
+#IMPORT gradio_interface
 from src.gradio_interface import demo
 
 # Launch the Gradio app

ctb-modal.py

@@ -0,0 +1,24 @@
+# modal_app.py
+import modal
+#IMPORT gradio_interface
+from src.gradio_interface import demo
+
+# Create a Modal app
+app = modal.App("ctb-image-generator")
+
+# Define the Modal image
+image = modal.Image.debian_slim().pip_install(
+    "diffusers", 
+    "transformers",
+    "torch",
+    "accelerate",
+    "gradio"
+)
+
+@app.local_entrypoint()
+def main():
+    with modal.enable_output():
+        demo.launch()
+
+if __name__ == "__main__":
+    main()

example-hello-world.py

@@ -0,0 +1,30 @@
+import sys
+
+import modal
+
+app = modal.App("example-hello-world")
+
+@app.function()
+def f(i):
+    if i % 2 == 0:
+        print("hello", i)
+    else:
+        print("world", i, file=sys.stderr)
+
+    return i * i
+
+@app.local_entrypoint()
+def main():
+    # run the function locally
+    print(f.local(1000))
+
+    # run the function remotely on Modal
+    print(f.remote(1000))
+
+    # run the function in parallel and remotely on Modal
+    total = 0
+    for ret in f.map(range(200)):
+        total += ret
+
+    print(total)
+

src/download_flux_modal.py

@@ -0,0 +1,34 @@
+import modal
+from pathlib import Path
+
+# Create or get existing volume
+volume = modal.Volume.from_name("flux-model-vol", create_if_missing=True)
+MODEL_DIR = Path("/models")
+
+# Set up image with dependencies
+download_image = (
+    modal.Image.debian_slim()
+    .pip_install("huggingface_hub[hf_transfer]")
+    .env({"HF_HUB_ENABLE_HF_TRANSFER": "1"})  # Enable fast Rust download client
+)
+
+# Create Modal app
+app = modal.App("flux-model-setup")
+
+@app.function(
+    volumes={MODEL_DIR: volume},
+    image=download_image,
+)
+def download_flux():
+    from huggingface_hub import snapshot_download
+    
+    repo_id = "black-forest-labs/FLUX.1-dev"
+    snapshot_download(
+        repo_id=repo_id,
+        local_dir=MODEL_DIR / repo_id.split("/")[1]
+    )
+    print(f"FLUX model downloaded to {MODEL_DIR / repo_id.split('/')[1]}")
+
+@app.local_entrypoint()
+def main():
+    download_flux()

src/img_gen_modal.py

@@ -0,0 +1,88 @@
+#img_gen_modal.py
+import modal
+import sys
+import os
+import random
+from datetime import datetime
+import random
+import io
+
+
+@app.function(
+    image=image,
+    #gpu="T4",
+    timeout=600
+)
+def generate_image(prompt_alias, team_color, model_alias, custom_prompt, height=360, width=640, 
+                  num_inference_steps=20, guidance_scale=2.0, seed=-1):
+    import torch
+    from diffusers import StableDiffusionPipeline
+    
+    # Find the selected prompt and model
+    try:
+        prompt = next(p for p in prompts if p["alias"] == prompt_alias)["text"]
+        model_name = next(m for m in models if m["alias"] == model_alias)["name"]
+    except StopIteration:
+        return None, "ERROR: Invalid prompt or model selected."
+
+    # Determine the enemy color
+    enemy_color = "blue" if team_color.lower() == "red" else "red"
+    
+    # Print the original prompt and dynamic values for debugging
+    print("Original Prompt:")
+    print(prompt)
+    print(f"Enemy Color: {enemy_color}")
+    print(f"Team Color: {team_color.lower()}")
+
+    # Format the prompt
+    prompt = prompt.format(team_color=team_color.lower(), enemy_color=enemy_color)
+    
+    # Print the formatted prompt for debugging
+    print("\nFormatted Prompt:")
+    print(prompt)
+
+    # Append custom prompt if provided
+    if custom_prompt and len(custom_prompt.strip()) > 0:
+        prompt += " " + custom_prompt.strip()
+
+    # Randomize seed if needed
+    if seed == -1:
+        seed = random.randint(0, 1000000)
+
+    # Initialize the pipeline
+    pipe = StableDiffusionPipeline.from_pretrained(
+        model_name,
+        torch_dtype=torch.float16,
+        use_safetensors=True,
+        variant="fp16"
+    )
+    pipe.to("cpu")
+
+    # Generate the image
+    try:
+        image = pipe(
+            prompt,
+            guidance_scale=guidance_scale,
+            num_inference_steps=num_inference_steps,
+            width=width,
+            height=height,
+            generator=torch.Generator("cuda").manual_seed(seed)
+        ).images[0]
+        
+        # Convert PIL image to bytes
+        img_byte_arr = io.BytesIO()
+        image.save(img_byte_arr, format='PNG')
+        img_byte_arr = img_byte_arr.getvalue()
+    except Exception as e:
+        return None, f"ERROR: Failed to generate image. Details: {e}"
+
+    # Save the image with a timestamped filename
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    output_filename = f"{timestamp}_{model_alias.replace(' ', '_').lower()}_{prompt_alias.replace(' ', '_').lower()}_{team_color.lower()}.png"
+    try:
+        image.save(output_filename)
+    except Exception as e:     
+        return img_byte_arr, "Image generated successfully!"
+    except Exception as e:
+        return None, f"ERROR: Failed to generate image. Details: {e}"
+    return output_filename, "Image generated successfully!"