a1111 example

api-txt2img/README.md

@@ -1,4 +1,4 @@
-# Image Generation Example with AUTOMATIC1111 API
+# Image Generation Example with AUTOMATIC1111 API (`sd3_medium` and `clip`)
 
 This example demonstrates how to use Python to generate images of a cat using the AUTOMATIC1111 Stable Diffusion API on a custom port. In this example, we'll generate **5 images** using a fixed prompt and save them locally as PNG files.
 
@@ -10,6 +10,20 @@ The provided Python script does the following:
 - Decodes the returned base64 image data.
 - Saves each image in the local directory as a PNG file.
 
+In addition to image generation, the example also shows how to use the server's CLIP Interrogator endpoint to generate captions from the created images.
+
+## Server Details
+
+For this example, the AUTOMATIC1111 server is running on the following network details:
+- **IP Address:** `172.30.200.3`
+- **Custom Port:** `35000`
+
+You can use these details to interact with the server for:
+- **Image Generation:** Sending requests to the `/sdapi/v1/txt2img` endpoint.
+- **CLIP Interrogation:** Sending requests to the `/sdapi/v1/clip-interrogate` endpoint.
+
+This means if you want to use the server for generating images or obtaining image captions via CLIP interrogation, you should point your API calls to `http://172.30.200.3:35000`.
+
 ## Prerequisites
 
 Before running the script, ensure you have:

api-txt2img/generate_cat_images.py

@@ -4,7 +4,8 @@ import io
 import base64
 from PIL import Image
 
-def generate_cat_images(ip, port, model_params, prompt="a cat", num_images=5):
+def generate_cat_images(ip, port, model_params, prompt="cat", num_images=5):
+    filenames = []
     for i in range(num_images):
         # Generate a random seed for each image
         seed = random.randint(0, 4294967295)
@@ -13,7 +14,6 @@ def generate_cat_images(ip, port, model_params, prompt="a cat", num_images=5):
         payload = {
             "prompt": prompt,
             "negative_prompt": "",  # Modify if needed
-            "sd_model_checkpoint": model_params["model"],
             "steps": model_params["steps"],
             "cfg_scale": model_params["cfg_scale"],
             "width": model_params["width"],
@@ -33,31 +33,59 @@ def generate_cat_images(ip, port, model_params, prompt="a cat", num_images=5):
             r = response.json()
             
             # Process returned images (the API returns a list in r['images'])
-            for img_data in r.get('images', []):
+            for img_data in r['images']:
                 # Remove any header if present (e.g., "data:image/png;base64,")
-                img_base64 = img_data.split(",", 1)[-1]
+                image = Image.open(io.BytesIO(base64.b64decode(img_data.split(",",1)[0])))
-                image = Image.open(io.BytesIO(base64.b64decode(img_base64)))
                 
                 # Save the image with a unique filename
                 filename = f"cat_{i}.png"
                 image.save(filename)
                 print(f"Saved {filename}")
+                filenames.append(filename)
         except requests.exceptions.Timeout:
             print(f"Timeout occurred while generating image {i}")
         except Exception as e:
             print(f"An error occurred on iteration {i}: {e}")
+    return filenames
+
+def interrogate_images(ip, port, image_files):
+    import requests, base64
+    for filename in image_files:
+        try:
+            # Open the image file and encode it in base64
+            with open(filename, "rb") as f:
+                img_bytes = f.read()
+            img_base64 = base64.b64encode(img_bytes).decode("utf-8")
+            payload = {"image": f"data:image/png;base64,{img_base64}", "model": "clip"}
+
+            # Send a POST request to the CLIP Interrogator API endpoint on the A1111 server
+            response = requests.post(
+                url=f"http://{ip}:{port}/sdapi/v1/interrogate",
+                json=payload,
+                timeout=30
+            )
+            response.raise_for_status()
+            data = response.json()
+
+            caption = data.get("caption", "No caption returned")
+            print(f"Image {filename} caption: {caption}")
+        except Exception as e:
+            print(f"An error occurred during interrogation of {filename}: {e}")
 
 if __name__ == "__main__":
     ip = "172.30.200.3"
     port = 35000  # Custom port for the API
-    # Define your model parameters (update 'model' to your actual checkpoint name)
+    # Define your model parameters (update as needed)
     model_params = {
-        "model": "sd3.5_checkpoint.ckpt",  # Replace with your actual model checkpoint file
         "steps": 20,
-        "cfg_scale": 7.0,
+        "cfg_scale": 7,
-        "width": 512,
+        "width": 1024, # Model specific, don't change
-        "height": 512,
+        "height": 1024, # Model specific, don't change
         "sampler": "Euler"  # Adjust based on your setup
     }
     
-    generate_cat_images(ip, port, model_params, prompt="a cat", num_images=5)
+    # Generate cat images and get the list of filenames
+    image_files = generate_cat_images(ip, port, model_params, prompt="cat, cute, perfect, anime", num_images=5)
+    
+    # Call the CLIP Interrogator function from main using the same ip and port
+    interrogate_images(ip, port, image_files)