diff --git a/a.jpg b/a.jpg
new file mode 100644
index 0000000..b8267af
Binary files /dev/null and b/a.jpg differ
diff --git a/b.jpg b/b.jpg
new file mode 100644
index 0000000..99dce7c
Binary files /dev/null and b/b.jpg differ
diff --git a/dedup.py b/dedup.py
index 7568b84..dc8d4be 100644
--- a/dedup.py
+++ b/dedup.py
@@ -346,6 +346,12 @@ def find_duplicate_with_rotation(img1, img2):
     
     # Return similarity score (lower is more similar)
     return len(matches)
+    
+def get_image_dimensions_cv(img):
+    if img is not None:
+        height, width = img.shape[:2]
+        return width, height
+    return None    
                     
 """
 xxhash is about 5–10x faster than SHA256, non-cryptographic.
@@ -383,6 +389,16 @@ if __name__ == "__main__":
     # Load large images
     large_img1 = cv2.imread(file1)  # e.g., 4000x3000 pixels
     large_img2 = cv2.imread(file2)   # e.g., 4000x3000 pixels
+    
+    w, h = get_image_dimensions_cv(large_img1)
+    w2, h2 = get_image_dimensions_cv(large_img2)
+    if w != w2 and w != h2:
+        print("Diffent Resolutions")
+        exit(0)
+    
+    if h != h2 and h != w2:
+        print("Diffent Resolutions")
+        exit(0)
 
     # Align with downscaling (initially process at 1/4 size)
     aligned, matrix, angle = align_with_downscaling(
@@ -397,12 +413,22 @@ if __name__ == "__main__":
     # Print debug info
     print(f"Detected rotation: {angle}°")
     print(f"Final transformation matrix:\n{matrix}")
-
-    score = find_duplicate_with_rotation(large_img1, aligned)
-    print(f"Score: {score}")
     
     # Calculate scores
     matrix_score = matrix_similarity_score(matrix)
+    
+    if len(sys.argv) > 3:
+	    is_score = sys.argv[3]
+    else:
+	    is_score = ""
+			
+    if matrix_score == 1.0 and is_score != "scores":
+        print("✅ Perfect Matrix score, should be identical")
+        exit(1)
+    if is_score == "scores":
+	    score = find_duplicate_with_rotation(large_img1, aligned)
+	    print(f"Score: {score}")	
+
     decomposed_score = decomposed_similarity_score(matrix, large_img1.shape[1])
     combined_score = comprehensive_similarity(large_img1, aligned, matrix)