Implement Nova Farma S1 Max: Layered terrain, water mechanics, Y-sorting, and asset cleanup

scripts/clean_and_resize_white.py

@@ -0,0 +1,83 @@
+import cv2
+import numpy as np
+import os
+import sys
+
+def remove_white_bg_and_resize(path, target_width=None):
+    if not os.path.exists(path):
+        print(f"File not found: {path}")
+        return
+
+    img = cv2.imread(path)
+    if img is None:
+        print(f"Error loading {path}")
+        return
+
+    # Convert to RGBA
+    img = cv2.cvtColor(img, cv2.COLOR_BGR2BGRA)
+
+    # Threshold for white (allowing slight off-white/compression artifacts)
+    # White is (255,255,255). Let's say anything > 240 in all channels.
+    lower_white = np.array([240, 240, 240, 255]) 
+    upper_white = np.array([255, 255, 255, 255])
+    
+    # Create mask: looking for white pixels
+    # Note: cv2.inRange checks ranges. For BGRA, A is 255.
+    # However, input usually doesn't have A yet if loaded as BGR. 
+    # But we converted.
+    
+    # Alternative: Grayscale thresholding is often safer for "almost white" backgrounds in illustrations
+    gray = cv2.cvtColor(img, cv2.COLOR_BGRA2GRAY)
+    # Create binary mask where white is 255, else 0
+    _, mask = cv2.threshold(gray, 240, 255, cv2.THRESH_BINARY)
+    
+    # Invert mask: We want to KEEP non-white. 
+    # Non-white = 0 in mask? No, thresholded white became 255.
+    # So we invert.
+    mask_inv = cv2.bitwise_not(mask)
+    
+    # Morphological cleanup (remove small white specks inside object if necessary, or smooth edges)
+    # For vector art, simple threshold usually works well.
+    
+    # Assign new alpha channel
+    b, g, r, a = cv2.split(img)
+    # Use mask_inv as alpha
+    final_img = cv2.merge((b, g, r, mask_inv))
+    
+    # Crop to content (Trim transparent borders)
+    coords = cv2.findNonZero(mask_inv)
+    if coords is not None:
+        x, y, w, h = cv2.boundingRect(coords)
+        # Padding
+        pad = 2
+        x = max(0, x - pad)
+        y = max(0, y - pad)
+        w = min(final_img.shape[1] - x, w + 2*pad)
+        h = min(final_img.shape[0] - y, h + 2*pad)
+        final_img = final_img[y:y+h, x:x+w]
+
+    # Resize if target_width provided
+    if target_width:
+        h_curr, w_curr = final_img.shape[:2]
+        if w_curr > target_width: # Only scale down
+            ratio = target_width / w_curr
+            new_h = int(h_curr * ratio)
+            final_img = cv2.resize(final_img, (target_width, new_h), interpolation=cv2.INTER_AREA)
+            print(f"Resized to {target_width}x{new_h}")
+
+    cv2.imwrite(path, final_img)
+    print(f"Processed (Cleaned White BG + Trim): {path}")
+
+TARGETS = {
+    'assets/DEMO_FAZA1/UI/badge_purchase.png': 512,
+    'assets/DEMO_FAZA1/UI/health_critical.png': 512,
+    'assets/DEMO_FAZA1/VFX/toxic_fog.png': 512,
+    'assets/DEMO_FAZA1/Environment/obstacle_thorns.png': 256,
+    'assets/DEMO_FAZA1/Structures/foundation_concrete.png': 512,
+    'assets/DEMO_FAZA1/Items/hay_drop_0.png': 128
+}
+
+if __name__ == "__main__":
+    base = os.getcwd()
+    for rel, w in TARGETS.items():
+        remove_white_bg_and_resize(os.path.join(base, rel), w)