diff --git a/lidar_pipeline/rendering.py b/lidar_pipeline/rendering.py
index c66de4c..6fbed65 100644
--- a/lidar_pipeline/rendering.py
+++ b/lidar_pipeline/rendering.py
@@ -204,7 +204,7 @@ def _apply_colormap(data, tif_file):
     # Find matching colormap
     for key, info in COLORMAPS.items():
         if key in name:
-            valid_data = np.asarray(data[~np.isnan(data)] if hasattr(data, 'compressed') else data.flatten())
+            valid_data = np.asarray(data.compressed() if hasattr(data, 'compressed') else data.flatten())
             valid_data = valid_data[~np.isnan(valid_data)]
 
             vmin = vmax = None
@@ -239,7 +239,7 @@ def _apply_colormap(data, tif_file):
             return data, info['cmap'], info['title'], legend, info['description'], False
 
     # Default: terrain colormap with percentile stretch
-    valid_data = np.asarray(data[~np.isnan(data)] if hasattr(data, 'compressed') else data.flatten())
+    valid_data = np.asarray(data.compressed() if hasattr(data, 'compressed') else data.flatten())
     valid_data = valid_data[~np.isnan(valid_data)]
     p2, p98 = np.percentile(valid_data, (2, 98))
     data = np.clip((data - p2) / (p98 - p2), 0, 1)
@@ -324,6 +324,10 @@ def tif_to_png(tif_file, vis_dir, resolution):
                 valid_data = np.asarray(data.compressed() if hasattr(data, 'compressed') else data.flatten())
                 valid_data = valid_data[~np.isnan(valid_data)]
 
+            # Convert MaskedArray to plain ndarray (NaN-filled) to avoid numpy warnings
+            if isinstance(data, np.ma.MaskedArray):
+                data = np.ma.filled(data, np.nan)
+
             # Apply colormap
             data, cmap, title, legend_label, description, is_rgb_result = _apply_colormap(data, tif_file)
 
diff --git a/lidar_pipeline/visualizations.py b/lidar_pipeline/visualizations.py
index 45b73a8..5c2e3da 100644
--- a/lidar_pipeline/visualizations.py
+++ b/lidar_pipeline/visualizations.py
@@ -6,6 +6,7 @@ parameters and returns the path to the output GeoTIFF file, or None on error.
 
 import logging
 import time
+import warnings
 from pathlib import Path
 
 import numpy as np
@@ -391,8 +392,10 @@ def generate_mslrm(dem_file, basename, vis_dir, resolution):
             lrm_stack.append(lrm_norm.astype(np.float32))
 
         lrm_array = np.array(lrm_stack)
-        with np.errstate(invalid='ignore'):
-            mslrm = np.sqrt(np.nanmean(lrm_array ** 2, axis=0))
+        with np.errstate(invalid='ignore', divide='ignore'):
+            with warnings.catch_warnings():
+                warnings.filterwarnings('ignore', message='Mean of empty slice')
+                mslrm = np.sqrt(np.nanmean(lrm_array ** 2, axis=0))
         mslrm[nan_mask] = np.nan
         _save_tif(output, mslrm.astype(np.float32), transform, crs)
         logger.info(f"  ✓ MSRM terminé ({time.time()-t0:.1f}s){gpu_tag}")
@@ -659,8 +662,10 @@ def generate_wavelet(dem_file, basename, vis_dir, resolution):
             response = response / std_val
             wavelet_stack.append(response)
 
-        with np.errstate(invalid='ignore'):
-            combined = np.sqrt(np.nanmean(np.array(wavelet_stack) ** 2, axis=0))
+        with np.errstate(invalid='ignore', divide='ignore'):
+            with warnings.catch_warnings():
+                warnings.filterwarnings('ignore', message='Mean of empty slice')
+                combined = np.sqrt(np.nanmean(np.array(wavelet_stack) ** 2, axis=0))
         combined[nan_mask] = np.nan
 
         _save_tif(output, combined.astype(np.float32), transform, crs)