diff --git a/notebooks/wut-web-dev.ipynb b/notebooks/wut-web-dev.ipynb
index 09c950f..4cd1d44 100644
--- a/notebooks/wut-web-dev.ipynb
+++ b/notebooks/wut-web-dev.ipynb
@@ -93,6 +93,63 @@
     "model = load_model(model_file)"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%%capture\n",
+    "def gen_image(test_data_gen,test_dir):\n",
+    "    test_image_gen = ImageDataGenerator(rescale=1./255);\n",
+    "    test_data_gen = test_image_gen.flow_from_directory(batch_size=1,\n",
+    "                                                       directory=test_dir,\n",
+    "                                                       target_size=(IMG_HEIGHT, IMG_WIDTH),\n",
+    "                                                       shuffle=True,\n",
+    "                                                       class_mode='binary');\n",
+    "    return test_data_gen"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%%capture --no-stderr --no-stdout\n",
+    "def gen_image_tmp(obs_waterfall_path):\n",
+    "    tmp_dir = tempfile.mkdtemp()\n",
+    "    test_dir = os.path.join(tmp_dir)\n",
+    "    os.makedirs(test_dir + '/unvetted', exist_ok=True)\n",
+    "    shutil.copy(obs_waterfall_path, test_dir + '/unvetted/')    \n",
+    "    \n",
+    "    img = im.open(obs_waterfall_path).resize( (100,200))\n",
+    "    display(img)\n",
+    "\n",
+    "    # XXX delete tmp dir down below\n",
+    "    #print(test_dir)\n",
+    "    #shutil.rmtree(test_dir)    \n",
+    "    return test_dir"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%%capture\n",
+    "def obs_wutsay(test_data_gen):\n",
+    "    prediction = model.predict(\n",
+    "        x=test_data_gen,\n",
+    "        verbose=0)\n",
+    "    predictions=[]\n",
+    "    prediction_bool = (prediction >0.8)\n",
+    "    predictions = prediction_bool.astype(int)\n",
+    "    \n",
+    "    return prediction_bool"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -118,26 +175,12 @@
     "        \n",
     "        obs_waterfall=os.path.basename(obs_waterfallurl)\n",
     "        obs_waterfall_path = os.path.join('/srv/satnogs/download', str(obs_id), obs_waterfall)\n",
+    "       \n",
+    "        test_dir=gen_image_tmp(obs_waterfall_path)\n",
+    "        test_data_gen=gen_image(obs_waterfall_path, test_dir)\n",
     "        \n",
-    "        tmp_dir = tempfile.mkdtemp()\n",
-    "        test_dir = os.path.join(tmp_dir)\n",
-    "        os.makedirs(test_dir + '/unvetted', exist_ok=True)\n",
-    "        shutil.copy(obs_waterfall_path, test_dir + '/unvetted/')\n",
-    "        img = im.open(obs_waterfall_path).resize( (100,200) )\n",
-    "        test_image_generator = ImageDataGenerator(\n",
-    "            rescale=1./255\n",
-    "        )\n",
-    "        test_data_gen = test_image_generator.flow_from_directory(batch_size=1,\n",
-    "                                                                 directory=test_dir,\n",
-    "                                                                 target_size=(IMG_HEIGHT, IMG_WIDTH),\n",
-    "                                                                 shuffle=True,\n",
-    "                                                                 class_mode='binary')\n",
-    "        prediction = model.predict(\n",
-    "            x=test_data_gen,\n",
-    "            verbose=0)\n",
-    "        predictions=[]\n",
-    "        prediction_bool = (prediction >0.8)\n",
-    "        predictions = prediction_bool.astype(int)\n",
+    "        prediction_bool=obs_wutsay(test_data_gen);\n",
+    "\n",
     "        print()\n",
     "        print('Observation ID: ', obs_id)\n",
     "        print('Encoding:       ', obs_transmitter_mode)\n",
@@ -146,19 +189,13 @@
     "            rating = 'bad'\n",
     "        else:\n",
     "            rating = 'good'\n",
-    "        print('wut AI rating:   %s' % (rating))\n",
-    "        \n",
+    "        print('wut AI rating:   %s' % (rating))          \n",
     "        print()\n",
     "        if obs_transmitter_mode == 'DUV':\n",
     "            print(\"Using DUV training model.\")\n",
     "        else:\n",
     "            print(\"NOTE: wut has not been trained on\", obs_transmitter_mode, \"encodings.\")\n",
-    "\n",
     "        print('https://network.satnogs.org/observations/' + str(obs_id))\n",
-    "        print()\n",
-    "        \n",
-    "        display(img)\n",
-    "        shutil.rmtree(test_dir)\n",
     "        #!cat $obsjsonfile"
    ]
   },
@@ -168,7 +205,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "print('Enter an Observation ID between', minobsid, 'and', maxobsid)\n",
+    "print('Enter an Observation ID between', minobsid, 'and', maxobsid);\n",
     "wutObs_slide = wg.IntText(value='1292461');\n",
     "wg.interact(wutObs, datObs=wutObs_slide);"
    ]