diff --git a/satpy/dataset.py b/satpy/dataset.py
index a1ffe38e54..18621e537b 100644
--- a/satpy/dataset.py
+++ b/satpy/dataset.py
@@ -15,8 +15,7 @@
 #
 # You should have received a copy of the GNU General Public License along with
 # satpy.  If not, see <http://www.gnu.org/licenses/>.
-"""Dataset objects.
-"""
+"""Dataset objects."""
 
 import sys
 import logging
@@ -171,6 +170,7 @@ class DatasetID(DatasetID):
     """
 
     def __new__(cls, *args, **kwargs):
+        """Create new DatasetID."""
         ret = super(DatasetID, cls).__new__(cls, *args, **kwargs)
         if ret.modifiers is not None and not isinstance(ret.modifiers, tuple):
             raise TypeError("'DatasetID' modifiers must be a tuple or None, "
@@ -269,6 +269,26 @@ def _to_trimmed_dict(self):
                 if getattr(self, key) is not None}
 
 
+def create_filtered_dsid(dataset_key, **dfilter):
+    """Create a DatasetID matching *dataset_key* and *dfilter*.
+
+    If a proprety is specified in both *dataset_key* and *dfilter*, the former
+    has priority.
+
+    """
+    try:
+        ds_dict = dataset_key.to_dict()
+    except AttributeError:
+        if isinstance(dataset_key, str):
+            ds_dict = {'name': dataset_key}
+        elif isinstance(dataset_key, numbers.Number):
+            ds_dict = {'wavelength': dataset_key}
+    for key, value in dfilter.items():
+        if value is not None:
+            ds_dict.setdefault(key, value)
+    return DatasetID.from_dict(ds_dict)
+
+
 def dataset_walker(datasets):
     """Walk through *datasets* and their ancillary data.
 
diff --git a/satpy/node.py b/satpy/node.py
index 3faaaa4992..c815b4d0de 100644
--- a/satpy/node.py
+++ b/satpy/node.py
@@ -20,6 +20,7 @@
 from satpy import DatasetDict, DatasetID, DATASET_KEYS
 from satpy.readers import TooManyResults
 from satpy.utils import get_logger
+from satpy.dataset import create_filtered_dsid
 
 LOG = get_logger(__name__)
 # Empty leaf used for marking composites with no prerequisites
@@ -410,7 +411,7 @@ def _find_compositor(self, dataset_key, **dfilter):
             compositor = self.get_compositor(dataset_key)
         except KeyError:
             raise KeyError("Can't find anything called {}".format(str(dataset_key)))
-        dataset_key = compositor.id
+        dataset_key = create_filtered_dsid(compositor.id, **dfilter)
         root = Node(dataset_key, data=(compositor, [], []))
         if src_node is not None:
             self.add_child(root, src_node)
@@ -434,16 +435,7 @@ def _find_compositor(self, dataset_key, **dfilter):
 
     def get_filtered_item(self, dataset_key, **dfilter):
         """Get the item matching *dataset_key* and *dfilter*."""
-        try:
-            ds_dict = dataset_key.to_dict()
-        except AttributeError:
-            if isinstance(dataset_key, str):
-                ds_dict = {'name': dataset_key}
-            elif isinstance(dataset_key, float):
-                ds_dict = {'wavelength': dataset_key}
-        clean_filter = {key: value for key, value in dfilter.items() if value is not None}
-        ds_dict.update(clean_filter)
-        dsid = DatasetID.from_dict(ds_dict)
+        dsid = create_filtered_dsid(dataset_key, **dfilter)
         return self[dsid]
 
     def _find_dependencies(self, dataset_key, **dfilter):
@@ -463,7 +455,8 @@ def _find_dependencies(self, dataset_key, **dfilter):
 
         # 0 check if the *exact* dataset is already loaded
         try:
-            node = self.getitem(dataset_key)
+            dsid = create_filtered_dsid(dataset_key, **dfilter)
+            node = self.getitem(dsid)
             LOG.trace("Found exact dataset already loaded: {}".format(node.name))
             return node, set()
         except KeyError:
diff --git a/satpy/tests/test_scene.py b/satpy/tests/test_scene.py
index 8081c97496..2aec7f1997 100644
--- a/satpy/tests/test_scene.py
+++ b/satpy/tests/test_scene.py
@@ -1047,6 +1047,28 @@ def test_load_multiple_resolutions(self, cri, cl):
         self.assertEqual(loaded_ids[1].name, 'comp25')
         self.assertEqual(loaded_ids[1].resolution, 1000)
 
+    @mock.patch('satpy.composites.CompositorLoader.load_compositors')
+    @mock.patch('satpy.scene.Scene.create_reader_instances')
+    def test_load_same_subcomposite(self, cri, cl):
+        """Test loading a composite and one of it's subcomposites at the same time."""
+        import satpy.scene
+        from satpy.tests.utils import FakeReader, test_composites
+        cri.return_value = {'fake_reader': FakeReader(
+            'fake_reader', 'fake_sensor')}
+        comps, mods = test_composites('fake_sensor')
+        cl.return_value = (comps, mods)
+        scene = satpy.scene.Scene(filenames=['bla'],
+                                  base_dir='bli',
+                                  reader='fake_reader')
+
+        scene.load(['comp24', 'comp25'], resolution=500)
+        loaded_ids = list(scene.datasets.keys())
+        self.assertEqual(len(loaded_ids), 2)
+        self.assertEqual(loaded_ids[0].name, 'comp24')
+        self.assertEqual(loaded_ids[0].resolution, 500)
+        self.assertEqual(loaded_ids[1].name, 'comp25')
+        self.assertEqual(loaded_ids[1].resolution, 500)
+
     @mock.patch('satpy.composites.CompositorLoader.load_compositors')
     @mock.patch('satpy.scene.Scene.create_reader_instances')
     def test_load_comp5(self, cri, cl):
@@ -1733,8 +1755,8 @@ def _test(self, sensor_names):
         scene.load(['comp11', 'comp23'])
         comp11_node = scene.dep_tree['comp11']
         comp23_node = scene.dep_tree['comp23']
-        self.assertEqual(comp11_node.data[1][-1].name, 'ds10')
-        self.assertEqual(comp23_node.data[1][0].name, 'ds8')
+        self.assertEqual(comp11_node.data[1][-1].name.name, 'ds10')
+        self.assertEqual(comp23_node.data[1][0].name.name, 'ds8')
         loaded_ids = list(scene.datasets.keys())
         self.assertEqual(len(loaded_ids), 2)
         self.assertIn('comp11', scene.datasets)