diff --git a/src/calibre/db/backend.py b/src/calibre/db/backend.py
index 37bc27f7cf..dab117d743 100644
--- a/src/calibre/db/backend.py
+++ b/src/calibre/db/backend.py
@@ -1029,12 +1029,12 @@ class DB:
     def restore_notes(self, report_progress):
         self.notes.restore(self.conn, self.tables, report_progress)
 
-    def import_note(self, field, item_id, html_file_path):
+    def import_note(self, field, item_id, html, basedir, ctime, mtime):
         id_val = self.tables[field].id_map[item_id]
-        return self.notes.import_note(self.conn, field, item_id, id_val, html_file_path)
+        return self.notes.import_note(self.conn, field, item_id, id_val, html, basedir, ctime, mtime)
 
-    def export_note(self, field, item_id, dest_dir):
-        return self.notes.export_note(self.conn, field, item_id, dest_dir)
+    def export_note(self, field, item_id):
+        return self.notes.export_note(self.conn, field, item_id)
 
     def initialize_fts(self, dbref):
         self.fts = None
diff --git a/src/calibre/db/cache.py b/src/calibre/db/cache.py
index 159a0ff38c..ce5f56bf26 100644
--- a/src/calibre/db/cache.py
+++ b/src/calibre/db/cache.py
@@ -706,12 +706,16 @@ class Cache:
         return self.backend.unretire_note_for(field, item_id)
 
     @read_api
-    def export_note(self, field, item_id, dest_dir):
-        return self.backend.export_note(field, item_id, dest_dir)
+    def export_note(self, field, item_id):
+        return self.backend.export_note(field, item_id)
 
     @write_api
     def import_note(self, field, item_id, path_to_html_file):
-        return self.backend.import_note(field, item_id, path_to_html_file)
+        with open(path_to_html_file, 'rb') as f:
+            html = f.read()
+            st = os.stat(f.fileno())
+        basedir = os.path.dirname(os.path.abspath(path_to_html_file))
+        return self.backend.import_note(field, item_id, html, basedir, st.st_ctime, st.st_mtime)
 
     @write_api  # we need to use write locking as SQLITE gives a locked table error if multiple FTS queries are made at the same time
     def notes_search(
diff --git a/src/calibre/db/constants.py b/src/calibre/db/constants.py
index e83cb811b6..0eebde06d0 100644
--- a/src/calibre/db/constants.py
+++ b/src/calibre/db/constants.py
@@ -13,6 +13,7 @@ NOTES_DB_NAME = 'notes.db'
 DATA_DIR_NAME = 'data'
 DATA_FILE_PATTERN = f'{DATA_DIR_NAME}/**/*'
 BOOK_ID_PATH_TEMPLATE = ' ({})'
+RESOURCE_URL_SCHEME = 'calres'
 
 
 @dataclass
diff --git a/src/calibre/db/notes/connect.py b/src/calibre/db/notes/connect.py
index 5211f7e97e..2bcfb76d6f 100644
--- a/src/calibre/db/notes/connect.py
+++ b/src/calibre/db/notes/connect.py
@@ -33,7 +33,6 @@ copy_marked_up_text = cmt()
 SEP = b'\0\x1c\0'
 DOC_NAME = 'doc.html'
 METADATA_EXT = '.metadata'
-RESOURCE_URL_SCHEME = 'calres'
 
 
 def hash_data(data: bytes) -> str:
@@ -467,26 +466,22 @@ class Notes:
                     report_progress('', i)
         return errors
 
-    def export_note(self, conn, field_name, item_id, dest_dir):
+    def export_note(self, conn, field_name, item_id):
         nd = self.get_note_data(conn, field_name, item_id)
         if nd is None:
             return ''
         from .exim import export_note
-        resources = {}
-        for rh in nd['resource_hashes']:
-            p = make_long_path_useable(self.path_for_resource(rh))
-            if os.path.exists(p):
-                for (name,) in conn.execute('SELECT name FROM notes_db.resources WHERE hash=?', (rh,)):
-                    resources[rh] = (p, name)
-        return export_note(nd, resources, dest_dir)
 
-    def import_note(self, conn, field_name, item_id, item_value, html_file_path):
+        def get_resource(rhash):
+            return self.get_resource_data(conn, rhash)
+        return export_note(nd['doc'], get_resource)
+
+    def import_note(self, conn, field_name, item_id, item_value, html, basedir, ctime=None, mtime=None):
         from .exim import import_note
-        def add_resource(path, name):
-            return self.add_resource(conn, path, name)
-        st = os.stat(html_file_path)
-        doc, searchable_text, resources = import_note(html_file_path, add_resource)
+        def add_resource(path_or_stream_or_data, name):
+            return self.add_resource(conn, path_or_stream_or_data, name)
+        doc, searchable_text, resources = import_note(html, basedir, add_resource)
         return self.set_note(
             conn, field_name, item_id, item_value, marked_up_text=doc, used_resource_hashes=resources, searchable_text=searchable_text,
-            ctime=st.st_ctime, mtime=st.st_mtime
+            ctime=ctime, mtime=mtime
         )
diff --git a/src/calibre/db/notes/exim.py b/src/calibre/db/notes/exim.py
index 0ff93ac8ca..b2b3bd542c 100644
--- a/src/calibre/db/notes/exim.py
+++ b/src/calibre/db/notes/exim.py
@@ -1,19 +1,20 @@
 #!/usr/bin/env python
 # License: GPLv3 Copyright: 2023, Kovid Goyal <kovid at kovidgoyal.net>
 
+import base64
 import os
-import shutil
 from html5_parser import parse
 from lxml import html
 from urllib.parse import unquote, urlparse
 
+from calibre import guess_extension, guess_type
+from calibre.db.constants import RESOURCE_URL_SCHEME
 from calibre.ebooks.chardet import xml_to_unicode
+from calibre.ebooks.oeb.transforms.rasterize import data_url
 from calibre.utils.cleantext import clean_xml_chars
 from calibre.utils.filenames import get_long_path_name, make_long_path_useable
 from calibre.utils.html2text import html2text
 
-from .connect import DOC_NAME, RESOURCE_URL_SCHEME
-
 
 def parse_html(raw):
     try:
@@ -22,42 +23,51 @@ def parse_html(raw):
         return parse(clean_xml_chars(raw), maybe_xhtml=False, sanitize_names=True)
 
 
-def export_note(note_data: dict, resources: dict[str, tuple[str, str]], dest_dir: str) -> str:
-    for rhash, (path, name) in resources.items():
-        d = os.path.join(dest_dir, name)
-        shutil.copy2(path, d)
-    root = parse_html(note_data['doc'])
+def export_note(note_doc: str, get_resource) -> str:
+    root = parse_html(note_doc)
     for img in root.xpath('//img[@src]'):
-        img.attrib.pop('pre-import-src', None)
+        img.attrib.pop('data-pre-import-src', None)
         try:
             purl = urlparse(img.get('src'))
         except Exception:
             continue
         if purl.scheme == RESOURCE_URL_SCHEME:
             rhash = f'{purl.hostname}:{purl.path[1:]}'
-            x = resources.get(rhash)
-            if x is not None:
-                img.set('src', x[1])
+            x = get_resource(rhash)
+            if x:
+                img.set('src', data_url(guess_type(x['name'])[0], x['data']))
+                img.set('data-filename', x['name'])
 
-    shtml = html.tostring(root, encoding='utf-8')
-    with open(os.path.join(dest_dir, DOC_NAME), 'wb') as f:
-        f.write(shtml)
-    os.utime(f.name, times=(note_data['ctime'], note_data['mtime']))
-    return DOC_NAME
+    return html.tostring(root, encoding='unicode')
 
 
-def import_note(path_to_html_file: str, add_resource) -> dict:
-    path_to_html_file = path_to_html_file
-    with open(make_long_path_useable(path_to_html_file), 'rb') as f:
-        raw = f.read()
-    shtml = xml_to_unicode(raw, strip_encoding_pats=True, assume_utf8=True)[0]
-    basedir = os.path.dirname(os.path.abspath(path_to_html_file))
-    basedir = os.path.normcase(get_long_path_name(basedir) + os.sep)
+def import_note(shtml: str | bytes, basedir: str, add_resource) -> dict:
+    shtml = xml_to_unicode(shtml, strip_encoding_pats=True, assume_utf8=True)[0]
+    basedir = os.path.normcase(get_long_path_name(os.path.abspath(basedir)) + os.sep)
     root = parse_html(shtml)
     resources = set()
+
+    def ar(img, path_or_data, name):
+        rhash = add_resource(path_or_data, name)
+        scheme, digest = rhash.split(':', 1)
+        img.set('src', f'{RESOURCE_URL_SCHEME}://{scheme}/{digest}')
+        resources.add(rhash)
+
     for img in root.xpath('//img[@src]'):
         src = img.attrib.pop('src')
-        img.set('pre-import-src', src)
+        img.set('data-pre-import-src', src)
+        if src.startswith('data:'):
+            d = src.split(':', 1)[-1]
+            menc, payload = d.partition(',')
+            mt, enc = menc.partition(';')
+            if enc != 'base64':
+                continue
+            try:
+                d = base64.standard_b64decode(payload)
+            except Exception:
+                continue
+            ar(img, d, img.get('data-filename') or ('image' + guess_extension(mt, strict=False)))
+            continue
         try:
             purl = urlparse(src)
         except Exception:
@@ -65,12 +75,13 @@ def import_note(path_to_html_file: str, add_resource) -> dict:
         if purl.scheme in ('', 'file'):
             path = unquote(purl.path)
             if not os.path.isabs(path):
+                if not basedir:
+                    continue
                 path = os.path.join(basedir, path)
             q = os.path.normcase(get_long_path_name(os.path.abspath(path)))
             if q.startswith(basedir) and os.path.exists(make_long_path_useable(path)):
-                rhash = add_resource(make_long_path_useable(path), os.path.basename(path))
-                scheme, digest = rhash.split(':', 1)
-                img.set('src', f'{RESOURCE_URL_SCHEME}://{scheme}/{digest}')
-                resources.add(rhash)
+                ar(img, make_long_path_useable(path), os.path.basename(path))
     shtml = html.tostring(root, encoding='unicode')
+    for img in root.xpath('//img[@src]'):
+        del img.attrib['src']
     return shtml, html2text(shtml, default_image_alt=' '), resources
diff --git a/src/calibre/db/tests/notes.py b/src/calibre/db/tests/notes.py
index 49805fc241..cc05925e02 100644
--- a/src/calibre/db/tests/notes.py
+++ b/src/calibre/db/tests/notes.py
@@ -135,14 +135,11 @@ def test_cache_api(self: 'NotesTest'):
         note_id = cache.import_note('authors', author_id, f.name)
         self.assertGreater(note_id, 0)
         self.assertIn('<p>test simple exim <img', cache.notes_for('authors', author_id))
-        edir = os.path.join(tdir, 'e')
-        os.mkdir(edir)
-        index_name = cache.export_note('authors', author_id, edir)
-        with open(os.path.join(edir, index_name)) as f:
-            self.assertIn(doc.replace('r 1.png', 'r%201.png'), f.read())
-        for x in ('r 1.png', 'r 2.png'):
-            with open(os.path.join(idir, x), 'rb') as a, open(os.path.join(edir, x), 'rb') as b:
-                self.assertEqual(a.read(), b.read())
+        exported = cache.export_note('authors', author_id)
+        self.assertIn('<p>test simple exim <img src="', exported)
+        from html5_parser import parse
+        root = parse(exported)
+        self.ae(root.xpath('//img/@data-filename'), ['r 1.png', 'r 2.png'])
 
 
 def test_fts(self: 'NotesTest'):