diff --git a/yt_dlp/YoutubeDL.py b/yt_dlp/YoutubeDL.py
index 7e065daa1f..c595021612 100644
--- a/yt_dlp/YoutubeDL.py
+++ b/yt_dlp/YoutubeDL.py
@@ -1596,9 +1596,13 @@ def process_ie_result(self, ie_result, download=True, extra_info=None):
             if not info:
                 return info
 
+            exempted_fields = {'_type', 'url', 'ie_key'}
+            if not ie_result.get('section_end') and ie_result.get('section_start') is None:
+                # For video clips, the id etc of the clip extractor should be used
+                exempted_fields |= {'id', 'extractor', 'extractor_key'}
+
             new_result = info.copy()
-            new_result.update(filter_dict(ie_result, lambda k, v: (
-                v is not None and k not in {'_type', 'url', 'id', 'extractor', 'extractor_key', 'ie_key'})))
+            new_result.update(filter_dict(ie_result, lambda k, v: v is not None and k not in exempted_fields))
 
             # Extracted info may not be a video result (i.e.
             # info.get('_type', 'video') != video) but rather an url or
@@ -2369,6 +2373,8 @@ def sanitize_numeric_fields(info):
 
         sanitize_string_field(info_dict, 'id')
         sanitize_numeric_fields(info_dict)
+        if info_dict.get('section_end') and info_dict.get('section_start') is not None:
+            info_dict['duration'] = round(info_dict['section_end'] - info_dict['section_start'], 3)
         if (info_dict.get('duration') or 0) <= 0 and info_dict.pop('duration', None):
             self.report_warning('"duration" field is negative, there is an error in extractor')
 
@@ -2604,10 +2610,11 @@ def to_screen(*msg):
             for fmt, chapter in itertools.product(formats_to_download, requested_ranges or [{}]):
                 new_info = self._copy_infodict(info_dict)
                 new_info.update(fmt)
-                if chapter:
+                offset, duration = info_dict.get('section_start') or 0, info_dict.get('duration') or float('inf')
+                if chapter or offset:
                     new_info.update({
-                        'section_start': chapter.get('start_time'),
-                        'section_end': chapter.get('end_time', 0),
+                        'section_start': offset + chapter.get('start_time', 0),
+                        'section_end': offset + min(chapter.get('end_time', 0), duration),
                         'section_title': chapter.get('title'),
                         'section_number': chapter.get('index'),
                     })
diff --git a/yt_dlp/extractor/common.py b/yt_dlp/extractor/common.py
index 3e3e557985..90af415758 100644
--- a/yt_dlp/extractor/common.py
+++ b/yt_dlp/extractor/common.py
@@ -385,6 +385,11 @@ class InfoExtractor:
     release_year:   Year (YYYY) when the album was released.
     composer:       Composer of the piece
 
+    The following fields should only be set for clips that should be cut from the original video:
+
+    section_start:  Start time of the section in seconds
+    section_end:    End time of the section in seconds
+
     Unless mentioned otherwise, the fields should be Unicode strings.
 
     Unless mentioned otherwise, None is equivalent to absence of information.