diff --git a/internal/meta/json_exiftool.go b/internal/meta/json_exiftool.go
index 4fd6a8cc9..b420d640c 100644
--- a/internal/meta/json_exiftool.go
+++ b/internal/meta/json_exiftool.go
@@ -15,6 +15,8 @@ import (
 	"gopkg.in/ugjka/go-tz.v2/tz"
 )
 
+const MimeVideoMP4 = "video/mp4"
+
 // Exiftool parses JSON sidecar data as created by Exiftool.
 func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 	defer func() {
@@ -150,6 +152,18 @@ func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 		}
 	}
 
+	hasTimeOffset := false
+
+	if _, offset := data.TakenAtLocal.Zone(); offset != 0 && !data.TakenAtLocal.IsZero() {
+		hasTimeOffset = true
+	} else if mt, ok := jsonStrings["MIMEType"]; ok && mt == MimeVideoMP4 {
+		// Assume default time zone for MP4 videos is UTC.
+		// see https://exiftool.org/TagNames/QuickTime.html
+		data.TimeZone = time.UTC.String()
+		data.TakenAt = data.TakenAt.UTC()
+		data.TakenAtLocal = time.Time{}
+	}
+
 	// Set time zone and calculate UTC time.
 	if data.Lat != 0 && data.Lng != 0 {
 		zones, err := tz.GetZone(tz.Point{
@@ -161,10 +175,10 @@ func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 			data.TimeZone = zones[0]
 		}
 
-		if !data.TakenAtLocal.IsZero() {
-			if loc, err := time.LoadLocation(data.TimeZone); err != nil {
-				log.Warnf("metadata: unknown time zone %s (exiftool)", data.TimeZone)
-			} else if tl, err := time.ParseInLocation("2006:01:02 15:04:05", data.TakenAtLocal.Format("2006:01:02 15:04:05"), loc); err == nil {
+		if loc, err := time.LoadLocation(data.TimeZone); err != nil {
+			log.Warnf("metadata: unknown time zone %s (exiftool)", data.TimeZone)
+		} else if !data.TakenAtLocal.IsZero() {
+			if tl, err := time.ParseInLocation("2006:01:02 15:04:05", data.TakenAtLocal.Format("2006:01:02 15:04:05"), loc); err == nil {
 				if localUtc, err := time.ParseInLocation("2006:01:02 15:04:05", data.TakenAtLocal.Format("2006:01:02 15:04:05"), time.UTC); err == nil {
 					data.TakenAtLocal = localUtc
 				}
@@ -173,8 +187,15 @@ func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 			} else {
 				log.Errorf("metadata: %s (exiftool)", err.Error()) // this should never happen
 			}
+		} else if !data.TakenAt.IsZero() {
+			if localUtc, err := time.ParseInLocation("2006:01:02 15:04:05", data.TakenAt.In(loc).Format("2006:01:02 15:04:05"), time.UTC); err == nil {
+				data.TakenAtLocal = localUtc
+				data.TakenAt = data.TakenAt.UTC()
+			} else {
+				log.Errorf("metadata: %s (exiftool)", err.Error()) // this should never happen
+			}
 		}
-	} else if _, offset := data.TakenAtLocal.Zone(); offset != 0 && !data.TakenAtLocal.IsZero() {
+	} else if hasTimeOffset {
 		if localUtc, err := time.ParseInLocation("2006:01:02 15:04:05", data.TakenAtLocal.Format("2006:01:02 15:04:05"), time.UTC); err == nil {
 			data.TakenAtLocal = localUtc
 		}
@@ -182,6 +203,18 @@ func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 		data.TakenAt = data.TakenAt.Round(time.Second).UTC()
 	}
 
+	// Set local time if still empty.
+	if data.TakenAtLocal.IsZero() && !data.TakenAt.IsZero() {
+		if loc, err := time.LoadLocation(data.TimeZone); data.TimeZone == "" || err != nil {
+			data.TakenAtLocal = data.TakenAt
+		} else if localUtc, err := time.ParseInLocation("2006:01:02 15:04:05", data.TakenAt.In(loc).Format("2006:01:02 15:04:05"), time.UTC); err == nil {
+			data.TakenAtLocal = localUtc
+			data.TakenAt = data.TakenAt.UTC()
+		} else {
+			log.Errorf("metadata: %s (exiftool)", err.Error()) // this should never happen
+		}
+	}
+
 	if orientation, ok := jsonStrings["Orientation"]; ok && orientation != "" {
 		switch orientation {
 		case "1", "Horizontal (normal)":
diff --git a/internal/meta/json_test.go b/internal/meta/json_test.go
index 3320cc55a..b28613c8d 100644
--- a/internal/meta/json_test.go
+++ b/internal/meta/json_test.go
@@ -2,6 +2,7 @@ package meta
 
 import (
 	"testing"
+	"time"
 
 	"github.com/photoprism/photoprism/pkg/fs"
 
@@ -48,7 +49,7 @@ func TestJSON(t *testing.T) {
 		assert.Equal(t, "2s", data.Duration.String())
 		assert.Equal(t, "2020-05-11 14:18:35 +0000 UTC", data.TakenAtLocal.String())
 		assert.Equal(t, "2020-05-11 14:18:35 +0000 UTC", data.TakenAt.String())
-		assert.Equal(t, "", data.TimeZone)
+		assert.Equal(t, time.UTC.String(), data.TimeZone)
 		assert.Equal(t, 270, data.Width)
 		assert.Equal(t, 480, data.Height)
 		assert.Equal(t, 270, data.ActualWidth())
@@ -72,8 +73,8 @@ func TestJSON(t *testing.T) {
 
 		assert.Equal(t, CodecAvc1, data.Codec)
 		assert.Equal(t, "2s", data.Duration.String())
-		assert.Equal(t, "2020-05-11 14:16:48 +0000 UTC", data.TakenAtLocal.String())
-		assert.Equal(t, "2020-05-11 12:16:48 +0000 UTC", data.TakenAt.String())
+		assert.Equal(t, "2020-05-11 16:16:48 +0000 UTC", data.TakenAtLocal.String())
+		assert.Equal(t, "2020-05-11 14:16:48 +0000 UTC", data.TakenAt.String())
 		assert.Equal(t, "Europe/Berlin", data.TimeZone)
 		assert.Equal(t, 1920, data.Width)
 		assert.Equal(t, 1080, data.Height)
@@ -99,8 +100,8 @@ func TestJSON(t *testing.T) {
 
 		assert.Equal(t, CodecAvc1, data.Codec)
 		assert.Equal(t, "4s", data.Duration.String())
-		assert.Equal(t, "2020-05-14 11:34:41 +0000 UTC", data.TakenAtLocal.String())
-		assert.Equal(t, "2020-05-14 09:34:41 +0000 UTC", data.TakenAt.String())
+		assert.Equal(t, "2020-05-14 13:34:41 +0000 UTC", data.TakenAtLocal.String())
+		assert.Equal(t, "2020-05-14 11:34:41 +0000 UTC", data.TakenAt.String())
 		assert.Equal(t, "Europe/Berlin", data.TimeZone)
 		assert.Equal(t, 1920, data.Width)
 		assert.Equal(t, 1080, data.Height)
@@ -870,4 +871,22 @@ func TestJSON(t *testing.T) {
 		assert.Equal(t, 1, data.Orientation)
 		assert.Equal(t, "", data.Projection)
 	})
+
+	t.Run("pxl-mp4.json", func(t *testing.T) {
+		data, err := JSON("testdata/pxl-mp4.json", "")
+
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		assert.Equal(t, "2021-07-12T22:56:37Z", data.TakenAt.Format("2006-01-02T15:04:05Z"))
+		assert.Equal(t, "2021-07-12T22:56:37Z", data.TakenAtLocal.Format("2006-01-02T15:04:05Z"))
+		assert.Equal(t, time.UTC.String(), data.TimeZone)
+		assert.Equal(t, 1080, data.Height)
+		assert.Equal(t, 1920, data.Width)
+		assert.Equal(t, float32(0), data.Lat)
+		assert.Equal(t, float32(0), data.Lng)
+		assert.Equal(t, 0, data.Altitude)
+		assert.Equal(t, 1, data.Orientation)
+	})
 }
diff --git a/internal/meta/testdata/pxl-mp4.json b/internal/meta/testdata/pxl-mp4.json
new file mode 100644
index 000000000..a9ee7bde9
--- /dev/null
+++ b/internal/meta/testdata/pxl-mp4.json
@@ -0,0 +1,71 @@
+[{
+  "SourceFile": "PXL_20210712_225625784.mp4",
+  "ExifToolVersion": 12.16,
+  "FileName": "PXL_20210712_225625784.mp4",
+  "Directory": ".",
+  "FileSize": "30 MiB",
+  "FileModifyDate": "2021:07:13 00:56:37+02:00",
+  "FileAccessDate": "2021:07:13 14:15:29+02:00",
+  "FileInodeChangeDate": "2021:07:13 14:12:26+02:00",
+  "FilePermissions": "rw-r-----",
+  "FileType": "MP4",
+  "FileTypeExtension": "mp4",
+  "MIMEType": "video/mp4",
+  "MajorBrand": "MP4 v2 [ISO 14496-14]",
+  "MinorVersion": "0.0.0",
+  "CompatibleBrands": ["isom","mp42"],
+  "MediaDataSize": 30925775,
+  "MediaDataOffset": 40,
+  "MovieHeaderVersion": 0,
+  "CreateDate": "2021:07:12 22:56:37",
+  "ModifyDate": "2021:07:12 22:56:37",
+  "TimeScale": 10000,
+  "Duration": "11.15 s",
+  "PreferredRate": 1,
+  "PreferredVolume": "100.00%",
+  "PreviewTime": "0 s",
+  "PreviewDuration": "0 s",
+  "PosterTime": "0 s",
+  "SelectionTime": "0 s",
+  "SelectionDuration": "0 s",
+  "CurrentTime": "0 s",
+  "NextTrackID": 3,
+  "AndroidCaptureFps": 30,
+  "TrackHeaderVersion": 0,
+  "TrackCreateDate": "2021:07:12 22:56:37",
+  "TrackModifyDate": "2021:07:12 22:56:37",
+  "TrackID": 1,
+  "TrackDuration": "11.11 s",
+  "TrackLayer": 0,
+  "TrackVolume": "100.00%",
+  "Balance": 0,
+  "AudioFormat": "mp4a",
+  "AudioChannels": 1,
+  "AudioBitsPerSample": 16,
+  "AudioSampleRate": 48000,
+  "MatrixStructure": "1 0 0 0 1 0 0 0 1",
+  "ImageWidth": 1920,
+  "ImageHeight": 1080,
+  "MediaHeaderVersion": 0,
+  "MediaCreateDate": "2021:07:12 22:56:37",
+  "MediaModifyDate": "2021:07:12 22:56:37",
+  "MediaTimeScale": 90000,
+  "MediaDuration": "11.15 s",
+  "HandlerType": "Video Track",
+  "HandlerDescription": "VideoHandle",
+  "GraphicsMode": "srcCopy",
+  "OpColor": "0 0 0",
+  "CompressorID": "avc1",
+  "SourceImageWidth": 1920,
+  "SourceImageHeight": 1080,
+  "XResolution": 72,
+  "YResolution": 72,
+  "BitDepth": 24,
+  "PixelAspectRatio": "65536:65536",
+  "ColorRepresentation": "nclx 5 1 6",
+  "VideoFrameRate": 30.051,
+  "ImageSize": "1920x1080",
+  "Megapixels": 2.1,
+  "AvgBitrate": "22.2 Mbps",
+  "Rotation": 0
+}]
\ No newline at end of file
diff --git a/internal/photoprism/convert.go b/internal/photoprism/convert.go
index 1002a93ce..237dcba06 100644
--- a/internal/photoprism/convert.go
+++ b/internal/photoprism/convert.go
@@ -144,7 +144,7 @@ func (c *Convert) ToJson(f *MediaFile) (jsonName string, err error) {
 
 	log.Debugf("exiftool: extracting metadata from %s", relName)
 
-	cmd := exec.Command(c.conf.ExifToolBin(), "-m", "-api", "QuickTimeUTC", "-api", "LargeFileSupport", "-j", f.FileName())
+	cmd := exec.Command(c.conf.ExifToolBin(), "-m", "-api", "LargeFileSupport", "-j", f.FileName())
 
 	// Fetch command output.
 	var out bytes.Buffer