Create timestamp synchronization test #14100

AviaAv · 2025-06-26T09:39:36Z

Tracked on: [LRS-1275]

Nir-Az · 2025-06-26T09:52:00Z

unit-tests/live/metadata/test-timestamps.py

+            ir1_frame_ts = ir1_frame.get_frame_metadata(rs.frame_metadata_value.frame_timestamp)
+            ir2_frame_ts = ir2_frame.get_frame_metadata(rs.frame_metadata_value.frame_timestamp)
+
+            log.d(f"Depth TS: {depth_ts}, IR1 TS: {ir1_ts}, IR2 TS: {ir2_ts}")


Please add color stream

And check with an acceptable tolerance, see the Jira ticket for bad and good numbers

You know what, since the original bug was for IR + Depth that's OK to keep as is.

Do we expect the exactly same TS?

Nir-Az · 2025-07-01T13:20:07Z

DO we know why the CI failed?
We expect it to pass right?

Copilot

Pull Request Overview

This PR creates a new test to verify timestamp consistency between different frame types (depth, infrared, and color) from RealSense devices. The test validates both global timestamps and frame metadata timestamps to ensure proper synchronization.

Adds a comprehensive timestamp consistency test for D400 series devices
Validates synchronization between depth, infrared (channels 1 & 2), and color streams
Includes tolerance checks for both global and frame-level timestamps

Copilot · 2025-07-29T12:00:34Z

unit-tests/live/metadata/test-timestamps.py

+            ir2_frame = frames.get_infrared_frame(2)
+            color_frame = frames.get_color_frame()
+
+            if not (depth_frame and ir1_frame and ir2_frame):


The condition doesn't check for color_frame validity, but color_frame is used later in the code (lines 41, 44, 49, 53, 59, 62, 67). This could cause AttributeError if color_frame is None.

Suggested change

if not (depth_frame and ir1_frame and ir2_frame):

if not (depth_frame and ir1_frame and ir2_frame and color_frame):

Copilot · 2025-07-29T12:00:35Z

unit-tests/live/metadata/test-timestamps.py

+            test.check(depth_ts == ir1_ts)
+            test.check(depth_ts == ir2_ts)


Exact equality comparison for timestamps may be too strict and could lead to flaky tests due to minor timing variations. Consider using approximate equality with a small tolerance instead.

Suggested change

test.check(depth_ts == ir1_ts)

test.check(depth_ts == ir2_ts)

test.check_approx_abs(depth_ts, ir1_ts, GLOBAL_TS_TOLERANCE)

test.check_approx_abs(depth_ts, ir2_ts, GLOBAL_TS_TOLERANCE)

Copilot · 2025-07-29T12:00:35Z

unit-tests/live/metadata/test-timestamps.py

+
+            # Check global timestamps
+            test.check(depth_ts == ir1_ts)
+            test.check(depth_ts == ir2_ts)


Exact equality comparison for timestamps may be too strict and could lead to flaky tests due to minor timing variations. Consider using approximate equality with a small tolerance instead.

Suggested change

test.check(depth_ts == ir2_ts)

test.check_approx_abs(depth_ts, ir2_ts, GLOBAL_TS_TOLERANCE)

Copilot · 2025-07-29T12:00:35Z

unit-tests/live/metadata/test-timestamps.py

+                test.check(depth_frame_ts == ir1_frame_ts)
+                test.check(depth_frame_ts == ir2_frame_ts)


Exact equality comparison for frame timestamps may be too strict and could lead to flaky tests due to minor timing variations. Consider using approximate equality with a small tolerance instead.

Suggested change

test.check(depth_frame_ts == ir1_frame_ts)

test.check(depth_frame_ts == ir2_frame_ts)

test.check_approx_abs(depth_frame_ts, ir1_frame_ts, FRAME_TS_TOLERANCE)

test.check_approx_abs(depth_frame_ts, ir2_frame_ts, FRAME_TS_TOLERANCE)

Copilot · 2025-07-29T12:00:36Z

unit-tests/live/metadata/test-timestamps.py

+                test.check(depth_frame_ts == ir1_frame_ts)
+                test.check(depth_frame_ts == ir2_frame_ts)


Exact equality comparison for frame timestamps may be too strict and could lead to flaky tests due to minor timing variations. Consider using approximate equality with a small tolerance instead.

Suggested change

test.check(depth_frame_ts == ir1_frame_ts)

test.check(depth_frame_ts == ir2_frame_ts)

test.check_approx_abs(depth_frame_ts, ir1_frame_ts, FRAME_TS_TOLERANCE)

test.check_approx_abs(depth_frame_ts, ir2_frame_ts, FRAME_TS_TOLERANCE)

remibettan · 2025-07-29T12:12:51Z

unit-tests/live/metadata/test-timestamps.py

+import pyrealsense2 as rs
+from rspy import test, log
+
+# This test is checking that timestamps of depth and infrared frames are consistent


please add about color

unit-tests/live/metadata/test-timestamps.py

remibettan

few comments - impressive python!

remibettan

LGTM

Nir-Az · 2025-07-31T10:57:17Z

unit-tests/live/metadata/test-timestamps.py

+
+# Tolerance for gap between depth and color
+GLOBAL_TS_TOLERANCE = 1  # in ms
+FRAME_TS_TOLERANCE = 100  # in microseconds


Is that too strict? where did you get 100 us from?

Nir-Az · 2025-07-31T11:07:52Z

@remibettan since I see the tests failed on CI I removed your approval.
Let's approve for merge when the test run and pass

added sleep, enable global ts if needed

AviaAv requested a review from Nir-Az June 26, 2025 09:39

Nir-Az reviewed Jun 26, 2025

View reviewed changes

Nir-Az requested review from Copilot and remibettan July 29, 2025 11:59

Copilot AI reviewed Jul 29, 2025

View reviewed changes

remibettan reviewed Jul 29, 2025

View reviewed changes

unit-tests/live/metadata/test-timestamps.py Show resolved Hide resolved

remibettan reviewed Jul 29, 2025

View reviewed changes

remibettan self-requested a review July 31, 2025 10:31

remibettan approved these changes Jul 31, 2025

View reviewed changes

Nir-Az reviewed Jul 31, 2025

View reviewed changes

Nir-Az requested a review from remibettan July 31, 2025 11:07

AviaAv force-pushed the timestamp-test branch from 7d3d371 to 989a840 Compare July 31, 2025 11:50

AviaAv force-pushed the timestamp-test branch from 21574f2 to ceec0be Compare August 13, 2025 12:49

AviaAv added 3 commits August 14, 2025 15:34

Create test-timestamps.py

21750f9

add color ts and does support check

43d48c7

increase tolerance, specify color resolution VGA

88ae985

added sleep, enable global ts if needed

AviaAv force-pushed the timestamp-test branch from ceec0be to 88ae985 Compare August 14, 2025 12:34

AviaAv requested a review from Nir-Az August 18, 2025 06:28

Nir-Az approved these changes Aug 18, 2025

View reviewed changes

Nir-Az changed the title ~~Create timestamp test~~ Create timestamp synchronization test Aug 18, 2025

Nir-Az merged commit 204572f into realsenseai:development Aug 20, 2025
25 of 26 checks passed

MartyG-RealSense mentioned this pull request Aug 20, 2025

Question about Realsense D415 frame timestamp synchronization and internal clock accuracy #14223

Closed

	if not (depth_frame and ir1_frame and ir2_frame):
	if not (depth_frame and ir1_frame and ir2_frame and color_frame):

		test.check(depth_ts == ir1_ts)
		test.check(depth_ts == ir2_ts)

	test.check(depth_ts == ir2_ts)
	test.check_approx_abs(depth_ts, ir2_ts, GLOBAL_TS_TOLERANCE)

		test.check(depth_frame_ts == ir1_frame_ts)
		test.check(depth_frame_ts == ir2_frame_ts)

Create timestamp synchronization test #14100

Create timestamp synchronization test #14100

Uh oh!

Conversation

AviaAv commented Jun 26, 2025

Uh oh!

Nir-Az Jun 26, 2025

Choose a reason for hiding this comment

Uh oh!

Nir-Az Jun 26, 2025

Choose a reason for hiding this comment

Uh oh!

Nir-Az Jun 26, 2025

Choose a reason for hiding this comment

Uh oh!

Nir-Az commented Jul 1, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Uh oh!

Copilot AI Jul 29, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jul 29, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jul 29, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jul 29, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jul 29, 2025

Choose a reason for hiding this comment

Uh oh!

remibettan Jul 29, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

remibettan left a comment

Choose a reason for hiding this comment

Uh oh!

remibettan left a comment

Choose a reason for hiding this comment

Uh oh!

Nir-Az Jul 31, 2025

Choose a reason for hiding this comment

Uh oh!

Nir-Az commented Jul 31, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants