Motion-dependent ghosting in D455 at low FPS (5 FPS) – disappears at 30 FPS

[camilo10k]

Camera Model:
Intel RealSense D455
Firmware version: 5.16.0.1

SDK Version: 2

We are observing motion-dependent “ghosting” artifacts on a D455 unit when operating at low frame rates (5 FPS).

The artifact appears as a semi-transparent duplicate or trailing shadow of the moving object in the direction of motion. When the conveyor direction is reversed, the ghost appears on the opposite side, which indicates the effect is strictly motion-related and not lighting-direction dependent.

The issue is significantly reduced or disappears when increasing FPS from 5 to 30.

We have another D455 Rugged (also know as D555) unit in the same setup that does not show this behavior under identical test conditions.

Test Setup

Conveyor belt system

Overhead camera mounting

Indoor environment

Controlled lighting

Object moving at constant speed

When reversing belt direction:

Ghost artifact flips direction accordingly.

This suggests motion integration or sensor timing behavior.

resolution:
width: 1280
height: 720
fps: 5

depth:
exposure_time: 10000

color:
enable: true
exposure_time: 16
brightness: 0
contrast: 50
gamma: 300
hue: 0
saturation: 64
sharpness: 50
white_balance: 5000

1. Is this expected behavior at low FPS due to motion integration or rolling shutter?

2. Is there any known firmware difference between D455 and D455 Rugged that could explain this?

3. Could this indicate a hardware issue in this specific unit?

[MartyG-RealSense]

Hi @camilo10k The D455 and D555 both use the D450 Depth Module circuit board for their sensor package. All of the sensors on this module have a fast global shutter that greatly reduces the risk of lag / blur occurring during capture of fast motion (such as fast-moving cars on a road) compared to the slower rolling shutter on the RGB sensor of D415 and the D435 type cameras. But fast motion capture is typically done at higher resolutions than 5 FPS.

When lag does occur, a workaround recommended in the past has been to set the RGB sensor to minimum FPS (5 FPS on D455) and also use manual exposure instead of auto-exposure and set an RGB manual exposure value of 78. This can enable RGB lag to be negated. It has not worked in every case but would be worth trying.

You could also try reducing a lower resolution of 848x480 if possible to see if reduced processing burden resolves the lag, or increasing the FPS from 5 to 15.

D555 has a different firmware driver to D455, and also benefits from a new-generation data processing circuit board with a best-in-class Image Signal Processor (ISP) that enhances the RGB with Geometric Distortion Correction (GDC) and Temporal Noise Reduction (TNR).

[camilo10k]

Thank you very much for the detailed explanation and suggestions.

We appreciate the clarification regarding the D450 depth module and the differences between D455 and D555, especially the information about the updated ISP and firmware differences.

Based on your recommendations, we will perform additional controlled tests:

Set RGB to manual exposure (including testing the suggested value of 78).

Compare behavior at 5 FPS vs 15 and 30 FPS.

Test at reduced resolution (848x480).

Run RGB-only streaming to isolate the effect from the depth pipeline.

.

[MartyG-RealSense]

You are very welcome. Please do update me about your test results later. Good luck!