[Issue] Astra 2 - Hardware D2C Alignment fails (640x480 Depth not supported)

[ozsoyu]

Hello Orbbec Team,

I am developing a multi-camera system using Orbbec Astra 2 and Gemini 335L cameras with pyorbbecsdk on Windows 10/11.

My goal is to enable Hardware D2C (Depth to Color) Alignment (OBAlignMode.HW_MODE) on the Astra 2 camera to fix parallax/offset issues on facial landmarks.

The Problem:
To use HW_MODE, I try to set the Depth resolution to 640x480 to match the RGB resolution. However, the SDK throws an error stating that this profile is not supported or does not support hardware D2C.

Code Snippet:

pipeline = Pipeline(device)
config = Config()

# 1. Color Stream
config.enable_stream(OBSensorType.COLOR_SENSOR, 640, 480, 30, OBFormat.RGB)

# 2. Depth Stream (Trying to force 640x480 for Align)
try:
    depth_profile = profiles.get_video_stream_profile(640, 480, OBFormat.Y16, 30)
    config.enable_stream(depth_profile)
except Exception as e:
    print(f"640x480 Depth Failed: {e}")
    # Fallback to Native 400x300
    config.enable_stream(OBSensorType.DEPTH_SENSOR, 400, 300, 30, OBFormat.Y16)

# 3. Hardware Alignment
try:
    config.set_align_mode(OBAlignMode.HW_MODE)
    pipeline.start(config)
except Exception as e:
    print(f"Start Failed: {e}")
Error Log:
code
Code
Cam 3: 640x480 Depth Failed! Invalid input, No matched video stream profile found!
Camera failed to start: Current stream profile is not support hardware d2c process
My Questions:
Does Astra 2 officially support Hardware D2C Alignment (HW_MODE)?
If yes, which Depth Resolution and Format pair must be used to enable it? (It seems 640x480 Depth is not available).
Is there a specific Firmware Version required to unlock 640x480 Depth or HW Alignment for Astra 2?
Environment:
OS: Windows 10/11
SDK: pyorbbecsdk (Latest)
Device: Orbbec Astra 2
Thank you for your support.

[gongye2]

Hi, @ozsoyu

Thank you for your question.
The issue occurs because Astra 2 does not support the 640×480 depth stream profile, so selecting this resolution will always result in an error and hardware D2C cannot be enabled with it.

To use Hardware D2C correctly, please obtain the supported depth profiles for the selected color profile via:

pipeline.get_d2c_depth_profile_list(color_profile, OBAlignMode.HW_MODE)

This API returns all depth configurations that are compatible with hardware D2C.
You may choose one from this list to start the pipeline successfully.

For reference, you can check the example in the v2-main branch:

examples/hw_d2c_align.py

We hope this helps. Please feel free to reach out if you have any further questions.

[ozsoyu]

Hello Orbbec Team,

I am building a 4-camera synchronization system using 2x Gemini 335L and 2x Astra 2 cameras connected via Orbbec Sync Hub Pro on Windows 10/11, using pyorbbecsdk.

Hardware Setup:

• 2x Gemini 335L (HW Sync Mode)
• 2x Astra 2 (HW Sync Mode disabled due to resolution limits, running Native 400x300)
• Connection: All cameras connected via USB 3.0 Hubs/Ports.

Stream Configuration:
To manage USB bandwidth, I am forcing MJPG format for Color streams:

• Gemini: Color 640x480 (MJPG) @ 30 FPS | Depth 640x400 (Y16) @ 30 FPS
• Astra: Color 640x480 (MJPG) @ 30 FPS | Depth 400x300 (Y16) @ 30 FPS

The Problem:
When attempting to poll frames from all 4 cameras simultaneously using a tight loop, the 4th camera often returns None (No Data) or disconnects visually.

Logs:
I observe the following errors in the console when the drop happens:
[error][WmfUvcDevicePort.cpp:435] setXu failed!
[error][SensorBase.cpp:393] Timestamp anomaly detected
code
Code
My Investigation & Solution:
I noticed that the stability depends heavily on the timeout_ms parameter in pipeline.wait_for_frames(timeout_ms).

1. Failing Scenario:
   If I use a short timeout (e.g., frames = pipeline.wait_for_frames(30)), the SDK cannot retrieve the frame in time for the 4th camera due to USB traffic latency, leading to "NO SIGNAL".

2. Working Scenario:
   If I increase the timeout to 100ms (frames = pipeline.wait_for_frames(100)), the system becomes stable, and all 4 cameras stream correctly without dropping, despite the high load.

Questions:

1. Is setXu failed a direct indicator of USB Bandwidth saturation, or is it related to internal command timeouts within the SDK?
2. For a 4-camera setup (mixed Gemini/Astra) on a single PC, what is the recommended "Safe" timeout value for wait_for_frames? Is 100ms an acceptable practice, or should I use a different threading approach (e.g., callbacks)?

Thank you for your assistance.