Camera::open() takes 5-7s for 1920x1200 SVO files with DEPTH_MODE::NONE (50x slower than 1080p)

[caiobarrosv]

Preliminary Checks

• This issue is not a duplicate. Before opening a new issue, please search existing issues.
• This issue is not a question, feature request, or anything other than a bug report directly related to this project.

Description

Camera::open() takes 5-7 seconds when opening SVO2 files recorded at the ZED X native resolution (1920x1200), even with DEPTH_MODE::NONE and camera_disable_self_calib = true. The same camera serial opening a 1920x1080 SVO file completes in 0.1-0.4 seconds — a ~50x difference with no depth or self-calibration involved.

Steps to Reproduce

import pyzed.sl as sl
import time

# Requires two SVO files from the same ZED X camera:
#   - one recorded at 1920x1080
#   - one recorded at 1920x1200 (native resolution)

FILES = {
    "1080p": "/path/to/recording_1080p.svo2",
    "1200p": "/path/to/recording_1200p.svo2",
}

for label, path in FILES.items():
    zed = sl.Camera()
    init = sl.InitParameters()
    init.set_from_svo_file(path)
    init.depth_mode = sl.DEPTH_MODE.NONE
    init.camera_disable_self_calib = True

    t0 = time.time()
    status = zed.open(init)
    elapsed = time.time() - t0

    if status == sl.ERROR_CODE.SUCCESS:
        info = zed.get_camera_information()
        res = f"{info.camera_configuration.resolution.width}x{info.camera_configuration.resolution.height}"
        print(f"{label}: open={elapsed:.2f}s, res={res}, serial={info.serial_number}")
        zed.close()
    else:
        print(f"{label}: FAILED ({status}) in {elapsed:.2f}s")

Expected Result

Camera::open() with DEPTH_MODE::NONE and camera_disable_self_calib = true should complete in under 1 second regardless of resolution, since no depth computation, neural model loading, or self-calibration is requested.

Actual Result

Camera::open() takes 5-7 seconds for 1920x1200 SVO files, even though:

• Depth mode is NONE
• Self-calibration is disabled
• GPU utilization is 0% during the call
• The files are smaller than the fast-opening 1080p files

This creates a significant bottleneck in SVO processing pipelines. In our case, processing 4 SVO files (one per camera angle) takes ~20-26 seconds wall time, with open() accounting for nearly all of it.

ZED Camera model

ZED

Environment

- Platform: x86_64 Linux (Ubuntu 22.04)
- GPU: NVIDIA RTX (CUDA 12.8)
- ZED SDK: 5.2
- Python API: pyzed
- Cameras: 4x ZED X (GMSL2), serials 42380488, 45973857, 46164128, 48026452
- Docker base image: `stereolabs/zed:5.2-devel-cuda12.8-ubuntu22.04`
- SVO files: Recorded at 1920x1200 (native), H265 compression, 5 frames, 30 FPS

Anything else?

Observed Behavior

File	Resolution	Serial	Size	open() time
test_1frame.svo	1920x1080	46164128	14.5 MB	0.39s
prod_0.svo2	1920x1200	46164128	1.2 MB	5.50s
prod_180.svo2	1920x1200	42380488	1.1 MB	4.84s
prod_270.svo2	1920x1200	45973857	1.4 MB	6.27s
prod_90.svo2	1920x1200	48026452	1.1 MB	5.16s

[github-actions]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment otherwise it will be automatically closed in 5 days