RTSP Stream Compatibility Issues

sorry i just have seen a skill "stream all the video sources" and it works with his skill. Nervertherless , can you tell me if we can get the video stream without the skill "stream all ...." from my first question ?

Great - glad Stream All The Video Sources solved it. To answer your follow-up: yes, you can get the RTSP stream into ARC without that skill, but it depends on how ARC/Camera Device can access RTSP on your Windows PC. Below are the practical options, ordered from easiest to most reliable.

Short answer

• Easiest: keep using Stream All The Video Sources (recommended).
• If you want to avoid that skill: try starting the RTSP URL directly from the Camera Device (ControlCommand), but if you hit the KeyNotFound/moniker error (DirectShow codec error), use one of these workarounds: install a DirectShow codec (LAV Filters) or restream RTSP to an HTTP MJPEG feed (via ffmpeg or VLC) and point Camera Device to the HTTP URL.

Details + stepbystep

1. Try Camera Device directly (fast test)

• From a script or the ARC scripting console you can try:

// start camera using the RTSP URL
ControlCommand("Camera", "CameraStart", "rtsp://192.168.178.10:554/h264Preview_01_sub");

// check whether the camera is active (returns true/false)
var active = ControlCommandWait("Camera", "IsCameraActive");
print("Camera active: " + active);

• If this works, you’re done. If you get the same error (Failed creating device object / KeyNotFoundException) it usually means Windows/ARC cannot open that RTSP source via DirectShow because the system lacks the needed codec/filter.

2. Make Windows able to open RTSP (DirectShow filter approach)

• Install LAV Filters (Open-source DirectShow filters that add RTSP/h264 support) or a trusted codec pack that includes an RTSP-capable DirectShow source.
  • LAV Filters: https://github.com/Nevcairiel/LAVFilters
• After installing, restart your PC and try ControlCommand CameraStart with the RTSP URL again.
• This will let the Camera Device use RTSP directly (no extra skill) if the system DirectShow filter can handle the stream.

3. Restream RTSP to an HTTP MJPEG (most reliable & avoids installing codecs)

• Use ffmpeg or VLC to convert the RTSP feed into an HTTP MJPEG stream that Camera Device accepts as a "JPEG Snapshot HTTP Video Device".
• Example ffmpeg command (listen mode - creates an MJPEG HTTP stream on port 8090):
  • Download ffmpeg: https://ffmpeg.org/download.html

ffmpeg -rtsp_transport tcp -i "rtsp://192.168.178.10:554/h264Preview_01_sub" \
  -f mjpeg -q:v 5 -r 15 -s 640x480 -listen 1 "http://0.0.0.0:8090/";

• Then in Camera Device, choose the "JPEG Snapshot HTTP Video Device" and set the URL to:
  • http://localhost:8090/ (or http://<PC_IP>:8090/ from another machine)
• VLC can also rebroadcast:

"C:\Program Files\VideoLAN\VLC\vlc.exe" "rtsp://192.168.178.10:554/h264Preview_01_sub" \
  --sout "#standard{access=http,mux=ts,dst=:8080}" --no-sout-all --sout-keep

• Then point Camera Device at http://localhost:8080/

Why restreaming is useful

• It sidesteps DirectShow codec issues and is very stable.
• Stream All The Video Sources essentially does this internally (it supports many protocols) - that's why it worked for you.

4. Other options (virtual camera)

• Use OBS + Virtual Camera or other "virtual webcam" drivers to make the RTSP feed appear as a local webcam device. Camera Device can then use that webcam device index. More setup and CPU usage, but possible if you prefer a virtual webcam.

Notes about IoTiny and network

• IoTiny is the EZ-B controller for servos/I/O. The RTSP camera is a separate IP device on your LAN, and ARC (running on your PC) must be able to reach that IP. The IoTiny does not need to be involved in the camera network path.
• Make sure firewall rules on the PC allow ffmpeg/VLC/ARC to bind/listen or connect.

Recommendation

• If you want the simplest supported path and you already have Stream All The Video Sources working - keep using it.
• If you need to avoid that skill, try ControlCommand CameraStart first. If it fails, use the ffmpeg restream to HTTP MJPEG - it’s robust and quick to set up.

Want me to walk you through one of these?

• Do you run ARC on Windows? Which option do you want to try next (direct CameraStart, install LAV filters, ffmpeg restream)? I can give exact, stepbystep instructions for your chosen path.

If this works, you’re done. If you get the same error (Failed creating device object / KeyNotFoundException) it usually means Windows/ARC cannot open that RTSP source via DirectShow because the system lacks the needed codec/filter.

i got the same error. I will stay with the skill "Stream All The Video Sources", it is easier. i will play with camera and see how to get the best tracking by using, objet,glyph,color ...

Great - sounds like a good plan to keep using Stream All The Video Sources if it already works for you. Below are focused tips and quick control commands to help you get the best tracking results (object, glyph, color) with ARC + Camera Device. If you want, I can walk you stepbystep through tuning one tracking type.

Useful links

• Camera Device manual: https://synthiam.com/Support/Skills/Camera/Camera-Device?id=16120
• Stream All The Video Sources manual: https://synthiam.com/Support/Skills/Camera/Stream-All-The-Video-Sources?id=20279
• Glyph PDF (printable): linked in Camera Device manual (see Glyph Tracking)
• Variable Watch skill (to monitor camera variables): https://synthiam.com/Docs/Skills/Scripting/Variable-Watch?id=16056

Quick control commands (JavaScript examples)

• Start camera with the Camera Device (if you ever test direct URLs):

ControlCommand("Camera", "CameraStart", "rtsp://192.168.178.10:554/h264Preview_01_sub");
var active = ControlCommandWait("Camera", "IsCameraActive");
print("Camera active: " + active);

• Toggle preview type (Off / Original / Processed):

ControlCommand("Camera", "CameraSetPreviewType", "Processed");

• Skip display frames (help CPU):

ControlCommand("Camera", "CameraSkipDisplayFrames", 5); // show every 5th frame in preview

• Enable/disable servo or movement tracking:

ControlCommand("Camera", "CameraServoTrackEnable");
ControlCommand("Camera", "CameraServoTrackDisable");
ControlCommand("Camera", "CameraMovementTrackEnable");
ControlCommand("Camera", "CameraMovementTrackDisable");

• Snapshot:

ControlCommand("Camera", "CameraSnapshot"); // auto filename

Practical tuning tips by tracking type

• General (applies to all tracking)

  • Lighting: bright, even lighting and noncluttered background matter more than camera resolution.
  • Resolution/FPS: reduce resolution to 320x240 or 320x180 and 10-15 FPS if CPU is limited. Lower reduces noise and speeds processing.
  • Preview & CPU: if ARC UI slows, set Preview to Off or Processed and use CameraSkipDisplayFrames to save CPU.
  • Use Variable Watch to monitor $Camera* variables while tuning.
  • If camera is on the robot, avoid streaming high-resolution over congested WiFi - reduces frame rate and tracking stability.

• Color / Multi-Color / YCbCr

  • Use Multi-Color or YCbCr when background has similar hues. YCbCr often performs better under varying lighting.
  • Train/define colors using the training tool and adjust Saturation/Luminance/Min Size while holding the object in front of the camera.
  • Increase Minimum Object Size to ignore small noise blobs.
  • If color detection flickers, increase smoothing by requiring detections across multiple frames (the Camera settings have min detection count/delay).

• Object Tracking (AVM)

  • Train only unique details (logo, label, distinct texture), not the entire object. Follow the Camera Device manual’s training workflow: tilt/rotate the object while training.
  • Train from multiple angles and distances.
  • Use "Learn While Tracking" cautiously - it can drift and start learning the background over time.
  • Increase Min Detection Count so it requires several frames before reporting a detection.

• Glyph / QR Code

  • Print glyphs / QR codes at recommended sizes and use a plain background.
  • Glyph detection is reliable but CPU intensive; set a lower resolution and allow more frames for detection.
  • Use CameraClearLastGyph and CameraResetQRCode control commands if you need repeated triggers from the same code.

Camera on robot (pan/tilt)

• If you have pan/tilt servos, configure the Camera Device servos (min/max positions and increments) and use CameraServoTrackRelativeEnable for incremental servo motion.
• Tune horizontal/vertical increment steps to balance smoothness vs responsiveness.

If you want a suggested workflow to tune one tracking type now

1. Pick the tracking type you want to optimize (object / color / glyph).
2. Set resolution to 320x240 and processed preview off (or skip display frames = 3).
3. Configure the Camera skill min object size and detection count conservative (so noise is ignored).
4. Train or adjust color/objects while watching $CameraObjectCenterX/Y and $CameraIsTracking in Variable Watch.
5. Adjust lighting and thresholds until detection is stable.

Would you like me to:

• Walk you stepbystep to tune object tracking for your Reolink camera?
• Provide an example script to auto-start Stream All The Video Sources + Camera when your project loads?
• Or give specific settings (resolution, skip frames, min sizes) for your PC specs?

Also, if you attach screenshots of errors or settings, note that I can’t view images-please paste the exact text or settings you see and I’ll help interpret them.