Progressive Latency Issue with NDI HX3 PTZ Cameras on TriCaster Mini 4K (Resolves Temporarily After PVW Switching)

[Phily]

I have 7 PTZ cameras of the same model, configured with a resolution of 4K 50P and output via NDI HX3 (version 6.2).
These cameras are connected to the same switch (model: H3C S5024X-HPWR-EI) and powered by PoE.
Similarly, I have connected my TriCaster Mini 4K to this switch, and all devices can achieve Gigabit speeds.

The problem I am encountering is: When no switching operations are performed, the PTZ camera feeds displayed in TriCaster exhibit extremely high latency, which increases progressively over time. However, the signal sources in the PGM and PVW windows remain unaffected.

The latency returns to normal only after I sequentially switch all PTZ input signals in the PVW window.
But if I keep the current signal source without any switching, the latency will start increasing again as time passes.

Could you please explain why this issue occurs and how I can resolve it?

 

[FrankPW]

my initial thinking:
7 PTZ 4k 50P streams on a Gigabit Ethernet setup could be the challenging factor in this your worflow.
I´d suggest investigate your NDI HX3 streams on your network setup using this tool from NDI

Analysis – Monitor, Debug, And Connect Better – Download Now

NDI Free Audio is a command-line utility that enables you to use any device with a soundcard to input or output NDI audio signals.

ndi.video

The Preview NDI video streams on TriCaster product units should be low res streams over the network, if your PTZ devices do support this NDI workflow feature.
This will change to a high res NDI stream as soon as you switch one of your PTZ sources to Program.

 

[kanep]

That is what I was thinking as well. Could the proxy (low res) NDI stream from this camera be slow, and then it resets when you switch it to Program. What is the model of the PTZ camera?

 

[Phily]

That is what I was thinking as well. Could the proxy (low res) NDI stream from this camera be slow, and then it resets when you switch it to Program. What is the model of the PTZ camera?

Hi, it's great to see both of your replies.

I made the following changes and re-run the tests, but the issue still persists:

1. Set all PTZ cameras to 1080 50P resolution with H.264 encoding;

2. Connected the 10G Ethernet port of the switch to the 2.5G Ethernet port of the Mini 4K using a media converter;

3. Changed the session settings of the Mini 4K to 1080 50P.

Test results show that there is almost no latency for the sources on PGM and PVW, while the latency of other sources gradually increases over time.
For example: When I keep the feed from PTZ Camera 7 (Input 7) on both PGM and PVW, the latency of PTZ Cameras 1-6 (Inputs 1-6) becomes increasingly significant over time, and their latency appears to be consistent. (However, at 4K 50P resolution, their latency varies in duration.)

When I switch each input source to PGM and PVW individually, the latency only disappears after 2-3 switches. At this point, all PTZ feeds appear synchronized and latency-free. If no further operations are performed, the sources not displayed on PGM or PVW will still experience gradually increasing latency over time.

Attached are the results from my NDI Analysis check.
According to the analysis, all latencies appear almost identical.
However, I'm unsure why they seem unsynchronized on the Mini 4K. Could this be related to the two monitors I have connected?
Both are set to a resolution of 1920*1080.
The version of my TriCaster Mini 4K is: 8-5-251202C V6.3.0.1
Just to add, my PTZ cameras are of the model BOLIN R7-420NX, and here is their official website:

https://bolin-av.com/product/r7-420nx-series-4k-ndi-ptz-camera/#tab-8

 

[FrankPW]

hmm ...,

1) 8-5-251202 is pretty new - did you see same issue with any previous SW build(s) ?

2) Maybe you want to contact BOLIN if they indeed did implememt already this NDI low res output stream feature on this PTZ model, which TriCaster will pick up then from the network for the UI Multiview windows. I think to remember some earlier posts on this topic when new PTZ models were released to market.

Edited:
but your NDI Analysis print out still states:
...
11:27:52.761: Video format changed. 3840x2160, codec=H.264, progressive, aspect ratio=1.78, frame-rate=50.00, no alpha channel.
...

plus
....
11:27:52.700: Source is using NDI 6.2.0.0 on the LINUX platform.
....
don´t know if this may be also an issue depending on the NDI SDK (LINUX ?) available to BOLIN for their NDI Support feature for these PTZ models.

And btw, did you update Nvidia driver on your Mini 4K too?
On my TC2E I did install driver version 581.80 (4Nov2025) on top of the 8-5-251202 update for GeForce RTX 3060 card in my TC2E unit.
And one more - don´t forget WIN(10?) updates available until that TriCaster build date.

P.S.
Yes this 8-5-251202 build does include new NDI Runtime libraries 6.3.x.x - first time I see NDI 6.3 in action.
P.S, P.S - so a new update for NDI Tools should be then around the corner soon - too.

 

[kanep]

I agree, it doesn't appear that the cameras are set to output 1080p50 as you are thinking. NDI Analysis says it is a 4K/UHD image size arriving from the camera.

Something else that might be worth checking is the Sync mode. The Mini 4K defaults to GPU, but I've found that using the Internal clock sometimes works better (depend on the monitors you have connected to the system). Go to SETUP, then Sync and change the Reference Clock to Internal (System Clock) and see if that makes any difference.

 

[FrankPW]

Hey Kane,

Just realized all 7 PTZ are on same port 5961, while e.g. all of the Mini-4K NDI Outputs are on different ports.
Is this indeed normal? I thought (I learned in the past) each NDI stream should get his own port?


Edited:
and
... (e.g.)
11:43:41.710: Video data rate (Mbps). Avg=123.24
...
this looks to me all of these 7 PTZ IP-streams indeed are still High-Bandwidth NDI streams, not HX3(2) H.264, HEVC / Proxy, correct?

 

[kanep]

Hey Kane,

Just realized all 7 PTZ are on same port 5961, while e.g. all of the Mini-4K NDI Outputs are on different ports.
Is this indeed normal? I thought (I learned in the past) each NDI stream should get his own port?

If you are talking about the port used on the camera, this is not a problem as each camera has it's own IP address. Since they are each creating a single NDI signal, it likely they will all use 5961 (next available port after 5960).

Edited:
and
... (e.g.)
11:43:41.710: Video data rate (Mbps). Avg=123.24
...
this looks to me all of these 7 PTZ IP-streams indeed are still High-Bandwidth NDI streams, not HX3(2) H.264, HEVC / Proxy, correct?

View attachment 221907

In the Analysis log, the Video Format lists the codec as H.264, so it is definitely a HX2 or HX3 signal. While SpeedHQ is pretty set by the NDI SDK, H.264 and HEVC values can be adjusted by the manufacture, the values listed above are recommended/typical values, but some devices can be different. This Bolin camera is listing its resolution as UHDp50 (3840x2160) in Analysis, in that case the values for an HX3 signal aren't that far apart (92mBit vs 123mBit).

 

[Phily]

Hi Kane and Frank：

Thank you again for your explanations—they’ve been incredibly helpful for my troubleshooting and testing.

I think my issue is now resolved. After testing over the past two days, I discovered the problem was caused by the frame rate of the proxy streams output by the PTZ cameras.

During this process, I tried various solutions: replacing the switch, using 2.5G Ethernet ports with higher transmission speeds, powering each camera independently, updating and rolling back the TriCaster software version, lowering the output resolution, updating the graphics card driver, and updating the Windows operating system. However, none of these fixed the signal delay issue.

Finally, when I changed the PTZ cameras’ output resolution to 4K 30P, I noticed their proxy streams switched from 25fps to 30fps. I then tested with a 4K 29.97 session, and the delay between the cameras and the session disappeared completely.

After that, while keeping the proxy stream frame rate at 30fps, I tested the cameras with outputs set to 4K 50P and 4K 30P, paired with TriCaster sessions in 4K 50P, 4K 25P, and 4K 29.97. In all cross-tests, no delay occurred.

But when I changed the sub-stream (proxy stream) frame rate back to 25fps, the delay gradually became more noticeable over time.

Anyway, my problem is now solved, and I can finally get a good night’s sleep! Thank you both again for your help—especially Frank, who reminded me that TriCaster uses proxy streams for local previews. That gave me the key insight for my troubleshooting!

 

[Phily]

Hi Kane and Frank：

Thank you again for your explanations—they’ve been incredibly helpful for my troubleshooting and testing.

I think my issue is now resolved. After testing over the past two days, I discovered the problem was caused by the frame rate of the proxy streams output by the PTZ cameras.

During this process, I tried various solutions: replacing the switch, using 2.5G Ethernet ports with higher transmission speeds, powering each camera independently, updating and rolling back the TriCaster software version, lowering the output resolution, updating the graphics card driver, and updating the Windows operating system. However, none of these fixed the signal delay issue.

Finally, when I changed the PTZ cameras’ output resolution to 4K 30P, I noticed their proxy streams switched from 25fps to 30fps. I then tested with a 4K 29.97 session, and the delay between the cameras and the session disappeared completely.

After that, while keeping the proxy stream frame rate at 30fps, I tested the cameras with outputs set to 4K 50P and 4K 30P, paired with TriCaster sessions in 4K 50P, 4K 25P, and 4K 29.97. In all cross-tests, no delay occurred.

But when I changed the sub-stream (proxy stream) frame rate back to 25fps, the delay gradually became more noticeable over time.

Anyway, my problem is now solved, and I can finally get a good night’s sleep! Thank you both again for your help—especially Frank, who reminded me that TriCaster uses proxy streams for local previews. That gave me the key insight for my troubleshooting!

View attachment 221908

Just to add a supplement: For this model of PTZ camera, the maximum configurable frame rate of its proxy stream is 30fps, with an adjustable range of 1–30fps. I only tested 25fps and 30fps and haven’t tried other frame rates in between.

 

[FrankPW]

GREAT - and again what learned !
Thank you for this your feedback - and have a good sleep now!