logo

Live Production Software Forums


Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

2 Pages12>
Options
Go to last post Go to first unread
seb666  
#1 Posted : Wednesday, May 23, 2018 4:41:08 AM(UTC)
seb666

Rank: Advanced Member

Groups: Registered
Joined: 12/10/2016(UTC)
Posts: 124
Man
Location: Frankfurt, Germany

Thanks: 93 times
Was thanked: 19 time(s) in 14 post(s)
Hi,

I recently saw a test video about the Newtek PTZ cam. They compared the latency of SDI with NDI on the Newtek cam with a Tricaster. The result was: it's about the same.

I own three NDI Sparks and a Panasonic AW UE70 with NDI update. When I compare the NDI inputs in vMix with SDI via Blackmagic Decklink, I can't say, that it's the same. There is a delay about 5 PAL frames between SDI and NDI.

Can I improve the delay in vMix? Do NDI|HX inputs in Tricaster have a smaller delay? Does it have something to do with the Newtek camera itself (it shouldn't because the Sparks have the same delay)?

Any help would be very nice! :)

Regards
Sebastian
thanks 1 user thanked seb666 for this useful post.
JoseL on 5/23/2018(UTC)
alexandrelinhares1  
#2 Posted : Wednesday, May 23, 2018 8:22:08 AM(UTC)
alexandrelinhares1

Rank: Advanced Member

Groups: Registered
Joined: 12/22/2016(UTC)
Posts: 47
Location: Brazil

Thanks: 15 times
Was thanked: 1 time(s) in 1 post(s)
seb666 wrote:
Hi,

I recently saw a test video about the Newtek PTZ cam. They compared the latency of SDI with NDI on the Newtek cam with a Tricaster. The result was: it's about the same.

I own three NDI Sparks and a Panasonic AW UE70 with NDI update. When I compare the NDI inputs in vMix with SDI via Blackmagic Decklink, I can't say, that it's the same. There is a delay about 5 PAL frames between SDI and NDI.

Can I improve the delay in vMix? Do NDI|HX inputs in Tricaster have a smaller delay? Does it have something to do with the Newtek camera itself (it shouldn't because the Sparks have the same delay)?

Any help would be very nice! :)

Regards
Sebastian



I have tua same issues in vmix!
JoseL  
#3 Posted : Wednesday, May 23, 2018 4:58:12 PM(UTC)
JoseL

Rank: Advanced Member

Groups: Registered
Joined: 4/15/2018(UTC)
Posts: 64
Man
Location: Spain

Thanks: 6 times
Was thanked: 19 time(s) in 13 post(s)
Sebastian 5 frames difference is really big latency. Can you try any other software?. I’m waiting to update to ndi our 4x ue70 until see a identical latency as sdi.
mjgraves  
#4 Posted : Wednesday, May 23, 2018 4:59:13 PM(UTC)
mjgraves

Rank: Advanced Member

Groups: Registered
Joined: 7/1/2015(UTC)
Posts: 1,150
Man
United States
Location: Houston TX

Thanks: 319 times
Was thanked: 263 time(s) in 233 post(s)
NDI <> NDI|HX.

NDI has less than one frame of delay.
- PTZOptics Gen 3 cameras use NDI.
- BirdDog Studio uses NDI.

NDI|HX adds several frames of delay.
- Newtek Spark uses NDI|HX.
- Newtek PTZ Camera (NewTek NDIHX-PTZ1) uses NDI|HX.
- The Panasonic cameras (AW-HN38, AW-HN40, AW-UN70 and AW-HN130) use NDI|HX.

NDI|HX itself causes the delay. It has nothing to do with vMix.

You can't go back in time. You can only delay OTHER sources to match the video from the NDI|HX sources.

thanks 1 user thanked mjgraves for this useful post.
Peter B on 5/23/2018(UTC)
seb666  
#5 Posted : Wednesday, May 23, 2018 5:45:48 PM(UTC)
seb666

Rank: Advanced Member

Groups: Registered
Joined: 12/10/2016(UTC)
Posts: 124
Man
Location: Frankfurt, Germany

Thanks: 93 times
Was thanked: 19 time(s) in 14 post(s)
mjgraves wrote:
NDI <> NDI|HX.

NDI has less than one frame of delay.
- PTZOptics Gen 3 cameras use NDI.
- BirdDog Studio uses NDI.

NDI|HX adds several frames of delay.
- Newtek Spark uses NDI|HX.
- Newtek PTZ Camera (NewTek NDIHX-PTZ1) uses NDI|HX.
- The Panasonic cameras (AW-HN38, AW-HN40, AW-UN70 and AW-HN130) use NDI|HX.

NDI|HX itself causes the delay. It has nothing to do with vMix.

You can't go back in time. You can only delay OTHER sources to match the video from the NDI|HX sources.



It has got nothing todo with going back in time.
I know that NDI|HX has latency based on compression and decompression, but it seems a Tricaster handles decompression better that vMix on an i9:


Ok, here I have the whole thread from NewTek's forum:
http://forums.newtek.com...hread.php?154681-Latency

Here we have a good latency comparison between SDI and NDI|HX:
https://vimeo.com/236308108/3a1201396f

The NewTek PTZ cam has the same delay on SDI and NDI, and a delay compared to the Sony cam with SDI.
I wouldn't care about 0.05 sec delay. I'm talking about 5 frames of 1080p25 production. That's a delay of 0.2 sec. And I think this can be improved...

But do you have a link to the Gen3-PTZOptics? Because the cams on their website also seem to be NDI|HX...
zenvideo  
#6 Posted : Wednesday, May 23, 2018 6:56:57 PM(UTC)
zenvideo

Rank: Advanced Member

Groups: Registered
Joined: 5/13/2014(UTC)
Posts: 514
Man
United Kingdom
Location: Manchester, UK

Thanks: 2 times
Was thanked: 183 time(s) in 130 post(s)
seb666 wrote:
But do you have a link to the Gen3-PTZOptics? Because the cams on their website also seem to be NDI|HX...

I'm pretty sure all the PTZOptics NDI-capable cameras are NDI|HX...
thanks 1 user thanked zenvideo for this useful post.
seb666 on 5/24/2018(UTC)
DWAM  
#7 Posted : Thursday, May 24, 2018 3:44:38 AM(UTC)
DWAM

Rank: Advanced Member

Groups: Registered
Joined: 3/20/2014(UTC)
Posts: 2,721
Man
France
Location: Bordeaux, France

Thanks: 243 times
Was thanked: 794 time(s) in 589 post(s)
Quote:
- PTZOptics Gen 3 cameras use NDI.

No it's NDI|HX

Remember that NDI|HX is simply some sort of fork of RTSP, that is Long-GOP H.264 encoding (GOP stands for Group Of Pictures) wrapped a little differently to be compatible with the NDI protocol. To me this is a bad joke for "professional" use cases. I never believed it would achieve the promised announced zero latency.

If you want to mix SDI and pure NDI for professional results, just forget about all things NDI|HX and stop believing marketing bullshit ! ;o)

Guillaume
thanks 2 users thanked DWAM for this useful post.
Zetor1980 on 6/1/2018(UTC), Ario on 6/20/2018(UTC)
seb666  
#8 Posted : Thursday, May 24, 2018 7:15:11 AM(UTC)
seb666

Rank: Advanced Member

Groups: Registered
Joined: 12/10/2016(UTC)
Posts: 124
Man
Location: Frankfurt, Germany

Thanks: 93 times
Was thanked: 19 time(s) in 14 post(s)
DWAM wrote:
Quote:
- PTZOptics Gen 3 cameras use NDI.

If you want to mix SDI and pure NDI for professional results, just forget about all things NDI|HX and stop believing marketing bullshit ! ;o)


Can you please stop derailing this thread?

There are latency tests showing a much lower latency with NDI|HX than I experience in my setup at the moment (just open that videos).
And I just want to know, what I can do, to improve latency with NDI|HX on a vMix machine.

Thanks!

P.S.: I know, that there has to more latency than SDI. I know it is not "real NDI". But it has it's use cases.
admin  
#9 Posted : Thursday, May 24, 2018 8:36:59 AM(UTC)
admin

Rank: Administration

Groups: Administrators
Joined: 1/13/2010(UTC)
Posts: 5,137
Man
Location: Gold Coast, Australia

Was thanked: 4135 time(s) in 1487 post(s)
Hi,

vMix uses the same NDI HX driver as the Tricaster and therefore should have pretty close to identical latency on the input side of things.
I think where the confusion lies, is different NDI HX devices have different latency.

I know NewTek's own PTZ camera seems to have quite low latency, but other devices such as the Connect Spark and some 3rd party cameras can be a frame or two more.

For the lowest possible latency make sure the frame rate in vMix matches the camera, so if the camera is sending out a 59.94p signal, set that in vMix as well.

Regards,

Martin
vMix
thanks 3 users thanked admin for this useful post.
mjgraves on 5/24/2018(UTC), seb666 on 5/24/2018(UTC), PFBM on 6/12/2018(UTC)
mjgraves  
#10 Posted : Thursday, May 24, 2018 12:40:16 PM(UTC)
mjgraves

Rank: Advanced Member

Groups: Registered
Joined: 7/1/2015(UTC)
Posts: 1,150
Man
United States
Location: Houston TX

Thanks: 319 times
Was thanked: 263 time(s) in 233 post(s)
The fact that the Newtek PTZ1 camera appears to offer NDI|HX that's actually in sync with its SDI output suggest that they might delay the SDI stream.

It would be interesting to see the camera compared to another camera, feeding the switcher via SDI. And both pointed at a rotary clock.
thanks 1 user thanked mjgraves for this useful post.
seb666 on 5/24/2018(UTC)
Ronald555  
#11 Posted : Thursday, May 24, 2018 1:08:45 PM(UTC)
Ronald555

Rank: Advanced Member

Groups: Registered
Joined: 1/25/2018(UTC)
Posts: 92

Thanks: 28 times
Was thanked: 3 time(s) in 3 post(s)
Are there still PTZ cameras with the original NDI?
Low latency (every frame counts) is much more important for me than a bit less bandwidth use.
seb666  
#12 Posted : Thursday, May 24, 2018 1:25:51 PM(UTC)
seb666

Rank: Advanced Member

Groups: Registered
Joined: 12/10/2016(UTC)
Posts: 124
Man
Location: Frankfurt, Germany

Thanks: 93 times
Was thanked: 19 time(s) in 14 post(s)
mjgraves wrote:
The fact that the Newtek PTZ1 camera appears to offer NDI|HX that's actually in sync with its SDI output suggest that they might delay the SDI stream.

It would be interesting to see the camera compared to another camera, feeding the switcher via SDI. And both pointed at a rotary clock.


I want to buy some PTZ cameras in the next weeks and hope, that I can try out a NewTek PTZ1 at my dealer.
I don't have a problem with some kind of delay. It's just that in my configuration it's over 200 ms. That doesn't work well with joystick control.
seb666  
#13 Posted : Thursday, May 24, 2018 1:27:42 PM(UTC)
seb666

Rank: Advanced Member

Groups: Registered
Joined: 12/10/2016(UTC)
Posts: 124
Man
Location: Frankfurt, Germany

Thanks: 93 times
Was thanked: 19 time(s) in 14 post(s)
admin wrote:
vMix uses the same NDI HX driver as the Tricaster and therefore should have pretty close to identical latency on the input side of things.
I think where the confusion lies, is different NDI HX devices have different latency.

I know NewTek's own PTZ camera seems to have quite low latency, but other devices such as the Connect Spark and some 3rd party cameras can be a frame or two more.


Thanks Martin. That sound's like a good reason to look into the PTZ1 as a camera for joystick control.

Regards
Seb
kane  
#14 Posted : Thursday, May 24, 2018 6:15:28 PM(UTC)
kane

Rank: Advanced Member

Groups: Registered
Joined: 3/24/2016(UTC)
Posts: 331
Location: Chicago, IL

Was thanked: 143 time(s) in 94 post(s)
Let me help answer a few things here.

Currently there are no cameras on the market with full bandwidth NDI built in. NDI|HX is taking existing encoding circuitry that is already built into cameras and re-programming with to output something that is low latency, but due to the LongGOP structure of the format, it will never be as low as full bandwidth NDI. I don't believe that NewTek ever said these units will be zero latency, in fact it is impossible to have zero latency thru any digital device no matter what signal we are talking about.

Once FPGA chips are available (which is what NewTek is doing with the Connect Spark Pro), then there will be something for camera manufactures to build into their designs.

As for NDI|HX latency, that all depends on the device. The PTZ cameras appear to have lower latency, because the compression chips on the unit have quicker access to the video data (right off the sensor) while Spark unit will appear a bit higher, because they don't get the signal until 'after' it comes out of the camera. Even HDMI and SDI outputs have some amount of image processing to get sensor data into a valid HDMI/SDI signal, so while we might call these outputs real-time, they too can be frame or two behind the image coming into the glass.

Attached is a quick latency test I took using a JVC camera via SDI, PTZ1 camera via NDI|HX and a HDMI and SDI Spark units coming off the JVC camera. There is no genlock, so the difference between the JVC and the PTZ1 can have some amount of difference because of that. Video format and frame rate can also cause differences in the latency, overall I recommend going for the highest video frame rate possible. In my test I used 1080p59.94 video input, sent everything into a TriCaster and put all four inputs into an M/E and then did a image grab of the M/E.

Kane Peterson
NewTek
kane attached the following image(s):
Latency Test.jpg (1,536kb) downloaded 88 time(s).

You cannot view/download attachments. Try to login or register.
thanks 6 users thanked kane for this useful post.
seb666 on 5/25/2018(UTC), Ronald555 on 5/25/2018(UTC), mjgraves on 5/25/2018(UTC), Peter B on 5/31/2018(UTC), Weckmuller on 2/10/2020(UTC), PaddyPatPat on 5/5/2021(UTC)
mjgraves  
#15 Posted : Friday, May 25, 2018 6:12:38 PM(UTC)
mjgraves

Rank: Advanced Member

Groups: Registered
Joined: 7/1/2015(UTC)
Posts: 1,150
Man
United States
Location: Houston TX

Thanks: 319 times
Was thanked: 263 time(s) in 233 post(s)
Without genlock I would expect one frame one delay on all sources, just to frame sync the signals.

One frame at 60 f/s is 0.01667 sec.

Looking at that sample image:

- JVC SDI is fastest
- PTZ1 NDI is 0.044 sec behind (2.64 frames)
- Spark is 0.089 sec behind (5.34 frames)

I'd still be interested to know how the NDI output of the PTZ1 compares to its own SDI output.
kane  
#16 Posted : Saturday, May 26, 2018 5:47:00 AM(UTC)
kane

Rank: Advanced Member

Groups: Registered
Joined: 3/24/2016(UTC)
Posts: 331
Location: Chicago, IL

Was thanked: 143 time(s) in 94 post(s)
mjgraves wrote:
I'd still be interested to know how the NDI output of the PTZ1 compares to its own SDI output.


I'll have to test it when I get back home. But I'm half a world away right now, so it will be a little bit.

Kane Peterson
NewTek
kane  
#17 Posted : Thursday, May 31, 2018 12:43:19 PM(UTC)
kane

Rank: Advanced Member

Groups: Registered
Joined: 3/24/2016(UTC)
Posts: 331
Location: Chicago, IL

Was thanked: 143 time(s) in 94 post(s)
Did a quick test of the PTZ1 SDI and NDI side by side. Got the same time into my TriCaster.

I should point out, that I am running beta firmware on my PTZ1. I'm not sure if this makes any difference in performance.
kane attached the following image(s):
Latency PTZ1 SDI vs NDI.jpg (1,023kb) downloaded 46 time(s).

You cannot view/download attachments. Try to login or register.
JoseL  
#18 Posted : Thursday, May 31, 2018 5:34:51 PM(UTC)
JoseL

Rank: Advanced Member

Groups: Registered
Joined: 4/15/2018(UTC)
Posts: 64
Man
Location: Spain

Thanks: 6 times
Was thanked: 19 time(s) in 13 post(s)
Thanks for the test.

Are you sure this clock is good?. A lot of internet stopwatch has 40- 50ms increments between steps.

Whats the latency between stopwatch and vmix monitor?. I ask because some cameras has very high latency on sdi. If you compare a camera with 4 frames latency, sdi or ndi, this is too much latency. Is better to change to a 1 frame latency camera.

Regards

Jose.

kane wrote:
Did a quick test of the PTZ1 SDI and NDI side by side. Got the same time into my TriCaster.

I should point out, that I am running beta firmware on my PTZ1. I'm not sure if this makes any difference in performance.

kane  
#19 Posted : Thursday, May 31, 2018 7:41:47 PM(UTC)
kane

Rank: Advanced Member

Groups: Registered
Joined: 3/24/2016(UTC)
Posts: 331
Location: Chicago, IL

Was thanked: 143 time(s) in 94 post(s)
I have no idea on how good the clock is, just something I found online. Beyond the timing you can get in JavaScript, there is also the refresh speed of the display, which I think for this tablet is 60hz (which is probably higher than what JavaSript timing is capable of) so you are really limited to that not matter what. However, it should give a reasonable idea on latency.
JoseL  
#20 Posted : Saturday, June 2, 2018 2:05:10 PM(UTC)
JoseL

Rank: Advanced Member

Groups: Registered
Joined: 4/15/2018(UTC)
Posts: 64
Man
Location: Spain

Thanks: 6 times
Was thanked: 19 time(s) in 13 post(s)
I tried your clock on my macbook pro and i found that time between each step of this clock is 44ms. 44ms=2,75frames at 1080@60p. I think is better to find a timecode generator or something similar, at least you will get 1frame between steps.
I think most important thing is total latency. If a camera has 5 frames latency on sdi and 6-7 frames latency on NDI, at least for me is too much. I want NDI like all of you, but please try to add total latency of camera on each output. Comparison between both outputs is not enough information in my opinion.

I attach what we usually do when compare latencies.

Regards

Jose.



kane wrote:
I have no idea on how good the clock is, just something I found online. Beyond the timing you can get in JavaScript, there is also the refresh speed of the display, which I think for this tablet is 60hz (which is probably higher than what JavaSript timing is capable of) so you are really limited to that not matter what. However, it should give a reasonable idea on latency.
JoseL attached the following image(s):
loa20x vs ue70.PNG (507kb) downloaded 32 time(s).
Loa20x vs he120.PNG (915kb) downloaded 37 time(s).

You cannot view/download attachments. Try to login or register.
thanks 2 users thanked JoseL for this useful post.
mjgraves on 6/4/2018(UTC), PaddyPatPat on 5/5/2021(UTC)
Users browsing this topic
2 Pages12>
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.