logo

Live Production Software Forums


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

Notification

Icon
Error

Options
Go to last post Go to first unread
Speegs  
#1 Posted : Monday, September 11, 2017 11:54:09 PM(UTC)
Speegs

Rank: Advanced Member

Groups: Registered
Joined: 8/3/2013(UTC)
Posts: 405
Location: Gold Coast, Australia

Thanks: 27 times
Was thanked: 76 time(s) in 58 post(s)
Hello,

I've put a bunch of tests on the Streaming Idiots Facebook Group:

https://www.facebook.com/groups/StreamingIdiots/

I now have tested both side by side.

Impressions so far:

I like both of them. If you are on a budget the HDMI Spark (or want some wifi) is going to appeal to you. If you want to be a pro the Birddog Studio is going to be appealing. With many things to come in future firmware versions I can see them both being excellent.

Web Interface: Spark is better

Low Latency: Birddog is better

Completeness of firmware: Spark is better

Coming features via Firmware: Birddog looks better

CPU usage in Vmix: Birddog is better

LAN usage in Vmix: Spark is better

POE Included: Birddog is better

Power Brick Design: Spark is better (nice screw in connection)

Tally: Birddog is bigger (IMO Better), both work well

This is by no means exhaustive, if you have an area where you want to figure out what is best for you feel free to ask my opinion. These are all my opinions and I'm not paid to promote either product. Keep up the good work guys because both products are GREAT considering this is new firmware and no doubt will need some time to mature.

I'm yet to figure out a video picture quality test, to my eye so far without freezing frames or looking at super high motion (jellyfish video) I can't tell the difference. I guess I need to put some really high quality car racing footage or something on to compare that. Birddog *should* look better being 110mbit vs Spark which is about 11mbit on high.

thanks 5 users thanked Speegs for this useful post.
ask on 9/12/2017(UTC), mjgraves on 9/12/2017(UTC), SportsNetUSA.net on 9/13/2017(UTC), lael on 9/20/2017(UTC), RobLambert on 11/2/2017(UTC)
DanMiall  
#2 Posted : Tuesday, September 12, 2017 5:57:19 AM(UTC)
DanMiall

Rank: Member

Groups: Registered
Joined: 7/24/2017(UTC)
Posts: 15
Location: Melbourne, Australia

Was thanked: 19 time(s) in 7 post(s)
Hi Speegs,

I can only speak for BirdDog here, but we think anything and everything NDI is very exciting!

Just a note that the pre-release units that have been sent out (including yours) are not running what will be the 'shipping' firmware. The shipping firmware will be lower latency again than what BirdDog is now, and will support all the 'core' features we announced way back when - SDI or HDMI input up to 1080p60, Video loop with cross conversion, Tally, PoE, 8 Channels embedded Audio, plus some more we will be announcing at IBC for immediate release along with some exciting 'bonus' features coming shortly to a BirdDog near you.
thanks 1 user thanked DanMiall for this useful post.
ask on 9/12/2017(UTC)
Speegs  
#3 Posted : Wednesday, September 13, 2017 2:49:28 AM(UTC)
Speegs

Rank: Advanced Member

Groups: Registered
Joined: 8/3/2013(UTC)
Posts: 405
Location: Gold Coast, Australia

Thanks: 27 times
Was thanked: 76 time(s) in 58 post(s)
Read on Facebook the spark has trouble on Mac Computers at the moment. Doubt that will worry a Vmix user :)
Videoguy16x9  
#4 Posted : Wednesday, September 13, 2017 9:15:12 AM(UTC)
Videoguy16x9

Rank: Member

Groups: Registered
Joined: 8/13/2017(UTC)
Posts: 17
Location: Minnesota, United States

Thanks: 13 times
Was thanked: 3 time(s) in 2 post(s)
Thank you for the info Speegs, I just got my Spark and Birddog test units yesterday, so this is very timely.

I have noticed with our test computers (we are building a 10 core i9 in the near future) that the "safe" formula for deployment is the number of physical cores on the computer minus one (n-1). Has this held true for you?

Our quad core i7 has "handled" up to seven inputs (four cameras a quarter mile away at the soccer field plus two NDI graphics sources) but the frame rates were anything but steady. Both the broadcasts and recordings have dropped frames and jerks.

Testing yesterday we were able to get consistently good results as long as we did not exceed three cameras or two cameras and an NDI graphics source.

I plan on taking my personal AMD Ryzen 8 core (just built a little over a month ago) in to the university on Wednesday to test and see - if the formula holds true, I should be able to handle five cameras and a pair of NDI graphics feeds fine, but an eighth source should begin dropping frames and a ninth source should really strain the system.

By the way with six NDI sources on the quad core (eight virtual cores) with 16GB RAM, I saw CPU use hit around 86% and Ram was consistently in the 56 to 64% range.

will update again when we get further along with testing if anyone is interested.
Videoguy16x9  
#5 Posted : Wednesday, September 13, 2017 11:29:28 AM(UTC)
Videoguy16x9

Rank: Member

Groups: Registered
Joined: 8/13/2017(UTC)
Posts: 17
Location: Minnesota, United States

Thanks: 13 times
Was thanked: 3 time(s) in 2 post(s)
By the way, I ran the Sparks and Birddogs through the studio waveform and vector scopes, and although they are very close, I do give a slight edge to the Birddogs.
thanks 1 user thanked Videoguy16x9 for this useful post.
Speegs on 9/14/2017(UTC)
Speegs  
#6 Posted : Thursday, September 14, 2017 1:51:27 AM(UTC)
Speegs

Rank: Advanced Member

Groups: Registered
Joined: 8/3/2013(UTC)
Posts: 405
Location: Gold Coast, Australia

Thanks: 27 times
Was thanked: 76 time(s) in 58 post(s)
Videoguy16x9 wrote:

I have noticed with our test computers (we are building a 10 core i9 in the near future) that the "safe" formula for deployment is the number of physical cores on the computer minus one (n-1). Has this held true for you?

Our quad core i7 has "handled" up to seven inputs (four cameras a quarter mile away at the soccer field plus two NDI graphics sources) but the frame rates were anything but steady. Both the broadcasts and recordings have dropped frames and jerks.


Your results seem about right to me, Video card, SSD and which capture devices you use some are more CPU friendly than others will all make a difference. I also stream via an external box, record to .TS MPEG2 mostly on Vmix because it seems to be low cpu (bigger files).

Finally frame rates make quite a difference. If you are not doing fast motion and are ok with 25 or 30fps. Use it, lower frame rates really help.

I don't use ISO recorder or Replay. Seeing approximately 8-9 cameras, plus 1 desktop capture, 1080p25, Magewell Capture devices, GTX 1060, streaming external box, .TS MPEG2 20mbit recording. About 40-60% CPU.

Mostly I use Vmix to switch and rely on some other things to do jobs.

I will be adding NDI to my workflow it's going to use more CPU no doubt and I'm not going to throw away those Magewells. As they get too old, I expect to be "replacing" Magewell capture devices with NDI. Some of my Magewell units are really old, all still work however.
lael  
#7 Posted : Wednesday, November 15, 2017 8:52:18 PM(UTC)
lael

Rank: Advanced Member

Groups: Registered
Joined: 2/18/2017(UTC)
Posts: 75

Thanks: 22 times
Was thanked: 4 time(s) in 4 post(s)
Early on (march) Martin posted that the Ryzen platform was showing high render times under moderate load - what did you see when testing your Ryzen 8?


Videoguy16x9 wrote:
Thank you for the info Speegs, I just got my Spark and Birddog test units yesterday, so this is very timely.

I have noticed with our test computers (we are building a 10 core i9 in the near future) that the "safe" formula for deployment is the number of physical cores on the computer minus one (n-1). Has this held true for you?

Our quad core i7 has "handled" up to seven inputs (four cameras a quarter mile away at the soccer field plus two NDI graphics sources) but the frame rates were anything but steady. Both the broadcasts and recordings have dropped frames and jerks.

Testing yesterday we were able to get consistently good results as long as we did not exceed three cameras or two cameras and an NDI graphics source.

I plan on taking my personal AMD Ryzen 8 core (just built a little over a month ago) in to the university on Wednesday to test and see - if the formula holds true, I should be able to handle five cameras and a pair of NDI graphics feeds fine, but an eighth source should begin dropping frames and a ninth source should really strain the system.

By the way with six NDI sources on the quad core (eight virtual cores) with 16GB RAM, I saw CPU use hit around 86% and Ram was consistently in the 56 to 64% range.

will update again when we get further along with testing if anyone is interested.

Users browsing this topic
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.