Subscriber Discussion

What Camera / Cameras To Cover A Basketball Arena / Concert Hall?

UI
Undisclosed Integrator #1
Jul 12, 2016

I am hoping for some feedback or suggestions regarding the best choice to achieve the following customer request:

"The hope is that we could continuously record the seating bowl in a way that we could later zoom in on any seat and ID a patron or staff. As it stands, you’d have to already be zoomed in and happen to catch the action."

This is an indoor college basketball arena that has countless different events(concerts, etc) in addition to basketball games. they currently have several Axis PTZ's, but realize that there needs to be some additional coverage in order to have the ability to digitally zoom in on a recorded image and potentially identify someone. Flashing stage lights, etc, could make image quality a problem. The campus security department is recommending using a couple of Axis M3006 cameras. If that model # is in fact being relayed to us accurately, I would have reservations about it being a proper solution.

I welcome your input.

(1)
JH
John Honovich
Jul 12, 2016
IPVM

The short answer is no.

Also, we have tested the M3006.

To determine how exactly it would do, we need to know:

  • How wide of an area will each camera cover? 100' wide? more? less?
  • What ppf is needed? Since the requirement is to "later zoom in on any seat and ID a patron or staff", I am guessing 50+ ppf.

With the M3006, it has a very wide 134° AoV, so a 100' width at 3MP, gives ~20ppf, way too low to ID a patron.

(1)
U
Undisclosed #2
Jul 12, 2016

No change the M3006 would work. Great camera, but not the right solution.

A few questions -

1.) What VMS are you using?

2.) Are there limitations on where cameras can be placed?

3.) Do you have approximate dimensions? Some arenas are substantially larger than others, so it's hard to be specific on what would truly allow someone to ID a patron.

Depending on the VMS, you have a couple of options that you could at least look at. The balance or trade-off is going to be quantity of cameras versus quality of cameras. Do you look at placing a couple dozen or more cameras at strategic spots to be able to capture the entire arena, or do you look at using a couple of really high-quality cameras.

Going the multiple-camera route is fairly straightforward. Select your camera of choice, determine where you can position it, ensure the lens provides the FOV you need, and ensure your pixel density is correct.

If you want to go with the fewer camera options, you have a few choices. Just keep in mind that your VMS will play a huge role.

In no particular order

  • Arecont multi-sensor cameras, if supported -- 20MP or 40MP cameras with 4-sensors that will capture a 180-degree panoramic view of an area at 5MP or 10MP apiece. Again, must do FOV/PPF calculations to make sure you're getting ID-level images.

    Concerns about these:

    • Low framerates. I believe the 40MP cameras are only 1.5fps. Trading resolution for framerate, but you also are capturing relatively motionless people.
    • Extremely high bandwidth requirements. Obviously a 40MP camera pushes a ton of bandwidth. Must have network configured properly.
  • Avigilon 16MP or 29MP cameras - single cameras that can cover a very large area and provide high pixel density. Multiple references for these for arenas and stadiums.

    Concerns:

    • Cost. Really expensive.
    • High bandwidth requirements.

I'm sure there are others, but those are first to mind. Once you provide your VMS, we might be able to give more details, especially because the VMS will determine how those cameras are handled/licensed.

(1)
Avatar
Jon Dillabaugh
Jul 12, 2016
Pro Focus LLC

This will be very challenging with differing events. The easiest will be basketball games where the arena is completely used and well lit. Usually when there are concerts, you will have a portion of the behind stage area obstructed (from the center POV) and lighting will be very challenging.

My best advice would be to get the 3D dimensions of the arena, including the scoreboard/Jumbotron in the middle. That will possibly be your best vantage point. You can possibly mount a ring of cameras from just below that area. The total number of cameras will be determined by many factors. AOV, PPF, light sensitivity, etc all come into play.

Sounds like a challenge. Just hope they have a decent budget to work with to go along with that high expectation.

(1)
UI
Undisclosed Integrator #1
Jul 13, 2016

It looks like the arena seating area is approx 200' x 300'.

The overhead structure that cameras would be mounted to is 90' above the floor. I really do not see any options for mounting as it is basically a large "bowl" with a structure over the top.

The VMS they are using is Milestone X Protect 2014. We did not install the original system, but they have been using us for some misc camera adds and repairs. We do not touch the VMS as that is completely controlled by their strict IT dept.

The variable lighting conditions are also a concern that I am not certain whether they fully embrace, yet.

Regarding budget, who knows? Like many large organizations, you never know what they might be willing to do...

I very much appreciate the input!

Avatar
Jon Dillabaugh
Jul 13, 2016
Pro Focus LLC

Not getting too exact with measurements and all, but even a 12MP camera with a 12mm lens is going to be insufficient for identification purposes at ~150'. You are likely to find that the low light abilities of the 4K cameras will also be insufficient. My best guess is you will need many (50+?) of 2MP super low light box cams with long lenses (~25mm). This setup will only cover about 35' horizontally and 20' vertically if left in wide format. I don't know how much vertically you have to cover. This would be a painstaking process to get the cameras all aimed and overlapping. Then keeping them organized in the VMS too.

Fun fun fun!

(1)
U
Undisclosed #2
Jul 13, 2016

If this were me, my next step (and the only step I would take) would be to spend time researching cameras compatible with Milestone and finding 3-4 reputable manufacturers that look like they could fit the bill, then call them in for a veritable shootout. These types of scenarios are far too specific to guess and cost far too much money to be wrong at.

Since Avigilon's Pro series is (not surprisingly) out, I'd probably call in Axis, Arecont, and maybe one other if you can find one that has some high-res or multi-sensor options to show their stuff and maybe provide an alternative input on design.

Good luck -- sounds like a fun project!

JH
John Honovich
Jul 13, 2016
IPVM

I disagree on the shootout until the math is worked out first.

Let's say you went with the Avigilon 30MP model (which is financially way too expensive even if it was supported), you still could not do it with one model. Here is the 30MP covering 200' wide:

You would need to use 2 of those to have a chance to ID. Now you are talking ~$15,000+ for just 2 cameras.

I recommend you first help your customer understand the relative tradeoffs of modern megapixel cameras with spaces that big. Walk them through it using our calculator, try different resolutions, FoV, widths, etc. and see what they then say.

(1)
(1)
U
Undisclosed #2
Jul 13, 2016
  1. I guess that is the value of using different approaches. I, personally, would not bother with the "math" until I talked with a couple of trusted manufacturers to get their perspective and see how they may have handled similar scenarios in the past. I would ask to see specific cameras in action, and, if at all possible, get recorded video clips using the cameras they'd recommend. From there, once we determined what is acceptable from a base level and what we deem the best solution, we back into the "math" required to get exactly what they want.

    TL;DR - Your recommendation is to show them your calculator. My approach would be to show them video of the scenario they're asking for.

  2. You mention $15,000+ for 2 cameras. Obviously, that's a lot.

    But, let's consider our options here, at a really high level --

    A. Use (2) 30MP cameras per "long" side and (2) 24MP cameras per "short" side. For the long sides, at 200 feet away and 7.5mm focal length, you get a HFOV of 160 feet at 46 PPF, which would give an approximate VFOV of roughly 90 feet if my super fast calculations are correct. Let's assume that is sufficient, just for the sake of discussion. So, with (4) 30MP cameras and (4) 24MP cameras, you're looking at roughly (camera+lens MSRP only) $84,000. Add in 8 camera licenses at $250 and you're getting a grand total of $86K.

    B. Use roughly (6) 3MP cameras per "long" side (21mm lens gives a 46' HFOV at 44.8PPF), plus (5) 3MP cameras per "short" side (same calcs as above). So, that's (22) total 3MP cameras. But, the problem here is that your VFOV in that case is only 26 feet. That means to capture the same ~90' that the 30MP captures, you'd actually need at least (18) 3MP cameras per "long" side, plus (15) 3MP per "short" side -- or 66 cameras total. Using an Axis P1357 camera at $900 MSRP plus a Computar lens that can give us 21mm at $250 MSRP....that's $75,900. Add 66 camera licenses at $250 each or $16,500, and we're at $92,400. And, don't throw the "Hikvision has 3MP cameras for $27" at me either....

    Now consider the cable drops, mounts, setup time (enjoy the aiming and focusing of all of those bad-boys to give us adjoined or overlapping views....), and annual PM for 66 cameras....not to mention the operator nightmare that would be....and add that to the cost as well.

I did that extremely fast so I'm sure someone can poke a few holes in those numbers. I didn't do that to say "YOUR ONLY OPTION IS TO USE A 30MP CAMERA". I just did that to show that there is occasionally more than one way to skin a cat, and when you consider every possibility AND the user experience at the same time, some options become more viable than others.

(3)
JH
John Honovich
Jul 13, 2016
IPVM

Well, first of all, you are approaching $100K for each of your options, so you probably want to check if the customer is open to something at that price. They might but they also might pass out.

As for your options consider C, which is to use 12MP cameras. There are a number on the market, e.g., the Panasonic 4K / 12MP (see test results) would be one to carefully consider. High image quality, high resolution, long telephoto varifocal autofocus lens built in. Most importantly, in terms of cost, much better overall pricing as that Panasonic is sub $2,000 for 12MP, which will slash total cost dramatically compared to your numbers.

(1)
Avatar
Jon Dillabaugh
Jul 13, 2016
Pro Focus LLC

John, I'm worried that the low light level in a concert scene would be an issue, based on your testing. Being that the center of the arena is probably close to 150' away from the furthest point in the scene, you won't be able to count on the IR. Also, the dark parts of the WDR tests give me pause as well. Do you see these as reasons to consider more quantity of lower res (1080p) super low light cameras? The overall cost of cameras should be lower this way, I think.

(1)
JH
John Honovich
Jul 13, 2016
IPVM

Jon, it depends on how dark the basketball arena / concerts are. That's something that would likely need to be tested but I agree it's a concern.

Even if you use lower resolution cameras, you still might need to add in external IR illuminators if it is quite dark and that far away.

(1)
U
Undisclosed #2
Jul 13, 2016

I completely agree. Panasonic would be a great 3rd option that I hadn't noted above.

My whole point with that rambling post (that I'm paranoid has mathematical errors in it that I can't go back and edit, so i'm not even reading it again...) is that there are a lot of different approaches, all of which can be "right" for a situation depending on various factors. And, I think everyone kind of has their own methods of handling things, many of which can be right as well.

Avatar
Robert Sprecher
Jul 18, 2016

Full Disclosure: I am Program Director for MorphoTrak and manage our Face and Video Analytics program. As others have said, use of cameras for a panoramic view of the arena and real time face recognition don't mix well. However, we have been using Axis Q-3505 PTZ and Panasonic WV cameras for post event face recognition and have identified recorded faces with very low resolution rates. We also only stream the face templates back to the matcher, so bandwidth is considerably lower than if we were streaming the entire acquisition. We have only integrated Genetec's Security System (5.2 or 5.3) VMS and Honeywell's Prowatch in this solution so far, but we are partner's with Milestone, so I don't see that as an impediment to integration.

I am not the technical resource that you would want to talk with, but I can put you in touch with one of my video solution engineers. I don't know how to arrange a conversation with one of them, without violating the terms of service for IPVM, so I am going to let John send me a message with instructions if you want to discuss face recognition for this use case.

New discussion

Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.

Newest discussions