When Would You Use 60fps?

We are in the middle of testing 60fps vs 30fps to determine what practical benefits it has. Obvious testing scenarios include money counting and highways/traffic - applications where things move fast.

Any specific use case that you are interested in? Any aspect you would like to see tested in this frame rate comparison?

You really need 60fps (or better, 120fps) for 3D. Have at it...

Come on Carl, would you be interested in 60fps for your application?

Not really. Our regulations, which are based on the NIGC MICS (Minimum Internal Control Standards), call for a minimum of 20fps. That's derived from "real-time VCRs", that typically delivered 20-24 fields per second with an 8-hour tape and special heads running at 1/3 of normal ep (slp) speed. In 2003, we chose to record all channels at 30fps to be safe and we're sticking with that frame rate.

30fps doesn't apply to all casinos and many have chosen 15fps, at least for back-of-house and less critical cameras, but we are sticking to 30fps for all.

We don't really need higher frame rates. 30fps, especially sans interlacing, is more than adequate for our purposes.

This would be great for detail of the action or event on playback or review. Eliminate any chance of missed detail.

How many times have your reviewed information or clips and really need about 2 more clips between the clips your reviewing? Detail would be convienent. Or the cash detail where you see the bill, the drawer, but what happend in the second between. You have to make assumptions.

I just reviewed a series of events the other night when looking at a police chase thru a hospital where the 8fps cameras cought the images, the 30 fps cameras were not where you needed them, and no cameras in the areas you really wanted them. Luckily there were cameras in areas where the chase had to slow down. When moving thru a confined space with limited access at a high rate of speed, it gets really hard to get the details and definition.

I we would just go to a 1" format in the survaillance field this would help solve this. Expecially in casinos where you want every second of time to see all parts of the scene. I know carl you cant talk openly about some cases, but wouldnt you rather have 60, not 30 fps?

Here is where *I* would promote 60 fps if I were an integrator or manufacturer dealing with such products:

1) In highly competitive bid scenarios. I would convince the customer that 60fps is the new 30fps and it's really the minimum they should accept. That would make it a good differentiator for me.

2) In any meetings with A&E's. Show some sample clips showing that you just can't know FOR SURE who pulled a knife, or pocketed a Rolex without a 60fps recording. Get A&E's to write "system must be capable of 60fps" into all specs.

3) To any customers who seemed to be not aware of new technologies, IP video, etc. 30fps is an "analog standard", 60fps is what many HDTV feeds operate from. Anyone specing in 30fps products is simply selling you old standards and is obviously not up to date, and therefore not qualified to install a "modern" system.

4) For any scenario where the possibility of the video being frequently used as evidence occurs. As we all know, in current legal cases anything less than 30fps is just absolutely inadmissable and will not stand up in court...with trends towards 60fps you would want to make sure your evidence video is the best possible and not thrown out in court, so therefore shouldn't you go with 60fps??

John, what is the camera MP for testing 60fps?

I assume 60 fps might be good solution for schools (in wake of the shootings)...not sure how you are going to test that :)

Actually It would not matter if the cameras are not in the right place, position, coverage, and the right resolution. Detail is most important. Problem is you will need to upgrade all the storage requirements ( Double ) creating other problems - Bandwidth , Network, Storage, Cpu processing

Brian K, that's a pretty thorough analysis. Unfortunately, it's a recipe to trick not users, not to help them :(

Pushkar, we will be testing a 720p/60 camera. What scenario in a school do you think can benefit from that? Is a gunmen running or?

Christopher, we will be testing bandwidth. I suspect, on average, going from 30fps to 60fps will be significantly less than doubling. Even from 15 to 30fps in simple scenes, the increase is more like 50% (see bandwidth vs frame rate test). However, in more complex scenes, like dark / high noise and intersections, it doubles (or close to it). That said, my hypothesis is that there are further compression efficiencies when going from 30 to 60fps and using interframe codecs like H.264 (simply because not much changes in 1/60s). In any event, it will be important to see what the bandwidth increase is.

John, as we have seen 30fps is good enough for most of the day-to-day situations. However, if someone wants to use a 60fps I think a school would be a good location, especially during a crisis when innocent lives are at stake and the situation is chaotic (kids and may be gunmen running and shot being fired).

If I am an authority at the school board and have enough $$ at my disposal I would rather go for a 60fps camera system. (Strictly talking about fps here and not the limitation of a lower MP camera, assuming only low MP cameras offer much higher fps today)

John - Basically I think 60fps is mostly a gimmick. Sure, there may be some cases where it's really useful, though I bet anything that truly needs more than say 15fps, probably needs something like 240fps or more. It seems in most cases, especially where security would be involved, you're either talking about a general overview, or a super-precise requirement. I don't think that 60fps offers all that much over 30fps for the general use case.

You know I was being facetious with my comment... The biggest benefit of 60fps would be to those that try to use it "against" customers in the ways I outlined. With that in mind, I think your test is a good one, as my guess it will show that the practical benefits are minimal, and the costs of storage and bandwidth could be substantial.

Very interesting (and entertaining). Brian shared two very important things (imo):

1. 60fps vs 30fps will be of insignificant value (compared to other, related cost increases) in almost all cases. I particularly liked how he used '15fps' (instead of the 'accepted' 30fps) when noting that anyone with an actual need for higher fps would need 240fps+...

2. His 60fps parody (of the currently popular belief that 30fps is needed for everything, for the particularly obtuse) is spot on - and yet the other commenters clearly are under the same misconception that he is parodying. :)

30fps has been packaged and sold to untold numbers of integrators and end users (in exactly the manner Brian used).

Except for specific niche applications, nobody needs anything above 15fps. It adds no value.

I'm tipping my hat to BK on this thread. Very well done! :)

BK's 'facetious post' is likely being read as a handbook by unscrupulous marketers!

Pushkar, I am not sure how 60fps is going to help people running in a shooting. How fast are they running? Is the Blade Runner guy involved? In all seriousness, what specific difference in a school shooting would 60fps rather than 30fps make?

John, I think it would help during the investigation when the authorities want to review the footage in slow(er) motion (may be to figure out where the bullet was fired from). I understand that higher fps are required when the camera moves relatively fast (example, when installed on a positioner and using long lenses) to avoid motion blurr. But, if the technology exists at a relatively low cost, why not use a higer frame rate camera in schools... don't you agree?

Pushkar -

I actually think schools are one of the worst examples where 60fps would have any utility. School incidents are statistically speaking very very rare, but of course the ones that DO happen get extreme publicity. It's not like there is any data that indicates a growing epidemic of school violence in regards to shooters (or similar attack scenarios) in a school.

We've also seen that many schools have very little security overall. The money would be better spent, IMO, on the basics first. Visitor access control, a lockdown system, improved door hardware, intercom systems between classrooms and the main office.

If you've ever seen a video of a gunfire, it's pretty obvious WHERE a bullet came from (and in the school scenario it appears to be typically a single-shooter scneario, or maybe 2 shooters though they might not be operating side-by-side). You'll see a bright muzzle flash, and then the initial speed of the bullet leaving the gun will be around 1500-5000 feet per second. You're not going to see a bullet at 30 or 60 (or 100) fps, especially not the very small size it would be in the typical CCTV FOV.

Places with high foot traffic, open areas where you have massive amount of people and want to use mp cameras .

Places where you would need high speed detail, fast transaction detail, or details of extremes: Traffic control pts, intersections, airports, transportation bottlenecks, monetary exchange pts where you need to watch the detail. Also, College, University, School Parking Lots with mp cameras

These are just some of the areas you would benefit with 60fps cameras and it would allow the average in the industry to afford to install.

Marty, we use the 8, 15, 30 rule now and we still have times when things are not quite right, we still have to retweak the cameras and recording solution after an inccident for the next time.

Why not give us the option upfront, so we can design it in at the start?

I disagree, the option is better than after the fact. 8 fps does not do the job, 15 does not always do the job, 30 sometimes is not enough.

"These are just some of the areas you would benefit with 60fps cameras" - Listing various categories does not prove your thesis point (nor even support it). Tell us what the benefit IS vs lower fps?

"8 fps does not do the job , 15 does not always do the job , 30 sometime s is not enough" - Generalizations are fine if you are in sales; however, technical folk generally require proof of concept. When is '30fps sometimes not enough'?

Camera looking at parking lot 8fps distance of 300',1.3 mp, man opens door, jumps out, start running ,has a gun, drops gun, other person inicent bystander picks up gun not thinking, you cannot see enough detail in 8 fps to make an accurate call as to who is what here.

If you are walking a normal pace or speed through a parking lot you can see great, if you are running and something small happens, you cant see detail. You see overal picture, visable facts, but you cannot go into the details at only 8fps

Money changing, some times to see slight of hand movements you need as much detail as posible to see the fast movements in the transactions.

Anything running thru a confined corridor or past a open space on the edge of the picture, you will be lucky if you get a little bit of the action, but not enough detail.

Quick Action events are hard to see the detail if you only have the industry standard. Due to this problem we use multiple cameras looking at the same thing at different angles.

"Camera looking at parking lot 8fps distance of 300', 1.3 mp, man opens door, jumps out, start running, has a gun, drops gun , other person inicent bystander picks up gun not thinking, you cannot see enough detail in 8 fps to make an accurate call as to who is what here."

Unless we're getting off topic, it seems like you're comparing 8fps to 60?

What is the benefit of 60fps over the current standard of 30fps, and is 60 actually *enough*?

BTW, I usually find your comments to be very high quality, but this first example seems to be contrived beyond general possibility. I actually think 8 fps would be more than sufficient there. You're talking about at least a 4 second time span for that scenario, that's 64 images, pretty sure you'd be able to make out the chain of events there. Also, this scenario is not generally something that the people buying/deploying CCTV systems are trying to solve for. A builder owner/CCTV system owner is going to pay increased cost to try and protect some innocent bystander that picks up a gun tossed by another guy running away from a scene?

Money handling scenarios *might* benefit from this, but in most of your examples I would not expect 60fps to be "twice" as good as 30fps...

30fps is more than sufficient for money handling. Unless you really need to see each bill freeze-framed when a stack is "fanned" (not a likely scenario), I can't imagine any use for 60fps. The only real applications would involve machine vision, where fast-moving parts need to be captured or maybe license plate capture on freeways - another limited-use application.

I'm sure I wouldn't have to try too hard to come up with an extreme hypothetical situation where even 60fps would be "not enough". The examples given thus for are, frankly, ridiculous.

Tbh. I see 60fps having some usage in the industry sector. It's becoming more and more common that people would like to have an extra camera which is faster then your regular one cause they can see better what their installation is doing. We actually got a high-speed camera here for those specific cases.

Rogier, can you elaborate on those specific cases? What is the scene? What objects are there, how are they moving? We are trying to figure out potential scenarios to test / measure performance difference.

I've been trying to see if I got any photo's but sadly I have not.

We have several rolling mills here where we roll the metal to thinner sizes. I'm sure you've all seen those rolls of metal coils which is the end result of that process.
In order to keep a constant process we weld a new coil to the excisting one which is already inside the installation. This way we can just pull the strip trough the installation. At the end of the installation we have a giant cutter. This one seperates the 2 strips from eachtother so you got your individual coils again.

Now sometimes the cutter doesn't work properly. The cutter however moves really fast and we noticed that with a standard 25fps camera, we miss it occasionally. Hence we got a highspeed camera for that.

Another case is our hot rolling mills. Here we have 500-700 °C steel flying in which just got rolled from a large steel slab into a strip. So here we don't pull it trough the installation, but it's being pushed forward with the mills. This can go up to 80 km/h at the end.
At the end we spool the strip up. But imagine a hot fast moving sheet of metal flying in which you need to pick up correctly in order to spool it. This occasionally goes wrong, and thus, also a situation where we use high speed camera's.

As you can see they're quite unique cases. One being a fast moving stationairy installation, one being a fast moving object.
I hope my english was decent enough to paint a decent picture.

PS: I once did a test with the highspeed camera to see how particles moved trough a conduit. It was a new system of injecting cokes into our ovens and they wanted to see how it moved, and how fast. So we had one of the conduits out in the open, and instead of using cokes (which is high flammeble) we used calcium oxide. You could see all the particles flying in slow-mo. Fun thing to do.

Rogier, that is a very fascinating and worthwhile example of super high speed frame rate. Not a common one, of course, but definitely makes sense. Thanks!

I would question the need for 60fps even in the case of license plate capture at a highway toll gantry.

With the typical camera mounting height and viewing angle for this type of application, and the camera focused on a single 12' wide lane, I would expect the vertical FoV to be about 25 - 30 feet. If we assume that only the closest 12 - 15ft will provide sufficient PPF for getting a definite read on the license plate, we're looking at a car, travelling at approximately 70mph (103 ft/s), passing through an area of about 12ft. That means the car moves through the useful FoV in about 0.1165 sec. If the camera is capturing 30fps, we should be assured of capturing that car in 3 frames. Is that not sufficient?

What is really needed for highway license plate capture is not particularly high frame rate, but rather very high shutter speed.

At 30fps, you're getting 1 frame every 0.0333 second. Other than Carl's example of industrial machine vision, when would that not be enough? Is there any "sleight-of-hand" thief that can move that fast? Doesn't the typical sleight-of-hand trick also involve distraction and/or misdirection? You can't really distract a fixed camera.

The math doesn't lie. In fact, from what I've seen of video toll plazas here, the cameras are even wider than Richard suggests, typically covering their "primary" lane as well as half a lane on either side at the mid-frame point (thus negating the urban myth that you can avoid the camera by straddling lanes). Wider shot = even more time that the car is in frame = even more images captured = even less need for extreme framerates.

Rogier application is more of a machine vision application and in manufacuturing sometimen machine vision cameras in manufacturing produce more than 200 fps. But surveillance appliations almost alway involve human movement and humans just can't move that far or fast in a second that capturing 30 frames in that second won't capture everything. Also, just because you have a camera set at 30 fps doesn't mean you are actually achieving 30 fps. Most camera spec sheets say "up to 30 fps". So any testing should be done in a way where actualy frame rate can be verified. For applications like schools where you typically have 50-150 cameras, go ahead and do the math on the storage requirement for multi-megapixel cameras recording at 60 fps. Just not practical.

Exactly... It's not that the high framte rate cameras have no value in general, it's just that their value in what would typically be considered a security/CCTV application is mostly impractical or pointless.

Brian: in practically all of my installs, even 30fps is impractical and/or pointless. Heck, we do systems for two major oil companies, and their *official spec* only calls for 1fps (with analog, it also doesn't require over CIF resolution, although we'll usually set 4CIF for things like ID shots, paypoint, etc., and bump the framerate up to 4-6fps on those cameras, especially with the storage available in newer systems).

I can see two reasons for using a camera that’s able to produce 60 fps:

Running two streams from the camera, one for local server based analytics and one for remote live view.

Previous cameras that we tested simply didn’t have the power to produce for example one stream of 15 fps mjpg , low compression at 1280*720 and one stream of 15 fps h.264 1280*720 medium compression. So having a camera with more processing power is welcomed.

The other reason has to do with picture quality . I might not “need” the 60fps that the camera can produce, but the processing power capability of a camera that can produce 60fps should be able to do a lot more in areas/scenarios where you might need a WDR camera, but then only set the camera to deliver 20 or 30 fps depending on available bandwidth and scenario.

// Snowman