What is the value of "full-frame-rate" video?
The evolution of IP video has placed a lot of attention on the resolution of video, as measured in the growing number of pixels in a frame. But another variable, receiving less attention, is the number of frames captured per minute (fps). We inherited the idea of “full-frame-rate” video from the analogue world, but increasing numbers of pixels (and more data!) have sometimes led to use of slower frame rates. We asked our Expert Panel: What is the value of “full-frame-rate” video? Absent specific compliance requirements, what might suffice as an acceptable frame rate (i.e., less than 30 frames per second [NTSC] or 25 frames per second [PAL])?
The more information you have in a video, the better is it for forensics purposes to figure out what happened after the fact. Following that theory, high frame rate should always be better than lower, right? Well, the surveillance industry has long accepted and talked in terms of 4 fps, 7.5 fps and 15 fps, which has stemmed from technical limitations and high costs, creating a real desire to have a lower frame rate. When looking at a certain storage capacity, of course, the user has to choose between retention time, frame rate, but also resolution and image quality. That being said, there will always be different scenarios and needs in different market segments. For instance, a casino has long required a full 30 fps, and looks at 60 fps as beneficial. In the future, as technology improves, it is possible that we could see 30 fps become the norm.
Higher frame rates are better suited in providing fluent video playback when monitoring fast-paced objects (e.g. moving cars, swift moving people, etc.). Attempting to monitor a fast-paced object at a slower frame rate will produce choppy video. Also, higher-frame-rate video increases the probability of capturing a usable frame (e.g. clear image of individual’s face, denomination of currency pulled from cash register, etc.). The average frame rate used in the field is 8 fps as it provides smooth video to the human eye while minimising storage costs. Six to 10 fps meets most requirements. However, the camera location many times will help determine the best frame rate for an application. If a camera has a tight or limited field-of-view, integrators tend to use higher frames rates 10 to 15 fps. For wide view or a large area that is less important, a lower frame rate is often used.
High-frame-rate video is valuable in applications that require the observation of fast-moving objects and people, such as license plate recognition on moving cars or gaming tables in casinos, where hands and cards move quickly. In security applications, full-frame-rate video is generally needed for applications in which the movement of the object and people needs to be very carefully observed. Two other factors must be considered: bandwidth/storage and shutter speed. The greater the frame rates, the higher the bandwidth and storage requirements. Shutter speed and its impact on video quality are often overlooked. Some believe that blurriness in the video is most commonly caused by slow frame rates, but most likely is caused by shutter speed. In my experience, for most common video surveillance applications, frame rates around 10 fps offer the optimal balance of video quality and bandwidth/storage.
As ever with CCTV, the answer depends on your footage’s purpose. “Spatial” resolution drives arguments about PAL/NTSC vs. HD vs. megapixel. How close together in space can two things be before your pixels are deemed too coarse (i.e., too few) to distinguish those details? Change space for time: “spatial” for “temporal.” How close together in time can two things be before your video fails to separate them? You must think how many images per second you need to keep the gaps short. I once took a client’s footage of a genuine street fight to check how many images per second were needed to prove who punched who – 25fps was fine but 12fps made the video evidence doubtful. Don’t forget, however many fps you choose, your shutter speed must be fast enough to prevent motion blur spoiling the details. All of this only has value when it meets your CCTV’s purpose.
The question refers to “full frame rate” and “acceptable frame rate;” however, the terms are very different. Full frame rate is a strict definition for the maximum frame rate supported for video surveillance applications, which is 30 frames per second for NTSC and 25 frames per second for PAL. Acceptable frame rate is application-dependent as well as subjective. For instance, in the gaming market, frame rate requirements are determined by gaming regulations. In many cases, full frame rate would be the strict requirement, and any lower rate will not be acceptable. This is the case in the state of Nevada. However, regulations in other states are sometimes more tolerant. Some casinos would indeed deploy lower frame rates to save on storage, although it is not common. With other applications – for instance, perimeter protection – some would accept much lower frame rates, for instance, 10 frames per second, depending on the case.
Full-frame-rate video is generally designated as 30 frames per second (fps). The value of full-frame-rate video is to deliver frames and images to meet the end user’s requirements. We see the highest frame rate requirements for gaming and some other specialty applications like airports. To insure that 30 fps are delivered, Arecont Vision offers “casino-mode,” which sets 30 fps as the minimum number of frames that will be delivered to the system. Many other “high frame rate” applications are in shopping malls, hotels, city surveillance, traffic monitoring and distribution centers. In all of these cases, the user wants to be able to go to the recorded video and forensically determine what occurred. Whether the end user is seeking 30 fps, 15 fps, or 7 fps, the key is whether the designated recorded fame rate will provide the data necessary for the investigation and actions that need to be taken.
As with any variable related to a video surveillance of security system, the question of frame rate comes down to determining the needs of the application. As higher pixel counts (and limited storage and bandwidth resources) put downward pressure on frame rates, it’s important that integrators and end users manage the compromises well. After all, what good is a really clear image if it misses action that happened between frames?