Testing Bandwidth vs Frame Rate

Author: Benros Emata, Published on Aug 30, 2011

*****PLEASE NOTE: DUE TO THE AGE OF THIS ARTICLE, WE WILL RE-RUNNING THIS TEST AND UPDATING IT IN APRIL 2018.*****

Frame rate selection has a major impact on bandwidth consumption for surveillance cameras. While most cameras have a maximum frame rate of 25 to 30 frames per second, users typically choose significantly lower frame rates, driven by a desire to minimize storage consumption. The most direct way to reduce storage is to use a lower bit rate stream for each camera with lower frame rates.

A key practical question is how much bandwidth is saved as the frame rate is dropped? Lower frames rates should result in lower bit rates but how significant are those changes and does this change depending on the scene?

Our Test

In this report, we share our result of a series of experiments we did with 3 IP cameras (Avigilon, Axis and Sony) in 3 different scenes: (1) indoor daytime, (2) indoor dark (< 1 lux), and (3) daytime outdoor (intersection). We did tests at 1, 5, 10, 15 and 30 fps to see how bandwidth consumption varied across a variety of frame rates.

We then produced a series of charts comparing bitrate vs frame rate and average frame size vs frame rate to provide two views on the relationship between these parameters.

With these test results, we answer the following questions:

  • How does bandwidth efficiency vary with changes in the frame rate?
  • Are low frame rate streams inefficient?
  • How can you project bandwidth consumption for various frame rates?
  • Does different scene types impact bandwidth vs frame rate efficiency?
  • How does the choice of VBR vs CBR impact the relationship between bandwidth and frame rate?

For background on frame rates, Pro members should review our frame rate fundamentals training. Also, for those interested in understanding the impact of bandwidth, review our sister report that examines the relationship between bandwidth and image quality.

Key Findings and Recommendations

Key Findings:

Get Video Surveillance News In Your Inbox
Get Video Surveillance News In Your Inbox

  • Bandwidth efficiency of lower frame rates can be modestly less efficient than higher frame rates when the I frame interval or scene complexity is low
  • As the scene becomes more complex (e.g., nighttime, noisy scene), bandwidth increases rather linearly as frame rate increases
  • No 'massive' bandwidth penalties exist for low frame rate H.264 encoding
  • Figuring out the right bitrate for a CBR camera at different frame rates is hard

Recommendations:

  • Unless your scene is quite simple (e.g., indoor office), you can ignore the impact of short I frame intervals or low frame rates impacting bandwidth efficiency
  • Even in simple scenes with short I frame intervals, bandwidth consumption is not a real concern as the bitrate will be quite low regardless (either very low or very very low)
  • When estimating bandwidth, using a fixed image size is generally a fair approximation but be careful because results could be off by up to 50% of projection.
  • When using CBR, if possible, try running in VBR mode first to get a baseline bandwidth level and use that as your CBR setting. Alternatively, export a few sample clips at different CBR bit rates, run this in a stream analyzer and pick the bit rate that results in a DRF of 32 or less

VBR Indoor Daytime (Low-Complexity)

Scene Overview

In this indoor daytime scenario the scene is fairly non-complex. Both cameras operate at VBR, but differ in their I-frame intervals. The Avigilon is using a 1s I-frame interval, whereas the Axis generates an I-frame for every 32 frames.

Bitrate vs. Frame-Rate

The Axis camera is considerably more efficient at the lower frames than the Avigilon camera. This is largely because the I-frame interval is effectively higher in the Avigilon at lower frame rates (at 1fps the ratio of Avigilon to Axis I-frames is ~30:1). At approximately 10fps the bitrates of the two cameras converge despite a 3:1 I-frame ratio. Another key point that may not be so obvious from the shape of the curves, is that bitrates increase less than linearly. As such efficiency improves at higher frame-rates (especially the Avigilon, which uses a per second I-frame interval).

Average Frame Size

At the lower frame rates the inefficiency of the higher I-frame interval is quite apparent. The Avigilon curve (red) indicates a considerable cost (in bytes) for I-frames relative to P-frames. This discrepancy is characteristic in less complex scenes, where P-frames are relatively efficient to produce (<10KB). The two curves converge around 10fps at roughly 10KB. The curves both continue to descend to a minimum at 30fps to ~7KB. At 30fps P-frame sizes dominate the average frame size metric (~30:1). With a value of ~7KB it suggests a low complexity scene and high compression efficiencies.

VBR Indoor Nighttime (High-Complexity)

Scene Overview

In this indoor nighttime scenario the scene is highly complex due to the presence of noise/gain. Both cameras operate at VBR, but differ in their I-frame intervals. The Avigilon is using a 1s I-frame interval, whereas the Axis generates an I-frame for every 32 frames.

Bitrate vs. Frame-Rate

At the low frame-rates the I-frame intervals between the two cameras are most significant. For example, at 1fps the Avigilon produces one I-frame for every one P-frame (or 1 every second), and the Axis camera produces one I-frame for every 31 P-frames (or 1 every 31 seconds). As such we would expect to see divergence of bit-rates at these lower frame-rate levels, but this is not the case as the curves follow each other in near identical fashion. It is clear that compression efficiency is very low due to the complexity level of the scene and that byte-wise cost of P-frames is not much 'cheaper' than I-frames, as a result. A key feature of these curves is the linear relationship between bitrate and frame-rate, suggesting no clear frame-rate where efficiency is at its greatest.

Average Frame Size

The minimum frame sizes of the two cameras of ~25KB to ~30KB is considerably higher than the ~7KB minimum found in the non-complex indoor daytime scene. The most interesting feature of this graph is the relative flatness of the Avigilon (red) curve. Because the Avigilon average frame size is influenced strongly by I-frame size at lower frame-rates and by P-frame sizes at higher frame-rates, its curve indicates the relative sizes of I vs. P frames. Given that the curve is mostly flat suggests little material difference between I and P frame sizes. This is characteristic and somewhat expected in highly complex scenes. As a result there is no clear frame-rate that is most efficient and as the previous graph depicts bitrates increase rather linearly as frame-rates are increased.

VBR Outdoor Intersection (Moderately High Complexity)

Scene Overview

In this busy intersection scenario the scene is moderately high in complexity. Both cameras operate at VBR, but differ in their I-frame intervals. The Avigilon is using a 1s I-frame interval, whereas the Axis generates one I-frame in every 32 frames.

Bitrate vs. Frame-Rate

As in the highly complex nighttime scenario, the curves of the two cameras provide no clear area or point of divergence. Their bitrate values are actually quite similar at each frame-rate setting. The bitrates reach a high of ~4mbps at 30fps, which is considerably lower than ~6.5mbps high observed during the very complex nighttime scene. This tends to suggest a somewhat lower level of complexity than the night scene.

The shape of the graph tends to flatten out going from lower to higher frame-rates, suggesting some efficiency gains at the higher P-frame saturated frame-rates.

Average Frame Size

As expected the Avigilon camera, which uses a 1 per second I-frame interval, exhibits a sharper decline in average frame size as frame rates increase. The deeper average frame size range of the Avigilon (~30KB to ~15KB) suggests that compression efficiencies can still be obtained by reducing the ratio of I-frames to P-frames (i.e., increase fps), and that the scene is not as complex as the nighttime scene.

The Axis curve (blue) is somewhat anomalous as its I- to P-frame ratio is a constant, and as such we would not expect its range to be as deep as experienced here in the intersection scenario. Indeed the Axis' range was ~5KB in the daytime and nighttime scenes versus the ~10KB range seen here (25KB to 15KB).

CBR Strategies for Bitrate vs. Frame-Rate Optimizations

Setting the right bitrate for a particular frame rate is significantly harder in CBR than in VBR. With VBR, the camera automatically adjusts the appropriate bit rate when you choose a different frame rate. By contrast, with CBR, the admin has to manually adjust the bit rate. For example, with a camera such as Sony, if the default bit rate is 2Mb/s, if you change the frame rate from the default of 30 to 15, the bit rate stays the same. What happens is that the camera will decrease the compression (or Quantization level) to decrease the amount of loss in the image. The key potential downside is that this can waste bandwidth. At a certain point, increasing quantization levels have little to no practical impact. That is, the mathematical loss decreases but the gains are imperceptible to the human eye. For background on this, please review our Bandwidth vs Image Quality report that digs into the issue of quantization.

When using CBR, if possible, try running in VBR mode first to get a baseline bandwidth level and use that as your CBR setting. Alternatively, export a few sample clips at different CBR bit rates, run this in a stream analyzer and pick the bit rate that results in a DRF of 32 or less. The final easy, low tech, alternative is to simply set different CBR bit rates, view the sample images and choose the lowest bandwidth image where you cannot spot any material differences in quality.

Methodology

Here are the cameras used in 'Bandwidth vs. Frame-Rate' report:

VBR

CBR

  • Sony CH140 (online $800) - 720p D/N; 1/3" CMOS; F1.2 Fujinon; 0.1 Lux (BW)

Here are the firmware versions of each camera:

  • Avigilon 1.0MP-HD-H264-B1 - fw 1.0.4.6(3445)
  • Axis P1344 - fw 5.22.2
  • Sony CH140 - fw 1.26.01

For VBR tests the two (2) cameras were left at their default VBR configurations. For each of the three key scenes (daytime indoor, nighttime indoor, and busy intersection), the cameras were recorded at different frame-rates (1, 5, 10, 15, and 30 fps). Average bit-rates and average frame sizes were obtained for analysis during each scenario.

5 reports cite this report:

Calculating Video Surveillance Storage / Bandwidth on Dec 29, 2016
Calculating surveillance bandwidth is complex, and inexperienced users can easily underestimate bandwidth, leading to reduced storage durations...
Frame Rate Guide for Video Surveillance on Aug 07, 2014
This is the industry's most in depth guide to frame rates in video surveillance. As a precursor, you need to know the speed of objects, most...
IP Camera Bandwidth Training on May 27, 2012
Understanding bandwidth is critical to using IP cameras. In this new training, we show how to measure bandwidth and how significantly bandwidth can...
Directory of Camera Shootout Series on Dec 27, 2011
The following directory lists all of our camera shootouts. These shootouts pit 4 - 8 different surveillance cameras in simultaneous tests on real...
Training: Frame Rates for IP Cameras on Apr 18, 2010
This is the industry's most in depth guide to frame rates in video surveillance. As a precursor, you need to know the speed of objects, most...
Comments : PRO Members only. Login. or Join.

Related Reports on Bandwidth

Dahua Dual Imager Dome Camera Tested (HDBW4231FN-E2-M) on Nov 07, 2018
Dahua has introduced a dual-imager dome model, the HDBW4231FN-E2-M, with two independently positionable sensors including integrated IR, not found...
Winter 2019 IP Networking Course on Nov 05, 2018
This is the only networking course designed specifically for video surveillance professionals.  Lots of network training exists but none of it...
Cloud Video Storage Usage 2018 on Oct 31, 2018
Storing email and documents in the cloud have long been common, with on-site email or file servers increasingly eliminated. However, what about...
Video Quality / Compression Tutorial on Oct 17, 2018
While CODECs, like H.264, H.265, and MJPEG, get a lot of attention, a camera's 'quality' or compression setting has a big impact on overall...
Last Chance - October 2018 Camera Course on Oct 04, 2018
Today is the last day to register for the October 2018 Camera Course, register now. This is the only independent surveillance camera course,...
Axis 4K vs. 5MP Camera Shootout (Q3518-LVE + Q3517-LVE) on Oct 04, 2018
Axis has released the 4K counterpart to their 5MP Q3517-LVE (see our test). Axis claims of both these cameras: Simply put, AXIS Q3518(17)-LVE...
SNMP / Network Monitoring For Surveillance 2018 on Aug 21, 2018
Surveillance systems typically rely on the the VMS to report issues, but this most often just means knowing a camera is "down" with no warning or...
Eagle Eye Networks Cloud VMS Tested on Jul 26, 2018
Eagle Eye has become one of the most significant players in the industry in the past few years: Eagle Eye's Owner Acquired Brivo Eagle Eye...
Powerline Networking For Video Surveillance Advocated By Comtrend on Jun 08, 2018
Powerline networking, using existing electrical wiring, has been around for many years. Indeed, over the years, some video surveillance providers...
H.265 / HEVC Codec Tutorial on Jun 07, 2018
H.265 support has improved significantly in 2018, with H.265 camera/VMS compatibility increased compared to only a year ago, and more manufacturers...

Most Recent Industry Reports

'Sticker' Surveillance Camera Developed (CSEM Witness) on Nov 16, 2018
The Swiss Center for Electronics and Microtechnology (CSEM) has announced what it calls the: world’s first fully autonomous camera that can be...
ISC East 2018 Mini-Show Final Report on Nov 16, 2018
This is our second (updated) and final show report from ISC East. ISC East, by its own admission, is not a national or international show, billed...
Facial Detection Tested on Nov 16, 2018
Facial detection and recognition are increasingly offered by video surveillance manufacturers. Facial detection detects faces in an image/video...
Throughtek P2P/Cloud Solution Profile on Nov 15, 2018
Many IoT manufacturers either do not have the capabilities or the interest to develop their own cloud management software for their devices....
ASIS Offering Custom Research For Manufacturers on Nov 15, 2018
Manufacturers often want to know what industry people think about trends and, in particular, the segments and product they offer.  ASIS and its...
Hikvision Silent on "Bad Architectural Practices" Cybersecurity Report on Nov 14, 2018
A 'significant vulnerability was found in Hikvision cameras' by VDOO, a startup cybersecurity specialist. Hikvision has fixed the specific...
French Government Threatens School with $1.7M Fine For “Excessive Video Surveillance” on Nov 14, 2018
The French government has notified a high-profile Paris coding academy that it risks a fine of up to 1.5 million euros (about $1.7m) if it...
Integrator Credit Card Alternative Divvy on Nov 13, 2018
Most security integrators are small businesses but large enough that they have various employees that need to be able to expense various charges as...
Directory of Video Intercoms on Nov 13, 2018
Video Intercoms, also known as Video Door-Phones or Video Entry Systems, have been growing in the past decade as more and more IP camera...
Beware Amazon Go Store Hype (Tested) on Nov 13, 2018
IPVM's trip to and testing of Amazon Go's San Francisco store shows a number of significant operational and economic issues that undermine the...

The world's leading video surveillance information source, IPVM provides the best reporting, testing and training for 10,000+ members globally. Dedicated to independent and objective information, we uniquely refuse any and all advertisements, sponsorship and consulting from manufacturers.

About | FAQ | Contact