Testing Vivotek Smart StreamBy: Derek Ward, Published on Apr 18, 2014
Reduce storage costs by 30%, claims Vivotek.
Their new "Smart Stream" feature intelligently compresses video streams such that moving objects get higher quality than background, stationary parts of the scene. Here is their promo video:
How Is This Done?
Recall all surveillance video is compressed. In H.264, one can choose the level of compression from 0 to 51 (least to highest amount of compression) and each manufacturer can choose how much compression (see our IP Camera Manufacturer Compression Comparison).
Camera manufacturers typically use a single compression level for the entire scene (for example, 28). What Vivotek is doing with Smart Stream is dynamically varying the compression level (for example, a 'low' 24 for moving objects and a 'high' 36 for stationary / background ones.
This absolutely saves bandwidth compared to using a single compression level. However...
The IPVM Test of Smart Stream
Just because one manufacturer is dynamically varying compression level does not ensure that it is lower than the compression approaches of rival manufacturers.
Here are our key findings from the test:
- Smart Stream was most effective in dark scenes, reducing bandwidth by an average of ~55% in our tests.
- In full light, savings varied, reducing bandwidth by ~15% to ~30% depending on the scene.
- Quantization reductions on moving objects occurred quickly, with no issues, even with vehicles moving ~40mph.
- Compared to other bullet cameras tested (Avigilon and Hikvision), bandwidth consumption with Smart Stream enabled was average competitively.
- Bandwidth consumption with Smart Stream disabled was far higher than average.
- IR detection range on the IP8371E was ~25 meters (~80'), ~5 meters short of their claimed IR range of 30 meters.
- Full light performance was similar to competitive models, with slight desaturation, though not problematic.
- Image quality of the Vivotek IP8371E was roughly the same as the previous generation Vivotek IP7361 in full light. In low light, the IP8371E's field of view had more IR illumination evenly covering the scene than the IP7361, and did not appear as washed out as the IP7361.
Note: readers highly interested in bandwidth savings techniques, should compare to our test of Arecont's Bandwidth Saving Mode
The Vivotek IP8371E is more expensive than most competitive zoom lens 1080p bullet models, such as the ACTi B45, Avigilon 2.0W-H3-BO1, and Samsung SNO-6084R, which cost costing ~$100-$200 less.
- Vivotek IP8371E: ~$780 USD Online
- Avigilon 2.0W-H3-B01: ~$650 USD estimated street price
- Axis Q1765-LE: ~$1,400 Online
- ACTi B45: ~$670 Online
- Samsung SNO-6084R [link no longer available]: ~$600 Online
Smart Stream is an interesting addition to IP cameras, with the potential to reduce bitrates, especially in low motion scenes. We recommend it be used in Vivotek cameras whenever it is available. However, even with it turned on, the IP8371 did not deliver bandwidth / storage savings compared to competitors.
We found the IP8371E's performance average, especially considering its price, higher than many competitors. Other options may deliver better IR performance with lower bitrates at similar or lower prices.
In this video, we review the physical features of the IP8371E IR bullet camera. The housing remains much the same as the previous generation IP7361, though the IR LEDs have been replaced with newer, larger, more powerful components.
Smart Stream works by increasing quantization on "background" areas while reducing it on "foreground" objects. By default, this is automatic, with moving objects considered foreground, but it may be applied to fixed regions of interest as well.
We review these options in this screencast:
This example image demonstrates "foreground" and "background" in an indoor field of view with a person walking through the scene:
Below, we have compiled a few example comparisons demonstrating the impact Smart Stream has on image quality. Users can click on an image to download a full-size comparison, to better distinguish the differences with smart streaming being off or on.
Indoors with walking subject
Indoors in full light (our conference room), Smart Stream "on" compressed the carpet and ceilings of the room, while our subject appears detailed and distinguishable. We have highlighted areas of noticeable compression in red.
Readers can see this in motion in this clip:
Outdoors with moving traffic
Moving the cameras outside, we shot the Vivotek IP8371E against a busy road to test Smart Stream in a real world application. Trees and fence are compressed in the background of this scene, as is the pavement in the foreground is also compressed. On average, bandwidth savings were ~0.5 - 1.0 Mb/s with Smart Stream turned on.
We gathered bandwidth measurements with the cameras normalized to 10 FPS, 1080p resolution with Q = ~28-30 for all cameras. Measurements are presented in Mb/s. Note the Vivotek IP8371E default and Smart Stream measurements, which provide various savings in different scenes between being on or off.
At night, we set up the cameras outdoors to accommodate a deeper FOV to test IR range. Below is an image of the FOV used.
Effective IR range of the IP8371E was ~80', well short of the ~100' (30m) specification. This outperforms the previous generation IP7361, effective to only about 65'. The Avigilon 2.0W-H3-BO1 beat both with ~100'+ range.
Test Chart -- IR and Low Light Cameras
Our subject then held the test chart in front of the cameras at a ~30' HFOV. From here, the Avigilon IR camera is able to produce the most details from the test chart, as well as out subjects eyes and head, while the Vivotek IP8371E was afflicted with large amounts of gain, making it difficult to produce details.
Finally, we tested the cameras indoors to compare image quality in a controlled environment. Below is the FOV used for testing.
In full light, the Vivotek IP8371E is comparable to other manufacturers tested, though moderately desaturated. It also appears oversharp compared to other cameras, with artifacts visible around the edges of objects.
Indoors, gain has less impact on the IP8371E's image (since IR illumination is reflecting off the white walls of the conference room). Also note the less concentrated illumination pattern of the Vivotek IP8371E compared to the IP7361's central hotspot.
We tested cameras using these firmware versions:
- Avigilon 2.0W-H3-BO1: 22.214.171.124
- Bosch NBN-932V: 12510590
- Hikvision DS-2CD2232-I5: V5.0.2
- Samsung SNB-6004: 2.22_131218
- Vivotek IP8371E: IP8371E-VVTK-0100j
- Vivotek IP7361: IP7361-VVTK-0201f
The following version of Exacqvision was used for this test:
- Version 126.96.36.199249