Shootout: 4K vs 1080p vs 720p Video

Published Jul 17, 2013 04:00 AM

4K is the next big thing in video. While 1080p has become the 'standard' for HD, the next step is 4K, quadrupling resolution / pixel count from ~2MP to ~8MP. It is already showing up in TV sets as well as consumer cameras. Indeed, the first 4K surveillance camera was announced this spring, with many more announcements certain to follow this fall.

Can 4K really make a difference for surveillance? Or is Axis right that the 'megapixel race is over'?

We bought a GoPro Hero3 Black camera [link no longer available] that delivers 4k resolution (indeed up to 12MP). What is especially interesting is that they use the same chip manufacturer (Ambarella) used by cameras from many well known manufacturers (like Avigilon, Bosch, Pelco, Hikvision, Honeywell, etc.).

We took the Hero 3 Black and put it to the test at FoV widths from 20' to 160' feet to understand 4K performance.

4K clearly made a material difference in image quality at each FoV width tested, though with different practical impacts:

  • At 160' wide, faces were blurry with 4K but delivered an outline of features, materially better than the blocky featureless ones from 1080p.
  • At 80' wide, facial features started to show with 4K whereas the clear blockiness of the 1080p video obscured some
  • At 40' wide, facial features were quite sharp with 4K while they were modestly blurrier with 1080p video. Moreover, smaller characters were clearly distinguishable with 4K though not with 1080p.
  • At 20' wide, no material difference in facial features existed but the tiniest lines on the eye chart were clearly sharper and easier to read in 4K than 1080p.

Key Potential of 4K

In wider areas (50 to 100 feet) we expect to see the most practical impact. This includes lobbies, parking lots, long hallways, etc. Bigger, wider, longer areas are likely to get more facial and license plate details. However, at shorter ranges (50 and under), unless super small details are of interest (like a tattoo on someone's hand or the serial number of a bill), 4K is likely to be overkill.

Low Light Note

Though the increase in resolution is great, similar to the jump from standard definition to megapixel, users should be aware that low light performance may suffer in 4K cameras, especially considering HD cameras have only truly begun to perform well in low light in the past 1-2 years. Low light performance of 4K cameras remains to be seen when production models become available.

The GoPro HERO3 Black has an f/2.8 which is quite high for surveillance camera standards, making low light relatively terrible. Certainly, surveillance cameras will support lower f stops but we do not know how low.

12MP vs. 4K Performance

In all of these comparisons, 12MP and 4K are fairly close in pixels per foot (4000 horizontal pixels vs. 3840). However, 4K images tend to look moderately worse for two reasons:

  • 4K, 1080p, and 720p images were captured from videos, so slight motion and encoding artifacts may occur. 12MP images were captured as such from the camera.
  • The GoPro changes field of view via differing digital crops from the sensor. Though all of these fields of view were set to "wide", there are slight differences, a matter of a few pixels to either the left or right edge of the image.

160' Field of View

First, we began with an extremely wide 160' horizontal field of view. 

At this width PPF ranges from 8 to 25 as resolution increases. 12MP and 4K resolutions are able to capture some details, such as clothing and hair style, as well as lines 1-2 of the chart. Line 1 is potentially readable at 1080p, though not guaranteed, and details of the subject are reduced drastically at this resolution and lower.

80' Field of View

Next, we  reduced the width of the field of view by half, to 80'.

At this level, the performance gap between 12MP/4K and 1080p is more noticeable. Details of the subject are easier to see, and the chart may be read to at least line 4 at 4K resolution, while only lines 1-2 may be made out at 1080p.

40' Field of View

Next, we further reduced the FOV to 40'.


 At this level, the benefits of 12MP/4K are lessened compared to 1080p. Letters on the chart may be read down to line 9/10 in the 4K image, while only lines 4/5 may be read at 1080p. However, details of the subject are easily visible even at 720p.

20' Field of View 

Finally, using only a 20' FOV:

In a field of view this narrow, even our 720p camera captures fine detail of the subject, with letters on the test chart visible down to rows 7/8, making 12MP/4K likely overkill for FOVs this narrow. 

Low Light Performance

We tested the GoPro in a ~2 lux scene to gauge low light performance, as seen below. However, readers should not consider this a true test of 4K surveillance cameras, as the Hero 3 uses a consumer sensor and F2.8 lens, both of which negatively impact low light performance.

Test Equipment 

For this test, we used a GoPro Hero 3 Black Edition [link no longer available]. Though a consumer camera intended for "adventure sports" applications, it reportedly shares some chips with upcoming surveillance cameras. Specs on the camera are as follows:

  • F2.8 wide angle lens, ~170° FOV
  • Ambarella [link no longer available] A7 [link no longer available] system-on-chip
  • Sony IMX1117CQT 12.4 MP image sensor

 

Comments are shown for subscribers only. Login or Join