Dahua Starlight 1/2" Camera Tested

By Ethan Ace, Published on Oct 16, 2015

In super low light, an area that the West has long dominated, can the Chinese manufacturers beat their Western rivals?

Dahua has a new 'Starlight' offering, featuring a 1/2" imager, priced ~$300 online.

We bought one, the Dahua Starlight IPC-HF8281EN and put it to the test against Hikvision's 1/2" Darkfighter as well as top low light performers from Axis, Bosch and Samsung.

Here is a preview of the low light results:

*** ***** */*" ********* camera's *** ***** *********** was ******** ****** *********'* */*" DarkFighter **** **** ***** a ******** **.* ****, but ************ *** ***** 1080p *** ***** ******* tested, ** **** ** the ***** *** ********* model.

***** *** ********* ***** low ***** **.** ****, the ***** ********* ***** produced ******** ****** **** when ***** *** ******* F1.2 ****, ** **** as *** ********* ***********, with ******** ***** **********, though ******* ****** ******* remain ************* *** ** digital *****.

*** *********** *** ******** worse **** *********** ***** WDR ****** ******* ****** backlight **** ****** ************, but ******* ** ** better **** ****** ** **** ***** of *** *****.

******** ** *** ***** (~0.01 ***) **** ********** lower **** *********'* *********** model, **** **** *********'* H.264+ ***** ***** **.

*** ****** ********** **** Exacq, *******, ********* *** Avigilon *****.

*******/************

*** ***-******* ***** ***~$*** *** ******, *** ********* ****, an ********** ~$***. **** that ** ** *** recommend *****'* ********* ****** due ** ********** **** F ***** (*****).

** ***** $*** *** for ****** *** ****, Dahua's ********* */*" ***** is *** ** *** lowest **** ***** *** light ****** *********, ********** in ***** ** ***** resolution */*" ****** ****** like *** ******* ***-****.

******** ** ***** */*" models, ** ** ************* less *********, **** **** half *** ***** ** Axis *** ****'* ***** low ***** ****** (*/*" and */*", *** ***** ~$150 **** **** *** 1/3" ******* ***-****, *** next ****** **** ********** tested.

***** ***-*******: ~$*** *** ****** (~$285 ****** + ~$*** lens)
**** *****: ~$*,*** ******, ******** lens
**** *****: ~$*,*** ******, ******** lens
********* **-***********-*: ~$*** *** ****** (~$600 ****** + ~$*** optional ****)
******* ***-****: ~$*** ****** ($*** camera + $*** ****)
**** ***-*****: ~$*,*** ******, ******** lens
******* *******-**: ~$*** ********* ****** price ($*** ****** + $150 ****)

***************

*** ***** */*" ********* HF8281E ** * ******** choice *** *** ***** imaging ******* ********** **. At **** *** ***** of ***********' ***** *** light *******, *** *********** ******* to *** ********* *********** 6026 *** ****** **** other ****** ***********, ** is **** ***** *********** in *** ***** *********.

Physical ********

*** ***** ******* ** similar ** **** *** featureset ** **** *** cameras. ** *** ***** below ** ****** *** physical ************ ** *** camera *** ******* ** the ********* **** *********** model. *******, *** ******** Hikvision **** ** **** larger **** ****, ****** the ******* *** ****** the **** ****.

******

***** ********** *** ****** for *** *******:

***-***********-***** [**** ** ****** available]: *** */*.*" ******, 4-18mm, **.*
***-***********-***** [**** ** ****** available]: *** */*.*" ******, 12-50mm, **.*

*******, **** **** ********** high * *****, **.* and **.*, ******** ** F1.2 ** *.* ***** in **** *** ***** lenses. ******* ** ****, we **** * ******** Computar M13VG288IR [**** ** ****** available] 1/2.7" *** **.* *****, recommended ** **** *************. Beware **** ***** ******* format ****** ** */*" imager *******, *******, ** vignetting *** ***** ** short ***** *******.

Low ***** ***** *******

*** ***-******* ********* ********* slightly ***** ** *** light ******** ** *** ********* DarkFighter **** **** ***** the **.* ****, **** increased ******* ***** *** dimmer ******. *******, ** performed **** ****** **** the **** *****, ******* SNB-6004, *** ***** *** starlight, **** ** ***** provided ****** ****** ** our *******.

***** *** ********* **.** DarkEye ****, ****** *** brighter **** **** *** Hikvision *********** ** *** same ***** *****, **** 3-4 **** ******* ***** of *** **** *****. However, *** *******'* **** remains ******** ** *****.

***

** ****** *** ******* our ******* **** ******** door *****:

*** ******* ********* ***** of ******* ****** ******* the ****** ********* ** the **** ******** ****, with ****** ************ ** our *******, ****** *** details *** ***** *****. In *** **** **** beside *** ****, ** was *** ********* *********, with ***** ****** ** both ******* *** *****.

Full *****

*****'* */*" ********* ***** ********* similarly ** ***** ****** in **** *****, **** no ******* ******.

*******

*****'* */*" ********* *****'* bandwidth *********** *** ********** higher **** *********** ** full ***** (***** ****** off), ** ***** ~*.* Mb/s (******** ** ~*.*-*.* Mb/s ** ***********). ******* Zipstream ** ** *** Axis ***** *** *.***+ on ** *** ********* 6026, ***** ******** **** to **** * ******** of *****.

*******, ** *** *****, its ******* ** ***** than *******, ** **** as **** *** ********* with ***** ***** ***** settings *** (********* *** H.264+). ******* ***** **, the **** *****'* ******* drops ***** *** *******, but ********* *********** ******* higher.

***** **** *** ******* Smart ****** ** *** of ***** *******.

Test **********

******* **** ****** ***** default ********, **** *** following ******** ******* ** *** cameras:

******* ***** *** *** to * ******* ** 1/30s
*.*** **** *******, ** FPS, ~** ************ *** used

*** ********* ********/******** ******** were ****:

**** *****: *.**.*.*
***** ***-*******: *.***.******* **.*.*, build: ****-**-**
********* **-***********-*: **.*.*
********* **-**********-**: **.*.*
******* ***-****: *.*********
***********: *.*.*.*****

Comments (26)

Ethan Ace

IPVM | 10/16/15 01:56pm

Several Samsung employees have asked why we used the Samsung SNB-6004 (1/3") instead of the SNB-6005 (1/1.9") in this test and others.

We are aware of that model, but as of the time this test was performed, it was not available and we are still having difficulties procuring one. We'll order one soon and put it in queue.

Undisclosed #1

IPVMU Certified | 10/16/15 02:27pm

Using the Hikvision super low light F0.95 lens, the Dahua starlight model produced brighter images....than even the Hikvision DarkFighter at the same light level

I'm pretty sure somebody at Hikvision wishes they had soldered that 0.95 lens directly to the body of the Darkfighter.

Undisclosed #2

10/17/15 03:53am

Kind of conserning when you're telling customers Axis is the best, but the tests prove otherwise. Luckily Axis has great support, selection, bandwidth, ACAP, and build quality. However, this would have me hesitant to do a image quality shoot out with a competitor that is half the cost pitching Hikvision. I'm curious how the Axis Q1635 and Sony SNV-VB635 would have done? It's Kind of like buying/sellling a Mercedes and getting beat by a Hyundai. Come on Axis, step up your game. Ugh...

Sebastian Kratzke

10/17/15 01:16pm

Unfortunately this ins another test where you better should have tested against Axis Q1635 with 1/2" chip.

I know you wrote you don't want to buy an Axis Q1635 but in my opinion but performance test results would have been much more fair and interesting then.

John Honovich

IPVM | In reply to Sebastian Kratzke | 10/17/15 01:22pm

Unfortunately for Axis, they charge nearly 3x what Dahua is charging for the same 1/2" imager class of camera. Let's say theoretically the Axis Q1635 came close to matching Dahua and Hikvision (not guarantened considering how poorly their high end 1/3" Q1615 did), it would still be $500 - $700 more for the same image quality.

Alternatively put, maybe it would be more fair to only test cameras with similar prices, which would further underscore how far Dahua and Hikvision have unforuntately caught up and past Axis.

John Honovich

IPVM | In reply to John Honovich | 10/17/15 01:44pm

That said, it is worth a test since more and more manufacturers have 1/2" imagers now.

I just bought one.

However, unless the Axis Q1635 does dramatically better in low light than Dahua and Hikvision, it will not show well for Axis.

Sebastian Kratzke

In reply to John Honovich | 10/17/15 10:17pm

That's a great move, John!

I see that from an economic point of view it's hard to beat Hikvision or Dahua.

But from a technical point of view I think it's most interesting to see a comparison with their direct pendant (1/2" chip).

Undisclosed Manufacturer #3

10/17/15 02:37pm

[Poster is from Bosch]

Test with Bosch cameras are not comparable due to different resoluton. You should set the camera to work @1080p (not the stream resolution but the camera install menu). Then you will double sensitivity and increase WDR. Last but not least if you test Bosch camera in night mode you should set Night optimized profile" and if you work in WDR you should set "iAE" profile and you get other 16db.

Thx

Ethan Ace

IPVM | In reply to Undisclosed Manufacturer #3 | 10/18/15 12:25am

We tested the 5MP starlight in 1080p mode awhile back and did not see a meaningful difference.

Second, unless something has changed, the night optimized profile has multiple issues in its default settings. If I remember correctly, it sets shutter speed to 1/15th second max, and increased DNR to higher levels, which definitely reduces noise, but introduces blur.

We did use Intelligent AE when testing. We typically do unless we see negative effects.

Finally, regardless of the camera's capabilities at 1080p, it's marketed as a 5MP camera, and users specifically asked how it would compare to 1/2" 1080p models which is why I included it. Generally users are expecting performance at a camera's max resolution. Also, It's a $1000-1100 camera (online), which puts it in the same spot as Axis, discussed by John above. Even if it's nearer to the DarkFighter and Dahua models, it's still 2-3x the price.

Tim Russell

In reply to Ethan Ace | 10/20/15 08:12am

So you used a 1/2" or 1/2.7" lens on a 1/1.8" sensor and you wonder why there is less light on the sensor ? The lux levels are also lower and WDR higher in 1080p. You should probably redo the test if you're presenting this as accurate.

Ethan Ace

IPVM | In reply to Tim Russell | 10/20/15 03:47pm

The 1/3" F1.2 lens we used on the Dahua starlight in this test passes more light to the sensor than 1/2" F1.6 lenses, including the one that ships with the Bosch 5MP starlight. I know because I tested it.

Again, as I said above, we have checked the Bosch 5MP starlight in 1080p mode in the past and not found much of a difference in low light performance, though in full light you have the obvious advantage of 5MP resolution. I will check it again in our next low light test with up to date firmware, as well.

Tim Russell

In reply to Ethan Ace | 10/22/15 06:55am

Cool, thanks. Surely by lowering the resolution, the larger pixels would gain more light ?

Undisclosed #1

IPVMU Certified | In reply to Tim Russell | 10/22/15 08:00am

Surely by lowering the resolution, the larger pixels would gain more light ?

Pixels don't change size.

Undisclosed #1

IPVMU Certified | In reply to Undisclosed Manufacturer #3 | 10/22/15 08:13am

You should set the camera to work @1080p (not the stream resolution but the camera install menu). Then you will double sensitivity...

The Bosch specification that I see only claims lux changing from from 0.0121 (5MP) down to 0.00825 (1080P). Since lux is a log scale and therefore 10 lux is actually only twice as bright as 1 lux, this indicates < 20% gain, not 100%.

Michael Miller

10/17/15 02:57pm

How people are still installing box cameras?

Undisclosed #1

IPVMU Certified | In reply to Michael Miller | 10/17/15 10:37pm

How people are still installing box cameras?

Clients who need higher performance.

I encourage you to take a look at Avigilon's HD Pro line, starts at 4K and it only gets better from there!

Michael Miller

In reply to Undisclosed #1 | 10/17/15 10:44pm

LOL

Tim Russell

In reply to Undisclosed #1 | 10/20/15 01:35pm

Doesn't 4k require 30 frames per second ? Not 12 ... Is Avigilons marketing is better than their maths ?

Undisclosed #1

IPVMU Certified | In reply to Tim Russell | 10/20/15 03:14pm

Is Avigilons marketing is better than their maths ?

English is their strong suit.

Tony Lagan

10/22/15 12:30am

I would also recommend that you try the new Hikvision darkfighters. The DS-2CD6026 has been superceded by the DS-2CD4026 which uses a larger 1/1.8" imager. From the testing we have done the 4026 has out performed the 6026.

Undisclosed #1

IPVMU Certified | In reply to Tony Lagan | 10/22/15 01:09am

Looks nice. And cheaper.

Not to mention another technological first, anti-thunder protection.

Take that Axis!

Tony Lagan

In reply to Undisclosed #1 | 10/22/15 02:10am

Hmmmm, well it is designed to stop the Dogs getting nervous in a Thunderstorm ;0)...

John Honovich

IPVM | In reply to Tony Lagan | 10/22/15 10:15pm

Tony, Hikvision says "both have a 1 1/1.9 sensor"

Bob Germain

In reply to Tony Lagan | 10/23/15 05:38pm

Disclosure: from Hikvision USA

I would like to help clarify this. The original datasheets for the 6026 stated 1/2" sensor. The actual size is 1/1.8" (John - my bad to you re 1/1.9").

The 4026 is a re-packaged version of the 6026 in order to bring a full line of DarkFighter cameras to the 4-series with the domes and bullets. It does have a different firmware file and defaults may be slightly different.

Tony and I have shared some dialog on this and his test results which in my view showed similar sensitivity but sharper images on the 4026. This could be due to changes in default values.

Undisclosed #1

IPVMU Certified | In reply to Bob Germain | 10/23/15 05:49pm

The actual size is 1/1.8" (John - my bad to you re 1/1.9").

Truly splitting hairs. ;)

Bob Germain

In reply to Undisclosed #1 | 10/23/15 05:50pm

I could round it to 1/2" :)

Read this IPVM report for free.

This article is part of IPVM's 6,602 reports, 890 tests and is only available to members. To get a one-time preview of our work, enter your work email to access the full article.