Do we really need 4MP? Dahua resolutions comparison.

[frank10]

It's finally arrived my first Ipcam, a Dahua IPC4421 with 3.6mm, firmware xxx.010R

I wanted to know how much difference there is from the max 2688x1520 to the classic 1920x1080 and intermediary res, 4MP and 3MP.
So I recorded with H.264H and extracted some frames and compared them.

Here the results:
https://imgur.com/a/EyCxH

The books for text comparing was at about 5m from the camera. The bigger text letters are between 7,5cm-4,5cm tall.
The Mb was the max available CBR, directToDisk in BI.
Just for fun, I uploaded also an old HP-USB (hp-4410) webcam at 1920x1080 (it seems similar lens, too): that is a great difference quality! Anyway it has its advantages: it's so tiny, you can hidden it marvellously.

To me it seems there is little increase in quality from 1920x1080 in the 3MP, but 4MP and more does seem to change nothing. BUT also in 1920x1080 you get pretty all the details you need: that could be a big CPU saving for BI!
So, at this point, why let's use 4MP res??
It's a pity that 4MP quality isn't there in major res and bandwidth... Maybe it should necessary a lot more Mbit in the higher res than 10Mb/s?

 

[MartyO]

I'm no expert but your logic make sense, probably the sweet spots are matching common monitors resolutions which is 1080p then 2160p (4k), which I think is 2MP and then 8MP when talking cameras. But four 4MP cameras displayed on a 4k monitor spec wise should look better than four 2MP on same monitor if you squint.

 

[frank10]

I don't know, I'm not convinced.
First, 4xcamera displayed on a 4k monitor, means you have 1920x1080x4, so the exact fullHD without rescaling, to completely fill the 4k monitor.
BUT I agree with you that new 4MP cameras are surely better than the old 2MP (but also 3MP) cameras. BUT NOT for the increased resolution, as I showed, but instead for the better electronics, better sensor, less noise, true WDR and so on.
So, my thinking is buying the new 4MP models BUT using them with lower resolutions, such as 3MP or still 1080p that's perfectly fine like the major res. The benefits could be huge in a pc with BI: you could set up a bigger system with less CPU resource used, with a lighter pc.

Anyway, this is my experience, I would like if someone else with a 4MP Hikvision could do a similar comparison, so we could see if this is a common trend for 4MP cameras (I think so). I saw also someone has 8MP: come on, let's see some comparison res between 8MP,4MP,2MP!

 

[LittleBrother]

Difference certainly seems small and I've no desire at all to upgrade from 3MP yet.

 

[frank10]

When I said buying the 4MP, I was thinking to build a system from scratch, having no camera at all.
If someone has some 3MP camera, I agree with you, it's not the case to upgrade at the moment.

 

[pal251]

I see nothing major in quality differences. I do have a 4k monitor but yet to get a 4k video card... soon


I think the differences in electronics also is the big difference. For the price difference now adays you might as well get 4k

 

[frank10]

For the price difference now adays you might as well get 4k

Do you mean 4MPvs3MP camera or 4K camera? Because 4k are a lot more expensive vs 4MP (minimum 3-4x).

 

[pal251]

I see nothing major in quality differences. I do have a 4k monitor but yet to get a 4k video card... soon


I think the differences in electronics also is the big difference. For the price difference now adays you might as well get 4k

I meant 4mp sorry

 

[stonewatch]

My preference is less resolution and a larger sensor. Unfortunately larger sensors are only used in expensive cameras. A 1/2" sensor 1080p is much more useful with low light than 1/2" sensor with 4Mp . resolution. Usually the larger each pixel is, the better the performance.

 

[MartyO]

My preference is less resolution and a larger sensor. Unfortunately larger sensors are only used in expensive cameras. A 1/2" sensor 1080p is much more useful with low light than 1/2" sensor with 4Mp . resolution. Usually the larger each pixel is, the better the performance.

I agree with you totally, quality is more important than quantity

 

[frank10]

I agree with both of you on general quality, that is linked to price, anyway.

BUT the point of my thread is not to compare different camera resolutions, sensor and quality, but only to check inside one camera itself, if it's really needed the greater resolution or if its equivalent also less res, saving CPU cycles.

I tested on Dahua 4MP, but it should be the same with a 8MP camera, or with a bigger sensor camera or with a more expensive and quality brand.
I asked for some other camera's snapshots, I hope someone will give other pics, so I can make a comparison here.

 

[tomw]

BUT the point of my thread is not to compare different camera resolutions, sensor and quality, but only to check inside one camera itself, if it's really needed the greater resolution or if its equivalent also less res, saving CPU cycles.

Electronics, lense and software make the difference. It is exactly the same televisions. You can even buy exactly the same panel in a TV the image will look very different across manufacturers depending upon the electronics.

 

[frank10]

True, that's the reason I would like to see some other brands pics to see how they work. At the moment the Dahua could be used at less res with quite same results. I quite think the same for hikvision up to 4MP. Let's see other brands or major resolutions like 4k.

 

[frank10]

Speaking of sensor dimensions and electronics, out of fun I tested 2 smartphones, Note2@1920x1080 and Note4@1920x1080 & @3840x2160 compared to the max res of Dahua @2688x1520.

Note2: 1/3.2
Note4: sensor is a sony IMX240, 1/2.6'', it's only some 15% larger than the average 1/3'' sensors used in other high-end smartphones.

It's not neither a 1/2'' sensor, it's only 1/2.6'' !!
The dahua is better than the Note2, but it's also @1920...
BUT with the Note4, even @1920 it's all another world! If you then look at 3840...
The light is not the same of the dahua test day, but it's similar (not direct sunlight).
The problem of course is nighttime... dahua is better in very low light, but if there is a mediocre lighting Note's also good. Maybe with proper electronic and reduction noise software it could be comparable.

So, we want such sensors in our ipcamera!

 

[NVR]

I don't think the method used offers enough info to conclude that there is no difference between a 3mp and a 4mp. But I still think we should have seen some differences in your images among different resolutions, which we dont and not sure why. Maybe not enough distance to see a difference, not really sure. Slightly visible in the smaller texts.


Here's a small test I did comparing my old 3mp Hik to the 4mp. This is exactly 7 feet away both cameras side by side.

 

[frank10]

One possible explanation is that comparing different models also of the same brand, the sensor changes of course, not only the resolution. So, also the image captured could be different due to the sensor quality, but also at the different lens quality.
Instead my test is related at the same sensor and I think it's possible that the camera always capture at the max resolution and then rescales it down to the lower res.
This could explain why there is so little difference from the max 4MP res and the 3MP but also the FullHD.
And again this is a test of Dahua cam, let's see some other brand or res. I'm waiting for pics from a Dahua 8MP...

But to compare correctly the images, you should scale them at the same apparent dimensions: some details aren't visible at lower dimensions whilst with equal dimensions they are, maybe defocused, but there. So I suggest you open those pics in different windows and zoom it at the same level comparable, then take a snapshot of the screen and look at them. This is more evident when comparing very different resolutions.
For example, if you have the hikvision 4MP, you could do the same test as mine: take a snapshot @max, @3MP, @FullHD, then open them in different windows (for example in Irfanview) or in Photoshop in different tabs; then zoom every image to the same apparent dimensions and take a snapshot of the screen, then post here the result. Thank you.

 

[NVR]

That was a quick snapshot the day I had everything out, agree with you, but when testing, one should be zooming in or cropping until there's notable difference, or distance the cameras far enough until pixelation starts. Resolution may appear the same at certain close distances and will only start to see variations until a further distance is reached. That unknown distance is an important mark when comparing cameras/resolutions, worthy or not, a 3mp should distort before a 4mp at a certain distance.

I was not aware you were using the same camera and only scaling down resolutions, likewise I don't think its a valid method when comparing cameras, as you mention, sensors change and vary in performance.

 

[zmx]

Day time images.... any 3-4Mp camera is going to do a pretty good job, Its low light that kills em. Thats where the good are sorted from the bad !

 

[frank10]

True.
At this point I made another quick test at night and I discovered that Note4 is way better than Dahua also at night! So, I was wrong before.
Of course Note4's software is not built to be a very low light system, so it lacks for example the great 3dNoiseReduction of Dahua, but the point of the test is to see what can be taken from the sensors. Then, when linked with the correct software and electronics, you can get better results.
So, here we go:
1) Note4
2) Dahua as is
3) Dahua with 3dNoiseReduction


View attachment Note4_6MP_Night_20151216_224337.jpg


I want Note4 sensor in a ipcamera!!

 

[techhi]

I have 4mp Dahua 2.8mm but mostly use them in 1080p so i can record more days on my hard drive and quality not much different from each other - the 4mp do have more features then the older cameras i replaced -
I also see some cams have H265 HEVC encoding and thats half the bitrate at the same quailty
Haven't tried them yet