Hikvision IR Panoramic Camera Tested

IPVM

You have to imagine that integrated IR panoramics will become commonplace in the next few years just like they have become so for domes and bullets.

Panoramic cameras inherently face low light problems - combine high F stops (f/2.0 or higher is typical) and higher resolutions (5MP common) and the cameras will struggle in low light. Enter IR.

John Bazyk

Command Corporation • IPVMU Certified

When testing IR performance, is there an option for "Smart IR"? I have seen it make a difference in domes and bullets but have never used this panoramic.

John Honovich

IPVM

In reply to John Bazyk

John, yes, some integrated IR cameras automatically adjust the IR power output depending on what is detected in the scene.

This goes by different marketing terms, such as 'smart IR' or 'adaptive IR', e.g., Axis calls it 'OptimizedIR', etc.

It's definitely useful when you only have one object in the scene. I suspect this would be trickier with panoramics because their super broad coverage area makes it more likely to have more objects in the scene simultaneously.

David Oleksy

•Feb 22, 2015

In reply to John Bazyk

These Hikvision fisheye cameras do have Smart IR. I wonder if it wasn't enabled for the IR tests.

Reactions:

Bob Germain

•Feb 22, 2015

In reply to David Oleksy

David- IPVM tests are done at Defaults (except maybe for normalizing shutter spedd for low light), so they normally would not enable Smart IR as it is not current default value.

Reactions:

Ethan Ace

•Feb 22, 2015

In reply to Bob Germain

All the images in this test were taken with Smart IR On. Though we did not see much difference between on and off.

Note that this test was a year ago and improvements could have been made in firmware. We'll be testing the Panasonic 4K panoramic camera soon, so will see if there have been changes.

Reactions:

Undisclosed End User #1

Do you need list the focus length of different camera, like 2.8mm or 3.5mm?

Here the "at close range (~6')" is the distance from camera to the subject,right?

thanks.

Ethan Ace

In reply to Undisclosed End User #1

The lenses vary slightly on panoramic cameras, but the field of view will always be ~180º. For example the Hikvision 6362 uses a 1.27mm lens, while Axis uses 1.3mm, Samsung and Hikvision both use 1.05mm, etc., but these all result in essentially the same field of view.

The measurements given ("at close range (~6')") is distance from the camera, yes.

Peter Pan

got it,thanks.

Jon Dillabaugh

Pro Focus LLC

Ethan,

Do these panoramic type cameras still cover a 16:9 area with respect to the field of view, or is it a circular area? What I am asking is, if you take your outdoor shot and place a number of cones on the ground at the edge of the FOV, is that patern on the ground subsequently a rectangular shape (16:9 matching the imager) or is it circular?

We are having issues with trying to model a panoramic camera in JVSG, which is the software we usually use to model all of our projects. Since I don't have one of these cameras in my hand, I am relying on you to help me figure this dilema out.

My guess is that it still covers a 16:9 area on the ground, hence you need to make sure you orient the camera's rotated view to match your desired FOV. So, for instance, I have a room that is 16' x 9' and I want a camera mounted in the center of the room, will I see an even coverage of that 16' x 9' room, or is the FOV actually circular and I will see more of a 9' circular FOV?

FOV A or FOV B?

Ethan Ace

In the case of most panoramics, including these, it's circular. The actual fisheye looks like this in the 6 MP in the room we test in:

You won't see much difference no matter which way you turn it because of that.

ImmerVision uses an elliptical fisheye to fill more of a 4:3 imager, seen here from our BrickCom test:

In that case, if you orient it to your room, you will see a small gain in pixels on target.

Neither of them are 16:9, though. I've always had issues calculating PPF in JVSG as well. I thought they were going to add panoramic cameras as a feature at some point, but I haven't seen it.

Rukmini Wilson

•May 18, 2014

Fisheye 360 degree lenses always distort right? Or do they? Consider a 360 spherical camera (not elliptical) placed at the center of the 'stage' below, facing upwards. Certainly the normal distortion effect would spherize the near (to us) side of the theatre, but what about the semi-circle stands? Would it be unchanged or look super curved? What if each step had the distance from camera printed on it? Any guesses?

Jon Dillabaugh

Pro Focus LLC

In an indoor setting, I think it will be harder to determine the actual FOV, due to the fact that you have walls that interfere with the ability to determine the true FOV. Do you have an image of an outdoor shoot that you could reference?

Basically, what I would do is set up a cam outdoors, or in a room large enough not to see the walls in the view, and setup cones (or something similar) at the edges of the cameras FOV. Is that cone patern a rectangular shape in reality (not what you see on the screen) or is it circular? What does that cone pattern appear to be in person, not what it appears like in the cameras view. Imagine you were flying overhead. What shape would those cones take?

The reason why I am asking this level of detail is that I don't know what I will get from one of these cameras, since they are not easy to model via software and I don't have one at my disposal at the moment.

I appreciate your help.

Ethan Ace

Well the FOV is 180 degrees or greater, both vertically and horizontally, so the FOV of panoramics isn't a cone. It's half of a sphere. If you mount them to the ceiling indoors you will in fact see the ceiling in many cases, and the FOV is bounded by the walls of the space.

So if you take the camera outside, the edge of the FOV is a horizontal plane extending from the camera til either: A) Infinity, or B) you hit an obstruction. I have clips from outside that would better show this, but I'll need to grab them for you from my other PC in little while.

Ethan Ace

Here's a still from a test I did last year which best illustrates this. Intersection daytime with a Vivotek FE8172V. They send a fisheye image which is square (1920x1920) because they crop out all the black space since it's pointless anyway. The sensor itself is 4:3. Keep in mind the size of the image and the actual FOV are two different things. Here the FOV is a 1920 pixel circle, but the image is square.

This was a flat area, and where you see the red arrows, the camera is seeing sky. Sky for miles if nothing is blocking it. Click the image for the full size version.

Stanislav Utochkin

•May 18, 2014

Thank you for the information. Could you answer for a more complicated question?

I suppose that the spatial resolution (pixel density, pixel per meter) is not flat along the FOV. There must be a mathematical dependence between the angle between the main optical axis of the lens and direction on a point in space and the pixel density. The more the angle is the worse the pixel density. Do you know the dependence? It would be interesting to made practical measuring the dependence. From my side I could develop modeling such cameras in VideoCAD.

Sean Nelson

Nelly's Security

Ive tested both of these cameras. I havent played with alot of fisheye cameras but when I tested these cameras, I thought the image quality was pretty poor. And Im talking about on the main "un-zoomed" panoramic view, the edges were very soft. Its almost like the lens they use is not providing a sharp enough image. Say for example the sharpness of their popular 2032 is way sharper of an image at the wide shot than these cameras. But like I said, I havent dealt much with fisheye's. Perhaps thats about as sharp as an image can be with that wide of a lens.

Jon Dillabaugh

Pro Focus LLC

Looking at the spec sheet from HIKVision for the DS-2CD6362F-I, it looks like the warped image size is 3072x2048, which is a 3:2 aspect ratio. This would lead to a rectangular FOV, if I am correct in my assumptions.

Undisclosed Manufacturer #2

Interesting looking at differences between cameras with Immervision and those using normal aspect lenses. There does not look to be much to be gained from Immervision - at least on the images shared.

It is interesting to put them head to head based on lens solution.

Good to know what lenses vendors are using. Many utilize the Sunex 360 lenses, some Immervision, some de-branded them. It also answers any questions on FOV and pixel mapping if that is known.

It's been a while since you last looked at an OnCam/Grandeye. Any plans to review their newer models, especially since Pelco have partenered ?

The Brickcom results seem to have gone downhill since the test against the Samsung SNF-7010 last month. And the Samsung has gone green !

This is one application where adjusting the Q factor (for the edge of the lens and blank areas which are constant areas on the sensor!) would be beneficial. I hope they are all doing that ?

John Honovich

In reply to Undisclosed Manufacturer #2

IPVM

"It's been a while since you last looked at an OnCam/Grandeye. Any plans to review their newer models, especially since Pelco have partnered ?"

No, because (1) there's nothing significantly different about OnCam/Grandeye cameras and (2) they are significantly more expensive than rivals. Finally, members overall seem to be a little tired of panoramic tests, so we are reluctant to do more tests here unless there is clear potential for something new (functionality, lower price, etc.)

Jon Dillabaugh

In reply to John Honovich

Pro Focus LLC

John, members are tired of panoramic reviews? I, for one, think it is an emerging product that I am trying to gauge if they are viable for my clients. I greatly appreciate the coverage your team has given.

John Honovich

IPVM

In reply to Jerome Humery

Jon, yes, overall members seem to be tired as the traffic numbers have declined on that series. My hypothesis is that we have done a lot of panoramic tests so that has been more than most care for.

Just like we don't test every box camera, we won't test every panoramic. However, whenever we see something different, we'll definitely prioritize a test.

Jerome Humery

•Aug 06, 2014

Basic question here:

I realize that there is a lot of distortion with panoramic cameras, but what is the alternative if your client wants a camera above an exit door in the middle of a facade and wants to see both sides? Is the norm to mount two cameras with a slight overlap in FoV?

Reactions:

Ethan Ace

•Aug 06, 2014

Normally, yes, you'd mount two cameras which overlap.

I've actually seen panoramic cameras used in the wild lately to cover a couple of exit doors. But even covering two doors, maybe only 10-15' wide, for example, the camera is going to provide far fewer details than a standard camera.

If you have other cameras covering a facility which will provide some identification of a subject and you're just using that camera to confirm which way they went or general actions, I'd say it's useful in that scenario. If you need absolute, no doubt, 100% ID quality video, though, I'd recommned against it.

Reactions:

(1)

Jerome Humery

•Aug 06, 2014