Subscriber Discussion

Why Cant Cameras Provide Uniform Pixel Density And Varifocal 180?

Avatar
Mark Jones
Mar 22, 2015

Since there is an awful lot of software in cameras, how hard would it be to design a camera that would provide a uniform pixel count over a given distance? 32 ppf at 50 ft, and 100 ft etc. Spread out the density using software.

While we are at it, why are imagers in 180 or 360 degree cameras always fixed? Why can't I move them? Mount them on a ring and let me set the position the customer needs?

U
Undisclosed #1
Mar 22, 2015

32 ppf at 50 ft, and 100 ft etc. Spread out the density using software.

The only real way to do this is to throw away some data at the closer distances. Why would you want to do that?

Avatar
Mark Jones
Mar 22, 2015

I am not sure why the data has to be thrown away, but if I could have a uniform pixels density over a given distance, I could live with it. The upside would be the ability to ID someone at 30, 50, 100, 150 feet. I think I would.

U
Undisclosed #2
Mar 22, 2015
IPVMU Certified

#1

#2 Did you plan on de-warping those off-axis 180 or 360 images?

Avatar
Mark Jones
Mar 22, 2015

Worth a read. I really enjoy this site. Thanks.

JH
John Honovich
Mar 22, 2015
IPVM

Pixel density is inversely proportional to the Field of View (FoV) width. FoV width increases the further you get from the camera, so the lower pixel density will be. Think about what happens the further you move from an edge of a triangle. So, by essence, the pixel density has to diminish.

The only way to get around that is to use a series of multiple cameras in a box, like Dallmeier does with their Panomera, but that's extremely expensive.

(2)
(1)
U
Undisclosed #1
Mar 22, 2015

Let's say you have enough resolution to get 80ppf at 200'. For closer distances, you are going to naturally have MORE ppf, since the field of view will get narrower, but the net pixel count (imager) stays the same.

The exact difference depends on the FOV angle, but let's say you have 80ppf at 200', you might have 160 ppf at 100'. If you want uniform pixel density across the image, you have to manage to the lowest ppf number, which is 80ppf in this example. So at the 150' range you'd be essentially throwing data away so that you can get uniform density across the image.

The alternate is to use multiple imagers setup to cover a distance, but even so you're going to have some areas with slightly more ppf than others. Again requiring you to throw some data away if you want truly uniform pixel density throughout the image.

Nobody that I've met ever really wants *less* resolution in any common part of the image, so you don't see this sort of setup. Additionally the multi-imager arrangement to cover longer distances (vs. wider arcs as in a 180 degree camera) is non-trival to configure and handle in the viewing software (read: more expensive).

Really what you do in the scenario you're describing is solve for pixel density at the maximum range, and then closer ranges just end up with an even better image.

JH
John Honovich
Mar 22, 2015
IPVM

"why are imagers in 180 or 360 degree cameras always fixed? Why can't I move them? Mount them on a ring and let me set the position the customer needs?"

If you do that, you will no longer have a 180 or 360, you will have a 150 or 130. Its just a fundamental principle, you 'move them' out, the lens is longer and the FoV / angle is narrower.

Practically speaking, I would assume it is hard, if not infeasible to build a lens that can varifocal from fisheye to telephoto.

In practice, varifocal surveillance lenses generally do not go wider than 120° or so (see Camera Finder varifocal wide angle results).

U
Undisclosed #2
Mar 22, 2015
IPVMU Certified

Is it clear to you that he means changing the focal length? It sounds like he could mean adjust the xyz positioning thru a gimbal mechanism.

Mark?

Avatar
Mark Jones
Mar 23, 2015

You are correct B.

The question started out almost rhetorically. I was blowing off some steam after hours of fruitless research. I understand that the futher way you move from a camera imager the fewer pixels you will have to work with, given that the field of view is larger on one end than on the other. The answer I was wishing for requires out of the box thinking.

I won't speak for everyone here, but lots, and I mean lots of customers watch too much NCIS or CSI. They want to "zoom" in on a recorded image and NOT have pixelation. You can have it, but it requires a lot of pixels.

My other issue with the current approach is that the field of view is conical in shape. Typically not nearly wide enough at the beginning and too wide at the other end. The multi-imager does "appear" to elliminate that problem. It can cover an entire parking lot by itself, back to front, side to side, offering more than adequate pixel density for ID over the entire image, all without trenching a foot, or tearing up any asphalt. If the customer doesn't have the cash, that is his problem. I have met his expectation. He has to make some decisions.

As for part 2 of my diatribe, I saw an add just last week for a new Axis camera that has imagers mounted on a ring. It was a PTZ and I really don't recall the model right now. However, if they can be positioned on a ring, why can't they be moved along that ring? maybe I don't want or need 360 degrees.

Having said that, I do like the Dallemier camera. I will be giving them a call tomorrow.

My apologies for spelling.

U
Undisclosed #2
Mar 23, 2015
IPVMU Certified

I think you mean the Q6000-E multi imager panoramic mounted concentrically with the Q-60 PTZ, reviewed here. Though I don't think you can buy the panoramic by itself.

Arecont has had a movable multi imager for a bit now, it's reviewed here.

Your two questions actually overlap a bit since you could look at the Arecont, if setup with lenses of different focal lengths, as being a 'poor mans dallamier'. 'Poor' because without the dallamier stitching software it couldn't be ePTZ'd as one single image.

What I was alluding to in my original response to your question about the 180/360 fixed imager, was that I think that for the dewarping to work well, that the camera needs to be on flush on the ceiling or wall. I could be wrong about that though, I'll try it on my q24 guinea pig and see.

DW
David Westberry
Mar 24, 2015
IPVMU Certified

"While we are at it, why are imagers in 180 or 360 degree cameras always fixed? Why can't I move them? Mount them on a ring and let me set the position the customer needs?"

I dont understand the need. If a 180* is mounted on a wall it sees everything wall to wall. A 360* sees all the way around. What purpose is there to move? If you need less viewing area you dont need these types of cameras. *confused*

U
Undisclosed #2
Mar 24, 2015
IPVMU Certified

Consider wall-mounting a 180 at a height of 10 ft. You look at the image and are mortified at the amount of space the ceiling takes up in the image (should have used the calculator.)

So, the natural tendency is to tilt the camera down a little, to put the center of the lens more on the subject below, even though what you lose in ceiling you gain in floor. You would gain a higher PPF for the subject.

Problem is, with my Mobotix at least, is that the dewarping seems to rely on the assumption that you are mounted on a plane, and the more that you are off plane, the more distorted the dewarped view looks.

New discussion

Ask questions and get answers to your physical security questions from IPVM team members and fellow subscribers.

Newest discussions