At one point Axis declared the megapixel race was over, but now parent company Canon is showing an imager that would be a rocket ship in a megapixel race. Canon says they are targeting surveillance applications, not photography, with this sensor.
Will this imager find its way into an Axis product? Is a 250MP surveillance camera practical?
Inside this report we detail Canon's prototype imager, and the impact it could have on Axis, Canon, Avigilon, and the security market in general.
Canon ***** ****** ********
** * ***** ********* * ***** representative ******** **** ************** *** *** imager, key ******* *******:
** ******* **** ******** ** ***** of ******** ****, ***** ******** ** other ***** *******, *** *****, ** low-light ************.
Technology *** *** *****, ******* *** *** ****
***** ***** **** ******** ****** *** technical ************** ** *** ****** ******, a *********** ******* ***** ********* **** under *** **** *****. **** ***** give **** ******** ****** *** *** highest ********** ******-****** ******, ********* *** company *** *** ********** **** **** to *****. **** ***** **** **** Axis ****** **** ** *** ****** leadership **********.
Technical **********
************ **** * **** ********** ****** would ****** ******* * ****** ** challenges. ***-***** ************ ***** ****** ********* ** limited, *** ** ****** **** ******** technology *** ****, ********* *** ******* costs ***** ** ***********.
*** ************ *** **** ** **** complex, *** ***** ********* **** **** produce **** ********** *******, ******** *** *******, **** ******* ****** ***** ******** to ***** *** ***, ********* ******** that ** ** *** ********* *** 'mainstream' ***** ** ******** ******* ******** super **** ********** ******* ******* ********* the **** **********. ** ******, **** could *** **** ** * ****** to ******* ***** *** *** ********, which **** **** **** ******** ********* in *** ******.
Avigilon ******
* ***** **** ****** **** ********* a ******* **** ********** ***** **** Avigilon's**** '**'****** **** ****** ********** ** **********. Though *** **** ****** ****** **** not **** ** * *********** ******* of ********'* *****, ** **** ******* the *********** ** ******** *** ******* as * ********** ****** ** ***** of ******-****** **********.
Market ******
***** **** ********** ******* ****** * very ***** ******, **** **** ***** resolution (** **********) ** ******* **** not ***** ***** ********. ********* ************ for * ***** ****** ***** ****** remain **** *******, ****** *** ***** of ************ ***** ** ***** ** beneficial (******* ********, ****** ********, ********/******) are ********** **** ****** ***********. ****, Canon/Axis *** ******** ** ******* *** market **** * ****** ***** ** this ******, *** ** ***** **** them ******** ********** ** ********** ******* over ******.
... a productized version would certainly come under the Axis brand.
That would make the most sense. But why are you so sure?
I realize that the Canon security line is now sold and supported by Axis, but has there been a statement made that CS will no longer produce any new products?
And Canon has often confounded those who assume that they will use their multiple security holdings to their best use.
Again, I do agree that they should do it, but will they?
Canon did not exhibit at ASIS, and is also not exhibiting at ISC West 2017, instead having their products in a rather small section of Axis' booth (at least at ASIS, we do not know what they will show at ISC West, just that they are not currently listed as having a booth).
The Axis 20MP camera that was co-developed with Canon was brought to market under the Axis brand.
And Canon has often confounded those who assume that they will use their multiple security holdings to their best use.
I think that many people, myself included, expected Canon to take a more active role in both Axis and Milestone, possibly combining them into a more unified offering. That has not happened, Canon has mostly left them alone. If Canon had made a number of obviously bad moves, then I might agree there was more risk of Canon doing something unbelievable, like marketing a new 250MP security camera under their brand instead of Axis'.
I think Canon realizes that their brand is not strong in security, but their technology innovation capabilities are significant. With that in mind, I think if they build a full product around this sensor, it would logically be under the Axis brand, and Canon has not done anything to show they would approach this differently.
...Canon has not done anything to show they would approach this differently.
I am not sure if you are offering that as supporting evidence to my statement, or as an argument.
The camera you posted in that link predates the shift for Axis to take over sales and marketing (outside of Japan) for Canon products, and that was also a relatively minor product, compared to existing things and available technologies.
...that was also a relatively minor product
Not as far as Canon was concerned:
This new camera will totally revolutionise low light surveillance and again show the world why Canon is a global leader in IP Network Cameras.
As for
The camera...predates the shift for Axis to take over sales and marketing...
That's true, it did by several months. Though it was still close to a year after Canon acquired Axis.
You had said
...it would logically be under the Axis brand. Canon has not done anything to show that they would approach this differently
My point was only that this was a counter-indication.
"I think Canon realizes that their brand is not strong in security, but their technology innovation capabilities are significant. With that in mind, I think if they build a full product around this sensor, it would logically be under the Axis brand.."
I hope so, or they will make the same mistake Schneider did, thinking his brand was stronger than Pelco, even for the security business
This one camera would replace 10,554 analog cameras right?
OMG. It is not PS, right ?
Some of these cameras are looking different angle.
And what about the lens? I don't know if any actual lens can achieve this optical resolution... INSANE!
So the sensor is 29.2mm wide with 19,580 horiz. pixels.
So ~1,500 pixels per mm.
So maybe a lens with 750 lp/mm resolution (1/2 of 1500 because they are line pairs) could keep up with the sensor.
Do you know of any 750 lp/mm lenses (MTF50)?
I would think since Canon is making the sensor and since Canon makes some of the best DSLR lenses on the market that they would have this figured out without issue.
I would think since Canon is making the sensor and since Canon makes some of the best DSLR lenses on the market that they would have this figured out without issue.
Don't know what you mean exactly by 'figured out without issue', but I don't think there is a Canon lens for APS-H size sensor, that has anywhere near a 750 lp/mm (MTF50) resolving power.
Which means the image resolution is limited by the lens well before its limited by the sensor.
Think about it, the pixel density of the 250mp is 10x of other high-end DSLRs, so the lens requirement is going to be 10x as well. And 100 lp/mm is on the high-end for the usual DSLR lens.
Its not a tragedy though; whether it's the sensor or lens, one always limits the other to a degree.
I hope you're right and they have a breakthrough lens coming out especially for this sensor. I'd welcome it.
Correct me if I wrong.
29.2mm with 19580 pixel
670.5 pixels per mm.
Lens need 335 lp/mm "only", right?
But I don't know which lens can do over 100 lp/mm @ MTF50......
yes, my math is backwards.
still the lens is hard to find. i think there is a new gopro lens with 400/lp, but its not big glass aps-h stuff.
The measurement of lp/mm is simliar to Lux in video surveillance
If the product got some lousy result from the test.
All of the manufacturers will do the same trick.
Using the nice number from a lower IRE/MTF.
400 lp/mm .... It's just too good to be true.
I believe the 500mm & 600MM L II Lenses resolve out to about 170 lp/mm. The 500mm lens is $9,000, and probably far more zoom than you would want for a 250MP security camera application.
I really don't think this one can resolve over 100 lp/mm (Not even the center point)
Canon MPF chart always using 10lp/mm for testing.
This one is good but not that good.
Any infomation about this test of 170lp/mm?
Maybe they were testing it under MPF 20?
I think maybe some kind of lens from boardcast grade can do it.
But I'm couldn't believe gopro lens or Canon EF can do it.
Theoretically, this camera would occupy the same bandwidth of 250 1 Megapixel cameras that would be arranged in a more clever and better distribution than just pointing to one same local.
When I tested my first video capture card, in 1997, it had only 1 fps and QCIF resolution. At that time, it had much better quality than a time lapse recorded image. Telephone line modems supported only a 56 Kbps bandwidth, but enough to transmit that images.
Today we would laugh from that; we have much better resolutions and bandwidth.
But we still don't have a bandwidth large enough to support several 250 Megapixels cameras. Certainly the day we will laugh of 10 megapixels cameras and 10 Mbps bandwidth will come.
Until there, except for a very few applications, it's a camera for the future, not for today.
Canon made a very fast car but too big to fit in our current highways.
I agree with you, Brian: This is more to put their flag in the top of the highest resolution mountain.
A 10G fiber plant and 40/100G switching would be costly but a fraction of the SSD SAN that would be needed.
How much floor space do I need in my data center to process and analyze video from each camera?
Just rent it from Amazon. They already have more than enough capacity to handle something like this.
Would you like a spoon of storage ?
http://www.extremetech.com/extreme/186797-the-liquid-hard-drive-that-could-store-a-terabyte-of-data-in-a-tablespoon-of-fluid
While I am sure there is a 250MP imager security camera in our future, I think the low frame rate (<5), bandwidth, storage and lens and probably lighting requirements may limit practical applications of it to a few niches.
Like others have pointed out, there aren't too many applications where this can replace 250 1 MP cameras. Still I would be curious to see if it could perform in the real world or is it just a bragging rights technology.
I agrre with all you mentioned, mainly because you said "probably lighting requirements..."
I have to confess I was a bit skeptical when I heard that Avigilon had launched an IP camera with 29 Megapixels, 0.1 Lux, since I didn't have good experiences with cameras of only 2 Megapixels in low illumination. It was 2 years ago.
To check if it was really possible or a big lie, I made a theoretical analysis of this camera features and I proved that, yes, it was possible a 29 Megapixels having a sensitivity as low as 0.1 lux.
You can check my analysis here: http://www.cctvinstitute.com.br/29-megapixels-ip-camera.html
Now, 2 years later, Canon, the very same company who supplied sensor and and lenses for that Avigilon camera, is launching a 250 Megapixels camera proclaming that it is very good at low illumination. Well, as technology had advanced a bit since 2 years ago, maybe it really is...
Hi Claudio,
The 29Mpx camera is so old, also 2 years ago. You may check the 30Mpx (7K) full-frame model. And it works below 0.1Lux (0.005 at F/1.4). It's only about the exposure time and lens aperture.
Really ? Work below 0.1lux (0.005 at F/1.4)?
Lux from cameras are absolute hokum.
Only two reasons why people use this.
They really don't know how to measure Lux.
Or they are intentionally misleading customer.
He did say
It's only about the exposure time and lens aperture.
so if you hold the shutter open long enough anything is possible, if meaningless at the same time.
Exposure control and lens aperture is a key factor to build a scene. You will set the exposure value and the lens aperture according to the scene requierements. No meaningles. Sometimes you need less fps, the relative movement in the scene is really low and you can set the exposure to really low values without blur. Every scene is a world.
But, only one question. Are you saying Avigilon is deliberately lying in the camera datasheet?
Are you saying Avigilon is deliberately lying in the camera datasheet?
I'm not talking about the datasheet, I'm talking about what *you* said:
And it works below 0.1 Lux (0.005 F/1.4)...
This statement is meaningless without knowing the shutter speed... You can capture distant stars with a slow enough shutter.
There are more questions even if a shutter speed is specified (like what constitutes a usable image), but without one it's a total non-starter.
Every one lie.
Too many factors can change the final result even public the IRE:
Lux meter spectrum, calibration, lights, Lux meter angle, reflectance of the object, shuttle speed, F-number......
What is "work" without IRE ? Who define it is OK or not ?
IF this picture capture under 0.0001lux.
Can I told everyone this camera works below 0.0001 lux ??
Well, for photography, to take still pictures, this is valid.
But for CCTV, images in movement, for real time monitoring the ideal is 30 fps.
For record purposes, with few exceptions, like casinos and money counting, above 5 fps is OK.
Regarding Avigilon, at least for their 29 Megapixels camera, in my article http://www.cctvinstitute.com.br/29-megapixels-ip-camera.html, I've checked that they may not lying.
I think what really matters is the sensitivity at 30 fps. Bellow that, images in movement become to be blurred, useless for identification, for CCTV purposes.
Some cameras specs says: 0,0001 Lux at 30 seconds exposure time.
30 seconds? Is this information helpful? What is the use for that? To record sloths in a zoo?
When someone specifies sensitivity, the correct is to also mention at which IRE level and exposure time.
There's no direct relation. The motion blur is about long exposure time. If you are going to 1fps but your exposure is about 1/60 or 1/100, your image will be detailed and not blurred (asuming you are not monitoring cars (you will need to go shorter in the expopsure time)
For record purposes, with few exceptions, like casinos and money counting, above 5 fps is OK.
Recording is the most importat part in a CCTV system. Is where you have the evidence. The 30/25 fps were important in the analog age, whit limited resolutions viewing small areas. In the megapixel age, fps are not so important because your monitored area is bigger. In fact, they are a waste of disk space. Only some applications as casinos (i.e.) may need real time movement. But then, record this stream too!
This is my opinion :)
Yes, frame rate is not the same as shutter speed.
But you still haven't given a SS for the F/1.4 0.005 lux measurement.
You're right: I made a mistake; blurring would occur with fast movements at slow shutter, not low fps. Problem with low fps is the lost of information.
But I disagree when you said: "The 30/25 fps were important in the analog age, whit limited resolutions viewing small areas. In the megapixel age, fps are not so important because your monitored area is bigger."
Since the time-lapse age, it was determined that, for security proposals, at least 5 fps is enough to not loose any important information.
I remember a video from Pelco showing a line of people passing by a table with an ashtray on its top.
At 1 fps, suddenly the ashtray 'disappears' and you can't know who took it because it happened between frames.
But at 5 fps you can clearly identify who took the ashtray.
Therefore, to detect quick and suspicious movements, doesn't matter how big or small is the area nor the quantity of information (resolution) you're recording, but the quantity of movement (fps) is being recorded.
Someone remember and still have this video? I've been looking for it to show in my trainings.
I don't have to give any SS!!! Maybe Avigilon... :)
