Comparison of Axis 223M and Mobotix 12M

[cglaeser 0]

Does anyone have experience with both the Axis 223M and Mobotix 12M or is there a review on the net that compares both? In particular, how do they compare wrt low light at night? The specs of each camera list the minimum lux level, but the 223M specifications do not mention the shutter speed used to determine that lux level, so it's difficult to compare based on specifications.

 

Best,

Christopher

[Monitor Your Assets 0]

Hi Christopher,

 

I know it's not exactly what you're after, but attached is a sample of the M12D I'm evaluating which has the same lens - the 43mm. Specifically, it is a M12D-IT-DNight with dual L43 lenses - 640x480 only.

 

The SUV driving by triggered the PIR sensor which took this shot. I'm playing around with the variables so that each image/video recorded will include the illumination and exposure settings so that I can dial it in.

 

Hope it helps!

[cglaeser 0]

I know it's not exactly what you're after, but attached is a sample of the M12D I'm evaluating

 

Wow, that looks pretty good! Do you know the shutter speed? The car is not blurred, but it may have been driving very slow in the snow.

 

I have an Axis 223M and although we don't know the lux value at your house vs mine, judging from this photo, I would say the M12D has much better low light capability compared to the 223M. Of course, the M12 has a sensor devoted to night, and I also think the M12D may have a much larger sensor as well.

 

I also like the M12D form factor compared to the 223M. The 223M requires a large enclosure, while the M12D is built for outdoors. Plus, the PIR sensor is included with the M12D, and I had to add a separate PIR for the 223M.

 

The M12D looks like a real winner. Feel free to post more information as you continue your testing.

 

Best,

Christopher

[cglaeser 0]

Here are some low light specs from the two cameras.

 

Axis 223M

1/2.7" sensor

B/W 0.2 lux F1.4

 

Mobotix M12D

1/2" sensor

B/W 0.1 lux 1/60s F2.0

B/W 0.005 lux 1/1s F2.0

 

- The M12D has two sensors, one each for day and night. The 223M has one sensor with an IR cut filter to cover both day and night.

 

- The M12D sensor is much larger than the 223M sensor, and thus has the potential for better low light characteristics, everything else being equal.

 

- The 223M specs do not indicate the shutter speed (exposure time). The camera is capable of a 2 second exposure time. Since they do not include the exposure time used to measure the low-light lux level, it's not clear if they used an excessively long 2 second exposure time or a more realistic 1/30s exposure. The M12D specs list the lux levels for two exposure times (all manufacturers should list at least one exposure time for the data to be meaningful).

 

- The M12D has five fixed lenses from which to choose. The aperture on all the lenses is F2.0 except the longest lens which is F2.5. The 223M has a varifocal lens from 4mm to 8mm. The aperture at 4mm is F1.5. Axis does not specify the aperture at 8mm, but the aperture is likely much smaller (higher F value) than 4mm, which would reduce the light to the sensor. Reducing the aperture by one stop will reduce the lux sensitivity (increase the minimum lux number) by one stop.

 

- It is easier to optimize the performance (sharpness in particular) of a fixed lens vs a varifocal lens.

 

Since the Axis specs fail to include the exposure time used to measure the low light lux rating, it's difficult to compare the night performance of the two cameras on paper, but the list of issues above make the M12D a compelling design compared to the 223M.

 

Best,

Christopher

 

PS: Kudos to Mobotix for including the exposure time in their product specifications.

[Monitor Your Assets 0]

Nice comparison of the two cameras Christopher - you know your stuff!

 

Here is a sample taken from the day lens - VGA provides enough detail to *just* recognize the license plate of the Impala in my driveway (first 4 characters blurred on purpose), but not plates on the road. Not bad from that angle, IMHO. In this picture, the Impala and Nissan are parked - the black Toyota is moving and is what triggered the PIR dectector (indicated by the colored "PI" at the bottom left).

 

 

I'm happy with the quality, especially being VGA and I hope to dial it in further. Aesthetically, the M12 blends in nicely above my garage door. If I disabled the LEDs, few would notice it. I've had a lot of fun with the two way audio too!

 

I don't recall the shutter speeds that were used for either picture, but am now experimenting with adding variables to each image (very easy) so that I can make comparisons and tweak it further.

[cglaeser 0]

I don't recall the shutter speeds that were used for either picture, but am now experimenting with adding variables to each image (very easy) so that I can make comparisons and tweak it further.

 

I am especially interested in the shutter speed (exposure time) used at night. The 223M produces great images of my driveway at night using a 1 or 2 second exposure time, but even very slow moving objects are just a blur at that setting. Please post your observations and images when available.

 

I placed an order for a RayMAX 50 Fusion IR Illuminator. No word yet on when it is due to arrive, but looking forward to using the 223M with an illuminator.

 

Also, where did you get the M12D? The Mobotix website has a list price of something like $1,195, but I have not seen them for sale on the net for that price. Did you get a loaner for testing? Would love to get a loaner myself.

 

Regarding the M12D form factor, I agree, I like the design compared to the standard CCTV enclosure. The M12D would blend in pretty well with the front of my house without too much clutter.

 

Best,

Christopher

[Sawbones 0]

Say... that's a nice, sharp image. Sharper than I thought it would be for only being 640x480.

 

I'd considered getting a Mobotix camera for over my driveway (currently have a Surveilux PTZ there), but wasn't sure whether the VGA model would be good enough (was leaning towards the 3mp "mega" model). I have to say your VGA image looks much better and sharper than I thought it would... and much sharper than my 4CIF PTZ.

 

Can somebody shed some light on this for me?

 

*************** EDIT ****************

 

Here's a sample image from my driveway:

 

[amirm 0]

Can somebody shed some light on this for me?

Thank you for posting your image. We can do a lot with that bit of information.

 

Here are some observations:

 

1. You see how the diagonal lines in your driveway are jagged/rough? That is due to an interlace (normal TV) camera sensor. In presence of slightest camera/subject motion/vibrations, you lose half your vertical resolution resulting in kind of artifact you see here. The camera or DVR software can filter these but then you get a softer image, not a sharper one. Advanced de-interlacing like used in better built A/V equipment does a better job here but I have not seen evidence of it being used in CCTV market yet.

 

The sensor in Mobotix is progressive and hence, has full 480 lines of vertical resolution.

 

Note that this kind of artifact makes the video harder to compress. So all else being equal, you also get more compression artifacts at the same data rate as that of a progressive sensor.

 

When picking an IP camera, it is super important to make sure it uses a progressive sensor. There is no reason (other than saving a couple of dollars in manufacturing) to use an interlaced sensor. I have noticed for example that some of the Acti IP cameras use interlace sensor (evidenced by the same artifact shown here in their demo material).

 

2. Cross-color/False-color visible in details. Look at the basketball net. You see the red/blue color bleeding there? That is an artifact of NTSC decoding not done right and/or fault of the NTSC format itself. same thing is visible in the bush branches near the mailbox. There, you see that artifact combined with problem #1, making a mess of that object.

 

Analog comaras sending their NTSC signal over a single composite (coax) wire will have this sort of artifacts which reduces effective resolution of the image once more.

 

As with #1, the extra "information" (false color) makes compression harder resulting in either larger files (if data rate is allowed to be higher) or more artifacts (if data rate is kept constant).

 

Again, the progressive sensor in an IP camera is sending pure data and not subject to NTSC flaws that come from sending that signal over coax wire (as opposed to a fraction of an inch inside an IP camera).

 

3. I see more compression artifacts. For example, let's look at the camera captions. We see what we call "ringing." These are halos that get created at the edges of objects (the sharper the edge, the more you see them). On top of that, you also see problem #2 in them with false pink shades bleeding into them as the signal had to be modulated into NTSC and demodulated back in the DVR. This again makes them harder to compress.

 

I see a lot of "blocking artifacts" in the cement driveway. These are square blocks that show up when that segment of the video is overcompressed. Another manifestation is image detail completley getting destroyed when these artifacts get too high. Look at the line at the intersection of the driveway and road. That line is barely visible. It is destroyed by combination of all three factors explanied here.

 

Grass is very difficult to compress and here, we see them becoming a sea of little blocks/blurriness, taking away their nature texture. On this front, I would say your image was harder to compress so on this bit, I think the comparison is not quite fair.

 

Partial olution for #3 would be to boost your data rate/quality setting. Set it way high for testing and you should at least eliminate problem #3, leaving you with #1 and #2.

 

4. NTSC signal sent over a coax wire CANNOT send 4CIF worth of horizontal resolution. The system is not cable of doing that, period. So your image has lower resolution than the sensor would indicate. Again, Mobotix doesn't use NTSC so it can send all the signals as it sees them. See my post in the middle of the page here: http://www.cctvforum.com/viewtopic.php?t=10415&start=45 (how do you link to single post here?). You are losing nearly 40% of your resolution this way!

 

Net, net, this is a classic example of how well a progressive IP camera works relative to an analog TV. Unless you come and tell me your camera was also IP in which case, I have to make a visit at this company and teach them a thing or two about proper design .

Edited January 25, 2009 by Guest

[cglaeser 0]

Can somebody shed some light on this for me?

 

Have you seen exceptionally good photos from a high-quality digital camera posted on the internet? The camera may be 10 megapixel or more, and then the resolution is reduced to VGA sized to fit on a web page. Photos taken with a high megapixel that are then down-resed typically look much much better than photos taken with low-end cameras with only a modest number of pixels.

 

There are other factors as well. Cheaper cameras not only have fewer pixels, but the pixels themselves may be very small, which means they collect less light and have more noise. In addition, high-end cameras typically have multi-element glass lenses, whereas low-end cameras may have plastic lenses.

 

Best,

Christopher

[Monitor Your Assets 0]

I am especially interested in the shutter speed (exposure time) used at night. The 223M produces great images of my driveway at night using a 1 or 2 second exposure time, but even very slow moving objects are just a blur at that setting. Please post your observations and images when available.

 

You bet!

 

Also, where did you get the M12D?

I am a Mobotix dealer and want to learn as much as I could - best way is to use one. I thought I'd start with a model that has a full feature set (built in DVR, especially), two lenses and the "worst case" images - VGA. PM me for more details as I do not wish to violate any forum policies!

 

the M12D form factor, I agree, I like the design compared to the standard CCTV enclosure. The M12D would blend in pretty well with the front of my house without too much clutter.

 

Here is a pic of the camera mounted above my garage door (top left), which keeps the camera out of the elements, but not the cold. Temperatures have dipped to -20c (-4f) with zero effect on the camera. The camera records the average hourly internal temperature over the last few days, which is interesting. Last night, the internal temperature dipped to -4c with an outside temperature of -18c.

 

 

amirm: excellent analysis of sawbones' picture! Similar to his grass, some detail is lost in my snow, but I think that's just a matter of dialing it in.

 

Great thread guys! In reviewing my post, I realize it's time for me to take my Christmas wreaths down!

Edited January 25, 2009 by Guest

[amirm 0]

Have you seen exceptionally good photos from a high-quality digital camera posted on the internet? The camera may be 10 megapixel or more, and then the resolution is reduced to VGA sized to fit on a web page. Photos taken with a high megapixel that are then down-resed typically look much much better than photos taken with low-end cameras with only a modest number of pixels.

You are correct in this point but I believe both images are posted at their sensor resolution. The second one seems to be set to PAL scan rate though. So it is a bit larger because of that. Here is a resize of it to VGA. It gets a bit sharper but all the issues remain:

 

 

There are other factors as well. Cheaper cameras not only have fewer pixels, but the pixels themselves may be very small, which means they collect less light and have more noise. In addition, high-end cameras typically have multi-element glass lenses, whereas low-end cameras may have plastic lenses.

 

Best,

Christopher

Indeed, these are all factors that also contribute. But they are harder to prove using just an image presented here. Nor do I believe they are the top reasons the image is so much worse than the other subjectively.

[cglaeser 0]

You are correct in this point but I believe both images are posted at their sensor resolution.

The M12D has two sensors and both have considerably more pixels than VGA.

Indeed, these are all factors that also contribute. But they are harder to prove using just an image presented here.

Take a look at the purple fringing and CA.

 

Best,

Christopher

[amirm 0]

The M12D has two sensors and both have considerably more pixels than VGA.

 

Not the "IT" model presented here:

 

Hi Christopher,

 

I know it's not exactly what you're after, but attached is a sample of the M12D I'm evaluating which has the same lens - the 43mm. Specifically, it is a M12D-IT-DNight with dual L43 lenses - 640x480 only.

 

Mobotix builds great cameras but has made a mess of explaining the resolutions/models of their cameras. Only the "Sec" models have megapixel resolution.

 

Take a look at the purple fringing and CA.

 

Best,

Christopher

It is hard to seperate CA (lateral chromatic abberation) from NTSC/PAL artifacts. My money is on the latter though .

[cglaeser 0]

Not the "IT" model presented here:

 

Thanks for bringing that to my attention, I missed that. That explains the excellent low-light photo (I thought it was a down-res'd megapixel). I'm interested in the SEC model. Would be interesting to compare the night photos of the SEC.

 

Best,

Christopher

[Monitor Your Assets 0]

Mobotix builds great cameras but has made a mess of explaining the resolutions/models of their cameras. Only the "Sec" models have megapixel resolution.

 

With only one exception: The M12M-Web-D43 which has a day only megapixel resolution (2048x1536). It is a stripper model that does not include several features such as internal DVR/storage, 2 way audio, telephony, external storage on Samba/NFS and video motion. It's more suited as a "web cam".

[Sawbones 0]

Just to add a second perspective, here's another screen grab from the identical scene, and the identical camera. This time, however, I took a screen grab with the camera detached from my embedded DVR and attached to an Axis 241q. Bit of a difference, wouldn't you say?

 

 

The Axis unit claims that it "de-interlaces" pictures. I don't know if that's an Axis euphemism for "progressive scan," but the picture is much cleaner through the Axis unit than through the DVR (a Dedicated Micros DS2).

[amirm 0]

Just to add a second perspective, here's another screen grab from the identical scene, and the identical camera. This time, however, I took a screen grab with the camera detached from my embedded DVR and attached to an Axis 241q. Bit of a difference, wouldn't you say?

Yes, big difference in reduction of compression artifacts.

 

The Axis unit claims that it "de-interlaces" pictures. I don't know if that's an Axis euphemism for "progressive scan," but the picture is much cleaner through the Axis unit than through the DVR (a Dedicated Micros DS2).

The two terms are not interchangable. Let me explain.

 

An interlaced sensor sends the horizontal pixels the same as a progressive sensor. However, in vertical dimention, in one shot ("field") we get the odd lines, and in the next one, even ones.

 

Now, if the camera and subject are 100% stationary, then nothing is lost in using interlace. We paint the odd lines in 1/60th of a second and even ones at the next 1/60th of second. Your eye would "filter" these into thinking all odd and even lines were sent at once.

 

But life is not made up of static objects and rock solid camera mounts. As soon as either moves, the odd and even lines are no longer from the same image, but different ones (due to things moving 1/60th of a second later). If the motion is fast enough, your vertical resolution reduces from 480 lines to 240 lines (i.e. odd and even lines capture entirely different images).

 

So in a nutshel, our broadcast standard has a variable vertical resolution. For motion, it has half the resolution of static objects.

 

There are two solutions here:

 

1. Average the two "fields" and created one "frame" out of them. This would smooth the differences between the lines, replacing the jagged line with a softer, albeit lower resolution image.

 

2. Using motion tracking to attempt to synthesize what could have been there if we did not have interlace. This is hard to do right and hence my comment regarding it not being as common in this space.

 

I don't know which scheme Axis is using but neither is the same as having a progressive sensor at the start. Interlace is a complex transformation of the image and it is very hard to undo its effect after the fact. The ONLY way to avoid it is to use an IP camera with progressive sensor.

 

Having said this, the Axis is doing a nice job there.

[Sawbones 0]

I didn't intend for my last post to sound like I'm pimping for Axis or anything... Those pictures were both taken today, they're both MJPEG, and at similar resolution (I think the Axis is 704x480, and the DM is 720x480) but the differences in picture quality between the Axis image and the Dedicated Micros DVR image are pretty stark.

 

Just as an aside... the information in this thread has been very valuable, and thank you to everyone who has been adding to it.

 

Amir? Your posts are absolutely stellar. Thank you very much for taking the time.

[Monitor Your Assets 0]

Just to add a second perspective, here's another screen grab from the identical scene, and the identical camera. This time, however, I took a screen grab with the camera detached from my embedded DVR and attached to an Axis 241q. Bit of a difference, wouldn't you say?

 

Curious - will you be using the Axis unit going forward?

[Sawbones 0]

Just to add a second perspective, here's another screen grab from the identical scene, and the identical camera. This time, however, I took a screen grab with the camera detached from my embedded DVR and attached to an Axis 241q. Bit of a difference, wouldn't you say?

 

Curious - will you be using the Axis unit going forward?

 

Probably so. I've never been completely satisfied with the picture quality my DS2 provided... I've kept it because I practically stole it from the guy who was selling it, it's convenient, provides most of the other features I want, and records to a standard hard drive (that I successfully upgraded to a Seagate SV35 series after the factory one failed). Unfortunately, the picture quality often disappoints, and running motion detection on more than 2-3 cameras slows the DVR to a crawl.

 

It also has some niggling faults that, in my opinion, simply shouldn't exist at the price point in which those units sell. For instance, it has the ability to send you an email with an image, but doesn't allow you to use a mail server that requires authentication. In essence, you have to run an open mail server on your network. Being a bit of a computer security geek, I absolutely REFUSE to do that. I'm struck by the irony of a security DVR expecting you to place such a large security hole on your network.

 

I'd been thinking for some time about going over to a completely IP-based system in the house. I already have a managed gigabit switch supplying the home network backbone, about 4.5TB of network storage on a heterogenous network, and a hand-tuned firewall/caching proxy that keeps script kiddies and other neer-do-wells out. It's all neatly rack-mounted in a locked, ventilated rack cabinet next to my alarm hardware (neatness counts). Putting my existing analog cameras (a mixture of Pano domes, Honeywell domes, etc) on an axis server or two and swapping in a couple of megapixel IP cams seemed like a natural fit. I access my camera views over the network 99% of the time anyway, and only run the monitor output to a TV in the master bedroom via an S-video-to-Cat5 balun. If something goes "bump" in the night, I can simply turn on the TV to see what's up.

 

I'm still thinking about a Mobotix over the driveway... that VGA image is a lot sharper than I thought it would be.

[cglaeser 0]

The Axis 223M megapixel day/night security camera has a minimum illumination of 0.2 lux at f/1.4 listed in the technical specifications. However, the technical specifications fail to mention the shutter speed (exposure time) used to measure that minimum lux value.

 

Using a photography light meter under a street light at night near my house, I measured a light value of ISO 3200, 1 second, f/8. Using a table I located on the net, this corresponds to EV 1 or 5 lux.

 

Objects under the street light look very good on the 223M if the shutter speed is set to 1/2 second, and look poor and grainy at 1/6 second (the camera does not have any settings between 1/2 and 1/6). A light value of 0.2 lux is between 4 and 5 stops less light than 5 lux (reducing light by one stop reduces the lux value by half). This means that Axis exposure time must be increased 4 to 5 stops to compensate for the reduced lux value (my lens is zoomed about half way, so that could account for one full stop there). Although the Axis brochure does not mention the exposure time required for the 0.2 lux rating, this analysis, if done correctly (please check my math) means Axis used the longest supported exposure time of 2 seconds to measure that 0.2 lux value.

 

In other words, a stationary building and a parked car look great on the Axis 223M at 0.2 lux. However, if you want to recognize people who are walking and otherwise moving at normal speed, 1/25 is pretty much a minimum requirement, and 1/60 would be considerably better. Of course, using a shutter speed of 1/60 requires considerably more light than 0.2 lux.

 

These values seem very far from the product specs for the Mobotix M12D which claim 0.1 lux at 1/60 (is this for both the IT and SEC? It's not clear from the brochure). The Mobotix has both a larger sensor and a sensor specifically designed for B/W low light use, so that could account for the large difference.

 

0.1 lux at 1/60. That seems pretty remarkable. Does anyone have access to a Mobotix SEC M12D and a light meter?

 

Comments? Please check my analysis and math.

 

Best,

Christopher

[Monitor Your Assets 0]

These values seem very far from the product specs for the Mobotix M12D which claim 0.1 lux at 1/60 (is this for both the IT and SEC? It's not clear from the brochure).

 

0.1 lux at 1/60. That seems pretty remarkable. Does anyone have access to a Mobotix SEC M12D and a light meter?

 

Hi Christopher,

 

The 0.1 lux at 1/60 is applicable for any camera with a night lens, for the M12 series, at least. I don't have access to a M12D-Sec, but here is an image that may help you.

 

I have added some variables to each image - at the top, you'll find the shutter speed used and the lux used at the time. In this sample, only street lights and the car's headlights were present - my house lights were off. In experimenting (remotely), 1/60 (used here) seems to be the sweet spot as it delivers a non-blurred image with acceptable noise. I'm going to experiment with the settings a bit more with targets walking around the porch (video motion box 2,3) because that is what matters most to me.

 

With sunny skies right now, the day lens reads 4686 lux and it's using a shutter speed of 1/477.

[Monitor Your Assets 0]

Forgot to mention, the following exposure times are available, max & min, per lens:

 

1/8000, 1/4000, 1/2000, 1/1000, 1/500, 1/250, 1/90, 1/60, 1/30, 1/30, 1/5, 1/3, 1/2, 1/1

[Sawbones 0]

I wanted to add that I have an NUUO 4-port H.264 card on order. I'll try to put up a picture of the same scene to go along with the Axis and DM images I've already posted.

 

You can never have too many images. Sometimes that's the best way to make a decision on what you really need.

[amirm 0]

Amir? Your posts are absolutely stellar. Thank you very much for taking the time.

Thanks for the kind words Sawbones.

 

While I am posting this, here is a pop quiz: what are those horizontal lines on the black and white picture Marc posted just above?