Member Discussion

What MP Camera Has 120db WDR Or Higher?

i'm working on a project and they need a Mega pixel camera with 120db WDR. i tired to use the finder but we can't select the WDR level in the finder any advise which camera can support this?

The Samsung Wisenet 2MP camera claims 120db WDR, as do the Hikvision Pro Series.

Panasonic has a number that specify 133 db, see here for an example.

We don't support searching by db for WDR because it's highly inaccurate and manufacturers can make it up (like minimum illumination for low light).

Samsung and Hikvison both rate some of their cameras for 120dB WDR. Vivotek rates some of theirs at 140dB. The newest Panasonic includes cameras at 133dB. Axis 'true' WDR Q16 cameras say 'up to 120dB'. Sony Gen 6 true WDR cameras say 'up to 130dB'.

If this is a bid spec, you can look at that but I really think you should push back against this as may unfairly knock out manufacturers that are good in low light but conservative in specs.

...manufacturers can make it up (like minimum illumination for low light).

Although related to minimum illumination levels, I believe dynamic range may be slightly more empirically grounded. The main problem with the low-lux spec, as you've lectured eloquently in the past, is that there is no accepted standard for when an image is deemed usable. Is it when the first signal level (over the noise floor) is detected or when some vague details can be discerned? Details according to whom? Obviously there is a lot of wiggle room there.

But dynamic range measurement is different as there is no subjective perception of image quality involved. The proof is in the pudding. For if a camera has a dynamic range of 120 dB, then one should be able to take, or the manufacturer should be able to supply a jpg which has a dynamic range of 120. Meaning that the range of continuous intensity values in the jpg should span that range without any missing gaps. This can be viewed with a standard histogram tool. Is their wiggle room, sure like what chart to use for instance. And of course manufacturers are know for fudging. But the more verifiable a spec the less they seem to stretch.

But taking out the subjective element of 'am I seeing a usable image' I think makes the spec about as trustable as SNR. Which I'm not sure how much you trust either... Do you agree even a little or do think that lux and DR dB are equally fictictious?

And how have you verified how manufacturers are rating / measuring their camera's dB?

My point is simple, since each manufacturer can decide on their own measurement process for WDR / dB, there's no way to know how a 120dB from one relates to a 120dB for another.

Or else do you really believe that Vivotek cameras max WDR is 140dB vs just 120dB for Sony's best?

Or else do you really believe that Vivotek cameras max WDR is 140dB vs just 120dB for Sony's best?

1. Sony's best is at least 130dB, you just said so in your other post above.

2. Vivotek is claiming 140db because they use 4 seperate exposures as opposed to 2 for traditional trueWDR.

Do you not believe they are doing this? Or do you believe that this technology cannot provide any real enhancement even if they are?

In any event my point is that this dB claim is verifiable by standard engineering tests for dynamic range. Namely the difference in intensities between the darkest and lightest pixels, without intensity gaps, of an image MUST be 140dB. Are there different accepted methods of insuring that there are no 'gaps in the intensities', that could give slightly differing results, yes, by maybe a few dB. But this is nothing like the fudge around lux.

IMHO, You are better off comparing dynamic ranges than to comparing lux. Even if Vivotek is stretching the interpretation of their test, 140 can be roughly compared to Sony's 130. But how do you compare 1 to .0000001? At all.

If you believe that dynamic range and min. illumination are equally fictictious and un-grounded, can you explain why we see differing values from 10 lux down to (last time I checked) .0000001 lux? Varying by factor of a million? And every day another zero.

Yet dynamic range can't even bust 200. Why not? Because if it did we would say bullshit, and if the Vivotek camera can't produce an image with 140db of difference in it, we should say bullshit now, just like we would if they had said it had a resolution of 10 gigapixels.

Vendors need to be held accountable for their dynamic range specification.

Sony new WDR cameras do 4 exposures as well.

"If you believe that dynamic range and min. illumination are equally fictictious"

Strawman argument. I am not saying they are equally fictious, lux rating are worse but neither are worth using.

On the Vivotek spec, take a look at this presentation (via google) which shows them in competition against Sony @ 140dB verses 130dB. Includes night time comparsion shots.
Vivotek use a Sony sensor in this camera, assume Sony do too :)


Can't believe the figure would be artifically inflated/measured for marketing/sales purposes.

I only see a single Sony vs Vivotek comparison shot, that's for a fairly well-lit street scene between the Sony SNC-VM601 and VIVOTEK FD8355EHV, copied below:

This is not convincing. The images are taken with different FoVs and at different times. Also, we have no idea what settings are used (i.e., if Vivotek optimized theirs in any way, etc.).

Also, this is not a true WDR scene and the Vivotek is being pointed out a window (notice the open window at the top of the Vivotek image (which is weird).

Is this the scene you are alluding to?

Yes. Only comparision shot in the document. It's a product briefing doc to show why Vivotek's solution is better. My take home is that they are on an equal footing given the image. ie: nothing earth shattering. It's also pretty poor for comparision purposes for the reasons you listed. The reason I mentioned that image was becasue of the lack of a comparison made when a WDR example is shown, yet the presentation highlights that the Vivotek solution is 140dB verses Sony's 130dB - so why no WDR comparision shot given the document's purpose?

My point is that even in the marketing literature, it's a stated number. There is no supporting inforamtion, yet it's a claim that could make or break many a tender because one of them "goes to 11".

An opportunity for you to clear this up on IPVM by testing many of the above listed cameras based on a standard definition and normalising all of the claims?

There are many high level country wide specifications which ask for specs just like 120dB+ without any validation at all of what that means.
For many bids it’s just a case of having the right info on the datasheet and the real world results are not important to the specifier.

There is no argument that will win over this logic. The spec is set. So expect the WDR figures to rise in order to meet these requirements.

Ethan and I have talked about doing true WDR greyscale testing, like so (snapped from an old Pixim doc):

We'll move this up our testing priorities list. This would be the most direct / quantifiable way to test though we need to be careful to see how this type of testing actually measures up to real world performance.

You've convinced me that DR values can and do get fudged enough by manufacturers to make them unreliable. But I do think that it's possible for a third party (IPVM) to arrive at a meaningful Dynamic Range estimate result based upon real-world tests, (unlike low lux where it is probably hopeless).

DpPreview created their own dynamic range test (for photographic still cameras) a while back for similar reasons.

Should be fun to see how the claims match-up..

Thanks for that link. We'll take a look at that.

Btw, we did an independent low light quantitative test last year, measuring TVL, but TVL alone does not fully capture low light performance.

This market is rife with must-meet project specs and without test verification, lower-quality cameras focused on meeting paper specs at lowest cost can win - hurting security and the industry overall.

For example, usable dynamic range is often limited by lens flaring and a theoretical number based on how many exposures are used (which is what some manufacturers use) is misleading, we all know that.
A 100dB spec camera with an appropriate lens, motion de-ghosting, and sophisticated tone mapping may outperform one specified to 120dB. Further, line/pixel-interleaved exposure perform better than frame-based due to less motion artifacts and a larger shutter window for low light.

The wide dynamic range dB number seems to be as valid as the low light lux number... IPVM can do some good here with a standardized dynamic range test.

I and the R&D team I was with at the time (2013-2014) tested and evaluated EVERY manufacturer claiming high Db for WDR.

Sony's 4-exposure Xarina provided the best overall combination of High Dynamic Range, Low-Light performance and Color accuracy, though it's better to move it to 2-Exposure in-low light to allow for longer exposure time.

I never once found Vivotek's claim of 140Db to provide actual better WDR than Sony Xarina. Vivotek's WDR is certainly excellent and scored top marks, but it's not the best WDR available.

Keep in mind that Vivotek DOES NOT develop its own Imagers, though they do develop some of their own Encoders/DSP silicon, whereas Sony Xarina is a combined, matched imager/DSP set. I'm not rah-rahing for Sony here, I'm just sharing with you the elevator speech of a long set of testing I did in 2013 and 2014 to evaluate actual relative WDR performance.

John's point is 100% correct, there is no way today to evaluate a WDR Db Spec relative to other manufacturers, because there is no universally accepted way to test WDR. Customers who make their buying decisions about WDR solely based on the Db rating on the Spec sheet would be well-advised to NOT do that, and TO ACTUALLY DO comparison video testing before they make their final buying decision. Particularly if WDR is the primary need of their application.

Though, for Sony and Vivotek at least, you seem to be saying that their stated DR numbers are at least generally indicative of top-notch WDR, no?

Who were the worst offenders in terms of specification vs. reality in your tests?

Also, can you sketch out what your methods were?

No I'm not saying Db numbers are generally indicative of top-notch WDR at all. I've seen plenty of no-name or unknown of manufacturers claiming 120Db with a product that couldn't possibly provide that high of Dynamic Range. What I'm saying is that you need to dive deeper into the technology under the hood, and ultimately do a shoot out. YOU SHOULD NOT ASSUME THE SPEC SHEET WITH THE HIGHER WDR DB RATING IS THE BETTER CAMERA.

I cannot sketch out methods, because to do so would violate confidentiality for my previous employer.

I also don't want to cast stones on the worst offenders. But generally speaking, if it's a company that's using Digital WDR and claiming a high Db, it should be suspicious. If the manufacturer cannot accurately explain to you the interaction between the imager and the DSP in creating solid WDR, you should be suspicious.

Thanks for the info!

My comment was in regard to the Sony and Vivotek models only, which you rated best and top notch/excellent respectively.

I asked for the worst offenders to get an idea if the inconsistency of the spec was due to abuse by the usual suspects of the 000001 low-lux gang, or if the discrepancy was mainly due the lack of a 'universally accepted' test.

Also, not only do I not assume that the "spec sheet with the higher WDR db is the better camera", (de-emphasis added), but that I do not even assume that the camera that actually has better WDR is the better camera.

Lastly, was your impression that the spec overreach was more a reaction by the digital WDR manufacturers trying to stay close to those of true WDR, or true WDR trying to distance itself from DWDR?

Thank you for your clarifications.

I think the spec overreach is predominantly done by those who are using true WDR to differentiate themselves from everyone else using similar technology, and with the knowledge that verifying the accuracy of the spec would be difficult or impossible (just like it is with low-lux ratings.)

People are spec-driven by nature. The same is true in the Consumer Electronics World, where TVs are selected (assuming two competing models are similarly priced) with things like brightness, contrast ratio and refresh rate.

Many times (and this is absolutely the case with camera specs like temperature rating and environmental IP rating), a customer will choose the highest spec without even needing that spec for their application, and will simply assume the higher spec'd product is "better". I see many jobs in the desert that demand IP67 even though it's the desert and the product will never be submerged in water, and an IP65 or IP66 product is more than adequate for their application. They'll likewise choose a product that has a 75°C rating over one that has a 70°C rating even though the hottest it will get in the Middle East is 54°C. Because the 75°C product has to be better, right?

Unscrupulous manufacturers know of this human condition and will spec their products accordingly.