Low Light Shooters

[Gavjenks]

DxO Mark is okay if you're careful with how you interpret it.

Their "Sports: Low light" score happens to be one of the stupidest metrics they use, and I highly recommend you ignore it. The definition is, "Low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits."

They provide absolutely no justification for why they chose 30db, 9 EVs, OR 18 bits of color depth. Nor any explanation why dynamic range OR color depth should even have anything to do with this at all. Fundamentally, low light noise is a function of SNR (signal to noise ratio) only. In other words, the score is 3 arbitrary numbers, 2 of which aren't even needed, stacked on top of each other and is thus pretty much meaningless.


However, a much better metric is also available, also from DxO Mark: Just SNR by itself is available to look at. You can find it by doing this:
1) Choose your comparison cameras and hit compare.
2) Just above the pictures of the cameras, there's a bar that has the options "Scores, specs, measurements, lenses tested" Choose "measurements"
3) Then choose "SNR 18%" This will give you a much better comparison between the cameras, at each ISO value (the whole curve matters, and sometimes a camera might have better low light performance at ISO 200 and the other has better at ISO 800, etc.)

 

[apaflo]

I understand the difference between FX and DX, but in terms of performance, what's the difference between the D600 and D610?

Similarly, what's the difference between the D300S, D7100 and D7000?

These web pages don't have data for every camera, but it is there for the most interesting ones. Also take a look at the graphs for ideal FX and DX sensors.

Photographic Dynamic Range versus ISO
Read Noise in ADUs versus ISO

Not quite as helpful, but certainly interesting:

Sensorgen - digital camera sensor data

 

[apaflo]

Their "Sports: Low light" score happens to be one of the stupidest metrics they use, and I highly recommend you ignore it. The definition is, "Low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits."

They provide absolutely no justification for why they chose 30db, 9 EVs, OR 18 bits of color depth. Nor any explanation why dynamic range OR color depth should even have anything to do with this at all. Fundamentally, low light noise is a function of SNR (signal to noise ratio) only. In other words, the score is 3 arbitrary numbers, 2 of which aren't even needed, stacked on top of each other and is thus pretty much meaningless.

That appears to be a fairly good metric. They are saying the image from the camera should be good enough that the JPEG format and an average monitor limits the quality more than the camera does.

 

[Gavjenks]

That appears to be a fairly good metric. They are saying the image from the camera should be good enough that the JPEG format and an average monitor limits the quality more than the camera does.

That's a nice theory, except:
1) Almost every monitor in fact provides 24 bit color depth (exactly one byte per channel, the jpeg, etc. standard), not 18 (....6 bit per channel???). So if that's what they're doing, why did they choose 18?
2) A crappy LCD monitor may indeed be aroudn 9 EV dynamic range, but if you are using RAW, that doesn't matter, because if you capture a higher dynamic range, you can compress it down in your converter without any clipping. So you can shoot at 11 EVs, and cram them into 9 for the monitor. Thus, it matters which camera has lower noise at ISOs that yield 11 EVs. So why are they doing their measurement only at 9? In fact, even if you're using JPEG, you can choose settings that will automatically compress it down to some extent (in Canon, you can choose slightly negative contrast as a jpeg mode setting, for instance), which would do the same thing, just more crudely.
3) the same logic in #2 also applies to color depth, yielding a second reason why 18 bits of color depth is a nonsense cutoff point.


Even the 30db cutoff doesn't really make much sense / is totally arbitrary as far as I can see. The only reason to use it appears to be so that they can publish one simple number for lazy people to compare together with a subtraction problem. But why would you do that when the entire graph is available? You're just throwing out 90% of your data for no reason. Surely, if you actually care about technical stuff like this at all, clicking a single link and looking at the graph is worth it when deciding how to spend hundreds and hundreds of your dollars on a new camera. And if you don't care about technical stuff, then why are you on DxO Mark at all?

 

[jaomul]

If all cameras shoot reasonably well in good light, but they don't all shoot well in low light, wouldn't it make sense to buy the best low light camera, thereby getting the best of both worlds?

What is your recommendation for a good low light camera?

I would say your own Olympus is well capable in low light, but all your lenses are slow. Add the panny 25mm f1.4 or similar and you should be set

 

[apaflo]

That appears to be a fairly good metric. They are saying the image from the camera should be good enough that the JPEG format and an average monitor limits the quality more than the camera does.

That's a nice theory, except:
1) Almost every monitor in fact provides 24 bit color depth (exactly one byte per channel, the jpeg, etc. standard), not 18 (....6 bit per channel???). So if that's what they're doing, why did they choose 18?

Because most monitors may have an input that is 24 bits, but the actual display is 6 bits per channel.

2) A crappy LCD monitor may indeed be aroudn 9 EV dynamic range, but if you are using RAW, that doesn't matter, because if you capture a higher dynamic range, you can compress it down in your converter without any clipping. So you can shoot at 11 EVs, and cram them into 9 for the monitor. Thus, it matters which camera has lower noise at ISOs that yield 11 EVs. So why are they doing their measurement only at 9? In fact, even if you're using JPEG, you can choose settings that will automatically compress it down to some extent (in Canon, you can choose slightly negative contrast as a jpeg mode setting, for instance), which would do the same thing, just more crudely.

A JPEG formatted RGB image will not be able to encode more than 9 stops of dynamic range.

3) the same logic in #2 also applies to color depth, yielding a second reason why 18 bits of color depth is a nonsense cutoff point.


Even the 30db cutoff doesn't really make much sense / is totally arbitrary as far as I can see. The only reason to use it appears to be so that they can publish one simple number for lazy people to compare together with a subtraction problem. But why would you do that when the entire graph is available? You're just throwing out 90% of your data for no reason. Surely, if you actually care about technical stuff like this at all, clicking a single link and looking at the graph is worth it when deciding how to spend hundreds and hundreds of your dollars on a new camera. And if you don't care about technical stuff, then why are you on DxO Mark at all?

It provides a quick and relatively meaningful tape measure, absent any need to actually understand all the details (which obviously isn't everyone's torte).

The 30 dB SNR has to do with being just less than the dynamic range of 6 bits per channel.

 

[Gavjenks]

Okay, so if you:
1) Market mainly to people with bad monitors (even the iphone displays 24 bits...)
2) Only ever display your photos on the internet, never in print
3) Use JPEG-only mode exclusively
4) Never use any adjustment settings like contrast (including the default one on most cameras!), even though you're shooting JPEG only and rely on settings (#3)
5) Never even so much as adjust the curves of your photos in photoshop

Then yes, that's a great metric for you! I wonder how many people in this thread/forum qualify under those requirments?
... *crickets* ...


It's the equivalent of weathermen nation-wide simply always reporting Manhattan weather, "because that's where the most people are efficiently experiencing the same weather, so for a single shotgun number, it's most likely to be correct!" And because surely nobody is smart enough to look up their own city (understanding details like whether data REMOTELY applies to you isn't everybody's torte after all).

That's about the level of accuracy we are talking here for a randomly chosen photographer looking at this number.

 

[apaflo]

Okay, so if you:
1) Market mainly to people with bad monitors (even the iphone displays 24 bits...)
2) Only ever display your photos on the internet, never in print
3) Use JPEG-only mode exclusively
4) Never use any adjustment settings like contrast (including the default one on most cameras!), even though you're shooting JPEG only and rely on settings (#3)
5) Never even so much as adjust the curves of your photos in photoshop

Then yes, that's a great metric for you! I wonder how many people in this thread/forum qualify under those requirments?
... *crickets* ...


It's the equivalent of weathermen nation-wide simply always reporting Manhattan weather, "because that's where the most people are efficiently experiencing the same weather, so for a single shotgun number, it's most likely to be correct!" And because surely nobody is smart enough to look up their own city (understanding details like whether data REMOTELY applies to you isn't everybody's torte after all).

That's about the level of accuracy we are talking here for a randomly chosen photographer looking at this number.

You are wandering off into absurd personal conclusions about things that in fact are relatively universally valid. For example, most monitors do in fact display 6 bits per channel. For DXOmark to recognize that is not surprising.

The point was that you weren't aware of why they defined it as they did, were unaware of the significance of the values they chose, and felt it was without any merit. The values are not arbitrary, and the measure does have merit beyond what you pointlessly argue.

 

[lennon33x]

The differences between DX and FX bodies are shrinking but FX still comes out slightly ahead of DX with better ISO performance. So if you got the cash, I would go with an FX body and some good glass. The D610 is a very capable camera and can do toe to toe with a D800 on some things. I went with the D800 for several reasons, I love the controls more: the ISO and WB buttons instead of the mode control wheel for example. The larger mp sensor, view finder window is nice as well, LCD protector is nice as well.

www.snapsort.com

 

[Derrel]

DxO Mark is okay if you're careful with how you interpret it.Their "Sports: Low light" score happens to be one of the stupidest metrics they use, and I highly recommend you ignore it. The definition is, "Low-light ISO is the highest ISO setting for a camera that allows it to achieve an SNR of 30dB while keeping a good dynamic range of 9 EVs and a color depth of 18bits."They provide absolutely no justification for why they chose 30db, 9 EVs, OR 18 bits of color depth. Nor any explanation why dynamic range OR color depth should even have anything to do with this at all. Fundamentally, low light noise is a function of SNR (signal to noise ratio) only. In other words, the score is 3 arbitrary numbers, 2 of which aren't even needed, stacked on top of each other and is thus pretty much meaningless.However, a much better metric is also available, also from DxO Mark: Just SNR by itself is available to look at. You can find it by doing this:1) Choose your comparison cameras and hit compare.2) Just above the pictures of the cameras, there's a bar that has the options "Scores, specs, measurements, lenses tested" Choose "measurements"3) Then choose "SNR 18%" This will give you a much better comparison between the cameras, at each ISO value (the whole curve matters, and sometimes a camera might have better low light performance at ISO 200 and the other has better at ISO 800, etc.)

Attempts to discredit DxO Mark's results, ranking, and scores cracks me up. Such statements are almost always given by a Canon user or Canon fanboy type.Why 30db, 9 EVm, and 18 bit color as criteria? Pretty simple: decent signal to noise ratio, adequate dynamic range, and decent color. In other words, the DxO Mark metric for Low Light/High ISO focuses on camera sensors that can deliver low noise; decent dynamic range, and good color images. Pretty. Obvious. Stuff. At least to somebody whose ego is not tied to the fact that "his brand" of cameras has multiple models that consistently score far,far below "the leading brand's models."Apparently, Gavjenks thinks that signal to noise is the ONLY valid criteria. The hell with dynamic range. The hell with color. Right? I mean, who cares if the dynamic range is 4 stops and the color is chit, as long as the noise figures are low, right? LMFAO.Sorry Gavjenks, but until Canon updates is sensor fabrication technology from its decade-old .50 micron process, and moves into the .19 micron era, like Sony and Toshiba has done, Canon cameras will CONTINUE TO SUFFER from poor dynamic range, noisy, God-awful shadow detail issues, and weak color at anything even remotely above base-level ISO. It's pretty simple: DXO Mark's testing procedure rewards the newer, higher-technology, better-performing sensors that Nikon and Pentax are buying from third-part sensor makers, and it penalizes Canon's home-brewed sensors because, for the most part, they are made using outdated sensor fabrication methods.

 

[kathyt]

Buy the best you can afford. I wouldn't go completely pro-grade, (like the X on the Canon side) but I would try a couple out and see what you think. Rent a few and find out which one works for you. I think the low light performance on the Canon 5D Mark iii is great, but I can't compare it to a Nikon because I have never touched one.

 

[Derrel]

DxOMark Camera Sensor

The above article is the newest one from The Luminous Landscape, in which DxO Mark's testing procedures are explained.

What you will find is that larger-sized sensors perform better than smaller sensors within a given generation. As the article mentions, however is that occasionally ,"There are other factors influencing sensor performance. These can be subtle differences between a company’s level of technical expertise, its R&D budget, its patent portfolio or its market strategy. Thus a smaller sensor (e.g. the 1.5× Sony NEX-7, August 2011) sometimes actually matches a larger and newer sensor (e.g. the newer Canon 5D Mark III, March 2012) – thus masking both previous trends. In general these are temporary anomalies because, in the end, the laws of physics and marketing will prevail."

Other highlights from the "horrible,awful, big, bad, mean old scientists at DxO Mark" include data that shows the following conclusions.

"

• The best scoring cameras are the Nikon D800 and D800E twins (highlighted in white). Both are full-frame cameras (blue dots). The D800E scores 1 point higher, although that difference is negligible.
• The 1.5× APS-C market segment is very full. Especially if you include the neighboring 1.6× APS-C models by Canon.
• Some newer 1.5× APS-C products score between 78 and 82 points (just like Canon’s best full-frame models). High-scoring APS-C models currently use sensors supplied by Sony:
  • Sony Alpha-580/NEX-7/Apha-77
  • Pentax K5/K-01
  • Nikon D3200/D5100/D7000
• For APS-C sized sensors, there is no significant performance difference between SLRs, Sony’s technology with a semi-transparent fixed mirror[19] or premium mirrorless models such as the Sony NEX-7[20].
• Except for the 1/1.7" segment, none of the Canon models are currently best-in-class[21]. This includes the recent Canon Powershot G1X and G15, and the EOS 650D, 5D Mark III and 1Dx. Canon is lagging behind particularly in its Dynamic Range at low ISO. Canon’s competitive 1/1.7” models incidentally use Sony sensors.
• Some sensor sizes are performing well given their size while others are lagging behind:
  • Recent 1/2.33” and 1/1.7” sensors are scoring well compared to their size.
  • The Sony RX100 (CX or 1/1” format) is scoring well compared to its size.
  • The (Micro) Four Thirds format has been dormant until recently. The two new well performing models are the Olympus PEN EPL5 (72 points) and the Olympus OM-D E-M5 (71 points).
  • The 1.6× format is struggling as well, but is a Canon-only format. So it reflects the fact that Canon is currently not state-of-the-art in terms of sensor noise.
  • The 1.5x APS-C format is doing well in some models due to the use of Sony sensors and likely due to the extreme competition.
  • The full-frame format is doing very well. The five highest scoring full-frame cameras are incidentally all by Nikon."

 

[Gavjenks]

Attempts to discredit DxO Mark's results, ranking, and scores cracks me up.

Except that I didn't.

I discouraged one DxO (summary) number, and encouraged the use of another, better DxO (full data) number.
I do not disagree that DxO in general is a great place to go for technical data on cameras. I'm simply suggesting that people who have half a head on their shoulders can almost always get much more relevant info for THEIR SITUATION by looking at the raw numbers in a graph, versus just going by the average, rough summary provided by DxO for their data.

Their summary numbers are aimed at newbie photographers who don't use any advanced processing or data storage techniques. The numbers assume you are:
* A jpeg shooter (since they assume you have no way to compress more than 8-9 stops of data into your final image)
* Using default settings (same reason)
* Exposing at the meter's suggestion all the time (such that those 8-9 stops are fully dispersed out in your range in the first place, evenly)
* Sharing all of your work on crappy monitors owned by friends over facebook/instagram/etc. (which are uniquely limited by the thresholds they use)




In other words, DxO's summary numbers assume you are an instagram photographer who doesn't know what they're doing and shoots all day on auto mode.
The raw data, however, can be interpreted with regard to whatever your actual skill level is.
Thus, it is almost certainly more useful for anybody who cares enough to be on this forum in the first place. Because you can say "Well when I shoot, I use XYZ parameters usually, and my equipment can handle XYZ range, so I look up those in the graph and find the data relevant to ME"




Why, Derrel, would you encourage people to take a wonderful resource like DxO Mark, and only pay attention to 10% or less of its data, that they picked out as being of the greatest interest to a hypothetical photographer who it not you? It's like going to a library, but intentionally ONLY looking up books whose authors' names start with "S." It's just mindlessly wasteful of information.

 

[deveaushawn]

I seem to have a knack for posting questions that start "lively" discussion! It's not intentional!!!

As Jaomul suggests, perhaps some faster lenses would allow me to achieve what I'm looking for with my EM-5.

In the MFT world, we have quite a few choices. Since I'm an Oly shooter, I already have IBIS, so I don't think it makes sense to pay 'extra' for a stabilized lens, unless there is a significant boost in IQ.

Given that, I've identified the following lenses as candidates:

Olympic 25mm f/1.8 @ $400CAD

Panasonic 25mm f/1.4 @ $600CAD

OR I could pick up the following 2 lenses:

Sigma 19mm f/2.8 @ $200CAD
Sigma 30mm f/2.8 @ $200CAD

Does anyone know if the Panasonic is worth an extra $200 to go from f/1.8 to f/1.4?

Would I be better off with the 2 Sigmas?

 

[Gavjenks]

I seem to have a knack for posting questions that start "lively" discussion! It's not intentional!!!

As Jaomul suggests, perhaps some faster lenses would allow me to achieve what I'm looking for with my EM-5.

In the MFT world, we have quite a few choices. Since I'm an Oly shooter, I already have IBIS, so I don't think it makes sense to pay 'extra' for a stabilized lens, unless there is a significant boost in IQ.

Given that, I've identified the following lenses as candidates:

Olympic 25mm f/1.8 @ $400CAD

Panasonic 25mm f/1.4 @ $600CAD

OR I could pick up the following 2 lenses:

Sigma 19mm f/2.8 @ $200CAD
Sigma 30mm f/2.8 @ $200CAD

Does anyone know if the Panasonic is worth an extra $200 to go from f/1.8 to f/1.4?

Would I be better off with the 2 Sigmas?

Depends. To shoot what subjects? Even then, depends on your style. 1.8 is significantly brighter and blurrier backgrounded than 2.8 (1 1/3 stop). And 1.4 is less of a difference, but still significantly more so than 1.8 (2/3 stop). So it's simply a matter of how low of light you're shooting in, and/or how blurry you like your backgrounds versus how hard it is to get stuff in focus, etc.