[Class A]

In an ideal world, having 14 bit raw files would give you more information and thus more wiggle room for PP. in the real world, the information in those last 2 bits is competing against the noise in the system.

In principle you are right, but Pentax's decision to implement ISO 80 and support 14-bit with the K-5, got them the DxOMark throne for APS-C sensor performance.

It may not be much of an advantage IQ-wise but the top spot at DxOMark counts for something too, especially if you have are playing catch up with Canikon in terms of sales.

[madbrain]

Edit: oh, and Magic lantern does focus peak on it already. Though I have not tried that feature yet on my T3i. Guess it's time.
Unified/UserGuide - Magic Lantern Firmware Wiki

I finally tried the focus peaking with Magic lantern. I tried it with a Pentax-F 50mm f1.7 lens mounted with a PK-EOS adapter on the T3i. I tried with the lens wide open.

Turns out the focus peaking is really imprecise. I cannot rely on it to get any sharp pictures. I constantly get blurry pictures of what I thought was in focus.
When magnifying the picture on the LCD, the focus peaking does not show.

Manual focusing with live view and magnification is *much* more reliable.

IMO, this is really more of a gimmick and not a useful feature at all, at least the way it's currently implemented in Magic lantern.
Maybe when the manufacturers themselves add it, they will have better algorithms that work more reliably.

[isaacc7]

In principle you are right, but Pentax's decision to implement ISO 80 and support 14-bit with the K-5, got them the DxOMark throne for APS-C sensor performance.

It may not be much of an advantage IQ-wise but the top spot at DxOMark counts for something too, especially if you have are playing catch up with Canikon in terms of sales.

Well sure, but that was the K-5, a much more expensive camera when it was released. My comments were about the K30. It isn't clear if the 14 bit raw contributed to the score, but in any case we should wait and see what Pentax does with their new flagship camera.

[bwDraco]

Then why would even the Canon EOS Rebel T3 (1100D)--Canon's least expensive production DSLR--output 14-bit raw files?

--DragonLord

[Pentaxor]

Then why would even the Canon EOS Rebel T3 (1100D)--Canon's least expensive production DSLR--output 14-bit raw files?

--DragonLord

marketing. and I don't believe that the T3 is better than the K-01.

[madbrain]

Then why would even the Canon EOS Rebel T3 (1100D)--Canon's least expensive production DSLR--output 14-bit raw files?

--DragonLord

Because for whatever reason, Canon decided not to cripple their camera processing chain after they put a 14 bit sensor, and Pentax did.

If everything else was equal, then 14 bit of course would be better than 12, but other things are not equal.

Pics from my Canon T3i at 18MP and 14-bit are not noticeably better than from my K-r at 12MP and 12 bit.
Actually the high ISO on the K-r is better. The SR also makes a lot of difference. My 800mm mirror lens is hand holdable on the K-r with SR during daytime, but completely useless on the T3i without tripod. The bitness really is way down the list of things that matter for picture quality.

Julien

[isaacc7]

Then why would even the Canon EOS Rebel T3 (1100D)--Canon's least expensive production DSLR--output 14-bit raw files?

--DragonLord

As others have said, it's most likely a marketing decision. Considering that it isn't clear that the T3 actually makes better files, the bit depth probably isn't all that important in the comparison.

Thinking out loud, I can imagine a situation where 12 or 14 bit raws could be a trade off of slightly better dynamic range under some conditions vs. more noise or possibly less accurate colors in some other conditions. I'm not an engineer, but I would think that higher bit depths could possibly have a benefit when dealing with lots of light. Any residual noise could be swamped by a strong enough signal coming off the chip. On the other hand, truncating the last two bits in a low light situation could avoid some of the effects of noise when there is a weak signal.

At any rate, those two bits have shown to be really marginal in any case. The point is that there are potential advantages going either way but having a good 12 bit output is always going to be easier, and therefore less expensive, then a good 14 bit output. When designing mass market electronics, a more consistent good is usually a better compromise than a potential better in some situations and worse in others. I don't know what the actual decision making was of course, but all of this makes sense to me, and a the price of the k30 I'm not going to gnash my teeth over a comparison to the k5.

[isaacc7]

The bottom line in the 12 vs. 14 bit raw decision is that it's complicated. Sampling theory, the effects of noise on digital signal processing, and issue of what things should look like like leave a lot of room for judgement calls to be made. There will always be compromises, and more of them in a mid level product. The good news is that it seems that the bit depth of the raw files are rather low on the importance of the general quality of the files made. Let's wait and see to see what the files look like. If they aren't good enough, then we can speculate about the decisions they made.

[creampuff]

isaacc7, sorry but I can't for the life of me understand what the bezeesus you're talking about? Especially when you go on and on about supposed issues about noise. Theoretical ramblings don't add clarity especially when not backed up by evidence.

I've got a K-5 and it delivers clean images up to a very high ISO with excellent dynamic range. The pre-production K-30 I tried gave very nice images but as it is a pre-production copy with not the final firmware it would be silly to make hasty conclusions on image quality much less speculate about something like the advantages of 14 bits vs 12 bits. Suffice to say the K-30 comes across as an amalgamation of some of the best bits of the current K-5, K-01 and K-r at an affordable price point. For the majority of users this camera is targeted to it is more than capable of meeting their photographic needs. To me it's like a slightly down-spec K-5 with colors in a more affordable package... what's there not to like?

[madbrain]

isaacc7, sorry but I can't for the life of me understand what the bezeesus you're talking about? Especially when you go on and on about supposed issues about noise. Theoretical ramblings don't add clarity especially when not backed up by evidence.

Agree there. It's just pure speculation. The argument that the lowest 2 bits on 14 bit cameras are just noise doesn't really hold water also, IMO. Who is to say the lowest 2 bits on 12 bit cameras aren't also mostly noise that should be cut off ? I think we can look at available test data on existing cameras such as DXomark

To refer just to the 2 cameras I own currently :

The K-r, a 12-bit camera, has 22.9 of effective color depth, vs the theoretical 36 bits (3x12).

DxOMark - Pentax K-r

The T3i, a 14-bit camera, has 22.1 bits of effective color depth, vs the theoretical 42 bits (3x14).

DxOMark - Canon EOS Rebel T3i, EOS 600D

So here we have a 12-bit camera with more effective dynamic range than a 14-bit camera.


For good measure, here are the K-5 results :

DxOMark - Pentax K-5

Effective color depth is 23.7 bits for the K-5 vs the theoretical 42 (3x14) bits .

DxOMark - Pentax K 01

Surprise - the "12-bit" K-01 comes in at 23.7 bits effective color depth, vs the theoretical 36 (3x12) bits.

Sure, I will look at the K-30 results when it comes out, but I don't think it will matter that much.
I'm much more interested in the ISO performance for one thing.

[slip]

Yes, K-30 looks nicer to me:

that is one HOT camera!

[isaacc7]

isaacc7, sorry but I can't for the life of me understand what the bezeesus you're talking about? Especially when you go on and on about supposed issues about noise. Theoretical ramblings don't add clarity especially when not backed up by evidence.

I've got a K-5 and it delivers clean images up to a very high ISO with excellent dynamic range. The pre-production K-30 I tried gave very nice images but as it is a pre-production copy with not the final firmware it would be silly to make hasty conclusions on image quality much less speculate about something like the advantages of 14 bits vs 12 bits. Suffice to say the K-30 comes across as an amalgamation of some of the best bits of the current K-5, K-01 and K-r at an affordable price point. For the majority of users this camera is targeted to it is more than capable of meeting their photographic needs. To me it's like a slightly down-spec K-5 with colors in a more affordable package... what's there not to like?

So sorry I didn't cite references... I have more experience with this when it comes to audio, but I assure you that when it comes to sampling, S/N ratio and noise considerations are the same no matter what medium you care to look into. If you want a good primer on these things in photography, you can find it here:

Noise, Dynamic Range and Bit Depth in digital SLRs

This is the discussion about Noise, Dynamic Range and Bit Depth.

Here's the summation of all the math and tests:

"Curiously, most 14-bit cameras on the market (as of this writing) do not merit 14-bit recording. The noise is more than four levels in 14-bit units on the Nikon D3/D300, Canon 1D3/1Ds3 and 40D. The additional two bits are randomly fluctuating, since the levels are randomly fluctuating by +/- four levels or more. Twelve bits are perfectly adequate to record the image data without any loss of image quality, for any of these cameras (though the D3 comes quite close to warranting a 13th bit)."

This is all bog standard stuff for people involved in making things digitally. Sorry I had assumed that people were familiar with things like signal to noise ratios and sampling. No need to get uptight about not understanding how things work, asking a few questions does wonders... My comments were trying to address the wailing and gnashing of teeth about how Pentax was "crippling" their cameras and that there wasn't a good reason to go to 12 bit raw files. I offered up some points as to why they might have done it, and I mused on some ramifications. We don't know the real reasons behind Pentax's decisions, but there's every chance that it is a solid engineering decision.

[isaacc7]

Agree there. It's just pure speculation. The argument that the lowest 2 bits on 14 bit cameras are just noise doesn't really hold water also, IMO. Who is to say the lowest 2 bits on 12 bit cameras aren't also mostly noise that should be cut off ?

I never said they were all noise, just that noise could be the reason for limiting the bit depth. Limiting noise is expensive, and there are other tradeoffs when it comes to larger files sizes, mostly involving speed but power of processor also comes into it. The more powerful the processor, the more heat is likely to be produced, which can lead to more noise, etc. Take a look at the links I posted before, my musings are hardly controversial. I was engaging in speculation, I was trying to calm some of the more worried folks about 12 bit raw files. Read those and you'll see that there can be some good reasons to limit the bit depth to 12 bits, and noise is primary one.

[madbrain]

So sorry I didn't cite references... I have more experience with this when it comes to audio, but I assure you that when it comes to sampling, S/N ratio and noise considerations are the same no matter what medium you care to look into. If you want a good primer on these things in photography, you can find it here:

Noise, Dynamic Range and Bit Depth in digital SLRs

This is the discussion about Noise, Dynamic Range and Bit Depth.

Here's the summation of all the math and tests:

"Curiously, most 14-bit cameras on the market (as of this writing) do not merit 14-bit recording.

These links provide good background information on noise, and good information about some 4-year old Canon and Nikon cameras.
I'm just not sure how much of it applies to the current crop of Canon, and released and unreleased Pentax cameras

[Pentaxor]

I never said they were all noise, just that noise could be the reason for limiting the bit depth. Limiting noise is expensive, and there are other tradeoffs when it comes to larger files sizes, mostly involving speed but power of processor also comes into it. The more powerful the processor, the more heat is likely to be produced, which can lead to more noise, etc. Take a look at the links I posted before, my musings are hardly controversial. I was engaging in speculation, I was trying to calm some of the more worried folks about 12 bit raw files. Read those and you'll see that there can be some good reasons to limit the bit depth to 12 bits, and noise is primary one.

so basically what you are saying that it depends on the hardware and the conversion that handles the bit rate quality. let's say that a newer hardware that handles the 12bit data would be better than the 14bit data converted by the current hardware on the K5. of course, 14bit would be better significantly, provided that it has the hardware that would optimize it's use. it's like optimizing the capability or usage.