[Wingincamera]

I am still on the fence for getting the K-S2 to replace my K-5 II. The main thing I like about the K-S2 is the tilt/swivel vari-angle LCD screen. But I am wondering if image quality suffers with it's 12 bit raw files vs the 14 bit K-5 or K-3 bodies? Just how much difference does the added two bits for the raw files make?

---------- Post added 07-29-15 at 01:51 PM ----------

After posting this, I see there is a number of posts about this question. I have more reading to do.

[Rondec]

I am still on the fence for getting the K-S2 to replace my K-5 II. The main thing I like about the K-S2 is the tilt/swivel vari-angle LCD screen. But I am wondering if image quality suffers with it's 12 bit raw files vs the 14 bit K-5 or K-3 bodies? Just how much difference does the added two bits for the raw files make?

---------- Post added 07-29-15 at 01:51 PM ----------

After posting this, I see there is a number of posts about this question. I have more reading to do.

I don't know how the K-S2 relates to the K5 II. The K5 II has the 14 bit and the K-01 (and K30) have 12 bit and the biggest difference that I can see was that the K5 II has better dynamic range at low iso. On the other hand, the K3 kept the same 14 bit processing, but lost the iso 80 and doesn't have as good dynamic range.

I think it does do some good with regard to dynamic range and color depth, but I would choose a camera purely based on that.

[mcgregni]

If you shoot RAW and need plenty of dynamic range, I think the extra bit depth is worthwhile. It doesn't translate into instantly visible results for all shots once conversion to JPEG has occurred, but I'd expect to find the 14bit files a little more 'pliable' in tonal and colour terms with that bit of extra 'latitude' ...for example on the histogram they'll give you a bit more before the graph starts to break up.

There is supposedly a DR increase .... I forget exactly what the scientific results are, something like 1 or 1.5 ev ..... (I think !)

[Wingincamera]

Thank you all for the replies. I guess it comes down to how much do I want the tilt/swivel LCD screen vs the higher bit raw file. Maybe I will just wait and save a little more money so I can keep my existing K-5 II and add the K-S2 to my kit.

[acoufap]

Using 8 bit you get 256 discrete steps (like jpeg RGB channels), 12 bit = 4096 steps, 14 bit = 16384 steps that divide the "brightness range" of the sensor (cell). So we don't talk about a little bit of difference. It's four times more steps.

In my opinion lesser bits doesn't mean lower dynamic range but more grainy structure. Dynamic range to me is a general property of the physical sensor cells. (But maybe I'm wrong.)

So with lesser bits the in camera raw converter or the raw converter software on your computer gets a more "grainy" brightness structure to work with. In the end the more bits you have, the more details should be possible to work out in an image. Theoretically an important advantage when processing the critical dark and highlight areas.

Depending on the scenery you capture this may lead to visible differences between 12 and 14 bit in the resulting RGB images. But I think there are a lot of other important influences like the capability of the raw converter you use and your experience in post processing.

[The Squirrel Mafia]

^ That makes sense. One thing I have noticed is that starting at ISO 6400, there is a noticeable difference in the fine details between the K-50 (12 bit) & the K-5IIs (14 bit). It looks like the K-50 RAW is compressed, while the K-5IIs RAW is not.

Jumping to ISO 12800, the difference is even bigger & you really start noticing it in the noise grain (luminance) & the fine details. The K-50 RAW looks more compressed at this point & the colors start smearing slightly. The K-5IIs still looks pretty good, but obviously with more grain.

If you grab some RAW files from Imaging Resource & play with them for a bit, you'll see what I'm talking about. The same thing is probably going to apply to the K-S2.

[clackers]

.

In my opinion lesser bits doesn't mean lower dynamic range but more grainy structure.

I think you're confusing resolution - a function of pixel quantity - with dynamic range, Acoufap.

That's the number of possible shades, and is what the bit range stores.

[Na Horuk]

I think if you compare K-5 and K-30 (same sensor, right? Just different bit depth recorded), the K-5 has better DR, according to tests. Maybe there are some other changes to the sensor, I don't know, but it seems that the 14bits do add something.
And there is one more thing - if you are editing a photo, with 14 bits you will get less posterization. This might be an indirect way of helping apparent DR.

[acoufap]

I think you're confusing resolution - a function of pixel quantity - with dynamic range, Acoufap.

That's the number of possible shades, and is what the bit range stores.

This was only an aside not my main point.

But since there seems to be some misinterpretation I'll go a bit deeper. I'm talking about the "time" where the rays doesn't even know that they will be part of pixels of an image .

In the first place I'm talking about the capability of a sensor cell (not a pixel!) to "recognize" light. That means the cell works within an voltage range and recognizing light means the cell produces a voltage. The brighter the light, the higher the voltage. And there is a maximum. So we have a lower boundary and an upper boundary. On this physical level everything depends on sensor design and used materials. If a signal is produced by a cell the next step is that it is digitized by functionality of the sensor (part of the sensor design).

To come back to the starting point of this thread. The digitization is done with 12 (K-30, K-S2) or 14 (K5, K3) bits. The lower digital boundary should be the value zero the upper the maximum of the 12 or 14 bit number. The range in between is devided into steps as I described in my first post. This can be done equidistant in linear steps or logarithmic steps. The latter is my guess.

Resume one cell is able to show (recognize, measure) the hole range of brightness (= dynamic range) of the sensor take all the cells of the sensor together and you get the brightness structure the sensor delivers before taking the bayer filter pattern into account. This is what I'm talking about in my first post.

Bits of an image and the (color) image itself are the product of interpreting the measurements of light after applying the bayer pattern. Interpretation is done using the in camera raw converter or a software raw converter. The interpretation is based on some parameters you set in camera or in the software raw converter.

After this conversion we can talk about resolution of an RGB image. And we can show the spreading of brightness of an image within a histogram. This is usually done on an jpeg image and basis are eight bit per color channel. At this point another story could start

That's my notion of how the things work for bayer pattern based sensors. It helps me to explain a lot of effects in digital photography.

I'm not talking about resolution or pixel quantity!

Dynamic range I see as an inherent property of the single sensor cell and the whole sensor design.

[Adam]

I am still on the fence for getting the K-S2 to replace my K-5 II. The main thing I like about the K-S2 is the tilt/swivel vari-angle LCD screen. But I am wondering if image quality suffers with it's 12 bit raw files vs the 14 bit K-5 or K-3 bodies? Just how much difference does the added two bits for the raw files make?

---------- Post added 07-29-15 at 01:51 PM ----------

After posting this, I see there is a number of posts about this question. I have more reading to do.

If you're already shooting with the K-5 II I'd recommend staying with the high-end body line and upgrading to the K-3. The K-S2 is great but you might miss the K-5's ergonomics.

[Wingincamera]

Thank you all, so much information to digest and think about. For me it's about the image quality of 14 bit or the tilting screen. One thing about the K-3 series I wondered about is the file size and my current computer status. I use LR & PS and sometimes get into panoramas and might have to upgrade laptop. I think I will download some K-3 raw files and play with them.