[rawr]

I suspect all these arguments about IQ vs megapixels must sound very familiar to Nikon D7000 > D7100 upgraders (or even D700 to D600/D800 upgraders).

I must visit some Nikon forums to see what the consensus position on the issue is.

[jsherman999]

So I'm wondering what the practical implications are for the modern day, professional digital photographer? And for that matter, anyone who enjoys digital photography? The amount of printing done these days is miniscule compared to the number of digital images produced. Even brides don't print their wedding photos that much anymore. But increasing the magnification of a photo on one's computer is certainly done proportionately higher statistically than printing the photo.

Yes, but you don't magnify to 100% and then paste that 100% crop to Facebook. ** You display the whole image in whatever medium you're using, and this is going to involve downsampling. When this downsampling happens, any more noise you see at the pixel level of the more magnified, higher-res image is compensated for. In other words, there's no IQ downside to more MP, assuming similar pixel QE - you'll either get the same noise performance with equal detail if radically downsized, or the same noise with more detail if slightly downsized. If printed or displayed at native size (really big,) a slight NR to the higher-MP image is going to clean up the noise, equaling it, while (probably) keeping more detail at the same time. Both sharpening and NR are easier to do, IMO, with more megapixels to work with - the results look more real when done sparingly and carefully.

Then there's cropping - something everyone does, to some degree. More MP allows you to crop to square, to 3:2, etc, while maintaining better detail. With a lot of MP you can crop heavily and still have fantastic large facebook images to share, if you're into doing that. (don't knock it, it keeps the extended family happy. ) And there's always the issue of future use for these images... IN 10 years, will we be able to buy a television with less than 4K resolution? That 4K (or more) displayed on a 60'' (or bigger) screen will be more demanding than what we're used to now.

** Even if you did post a 100% crop to facebook, one from the K-5 and one from the K-3, the K-3 crop would be larger dimensionally, like:


(24MP vs 12MP 100% crop, images from diglloyd)

- if you wanted to make those crops the same size, your resizing of the K-3 crop would involve downsampling, or upsampling of the K-5 image, and upsampling isn't going to favor the lower-res image either:

But perhaps the megapixel race is still ahead of the ability to control noise? Which would support panoguy's suggestion that if lower noise photographs (as viewed on a computer screen) are more advantageous to the photographer's work, then perhaps the lower MP camera is a better choice?

That would only be true if the K-3's pixel-level efficiency is actually worse than the K-5's, in which case a downsampling to 16MP would give you worse performance (more noise) than a native K5 image. That's possible, but I don't think we've seen that happen recently with sensors from the same generation but with higher MP. It did happen with the Sony A900 vs. Nikon D700, the higher-MP for that 24mp sensor didn't quite match the D700 even when downsampled to 12MP - the D700 was still cleaner. But since then, that hasn't happened AFAIK.


.

[Winder]

So far 6400 has cleaned up very well. DXO Pro Optic 9 doesn't recognize the K-3 DNG files but LR 4.4 is doing fine with a little adjustment. Topaz DeNoise is really not needed at 6400 unless you are pixel peeping.

[jsherman999]

I suspect all these arguments about IQ vs megapixels must sound very familiar to Nikon D7000 > D7100 upgraders (or even D700 to D600/D800 upgraders).

I must visit some Nikon forums to see what the consensus position on the issue is.

Actually that isn't a bad idea I learned to not fear the MP from the guy who runs sensorgen, from Luke Kaven, Bob Sheehy, and Joseph James (guy who coined 'equivalence') and others on the Nikon FF forums during the run-up to the D800 release a couple years ago. A lot of bedwetting from some D700 owners that more MP would equal more noise, that more MP was a waste, etc. I learned a lot from those threads, and it changed my mind about a few things.

[Tanzer]

So I'm wondering what the practical implications are for the modern day, professional digital photographer? And for that matter, anyone who enjoys digital photography? The amount of printing done these days is miniscule compared to the number of digital images produced. Even brides don't print their wedding photos that much anymore. But increasing the magnification of a photo on one's computer is certainly done proportionately higher statistically than printing the photo. The question of increased noise vs increased megapixels is part of the digital photographic landscape. I think we all hoped/assumed that advances in technology would allow for higher resolution without the compromise of increased noise, and this has been true to some extent. But perhaps the megapixel race is still ahead of the ability to control noise? Which would support panoguy's suggestion that if lower noise photographs (as viewed on a computer screen) are more advantageous to the photographer's work, then perhaps the lower MP camera is a better choice?

No, don't confuse a lower MP count with less noise. You may be able to "coax" more noise from a K-3 image than a K-5 image, but only because the K-3's image has a higher resolution, which allows one to crop so extensively to begin with. For a given scene, without any cropping, or by cropping both images an equal amount, the K-3's images should actually be less noisy, due to the fact that the sensor is a newer generation -- regardless of pixel count.

Many wedding photos and the like, which are delivered electronically, are a scaled JPEG, almost never the raw file, so that will further limit how much the end user can crop.

The biggest downside I can think of with a higher megapixel count is the larger filesizes.

[Class A]

And of course I'm looking at the files at 100%, that's where you see noise or other artefacts!

Image reviewing at 100% is fine, but it is not fair to compare a K-5 to a K-3 by choosing the same magnification (100%) for both.

Stating that the K-3 is noisier based on comparing 100% crops is like saying "My lenses are sharper on my K-5 compared to the K-3."

Comparing the sharpness of images when dealing with different amounts of MP only works by first normalising the number of MP (e.g., downscale one image, or downscale them both). Otherwise the higher-MP camers provides a magnifying glass effect on what the lens delivered and of course with increasing magnification, the lens flaws will become more visible.

The very same holds true for image noise. If you increase the magnification (e.g., by using more MP) then you'll see more noise.

The DxOMark results correspond to actually perceived image noise when you print (or view at) the same size, so it is worth accepting them as a true measure of a sensor's performance (unlike comparing sensors at 100% each).

The K5/K5II's raw file dynamic range capability seems almost unprecedented. I'm not sure if I will buy the K3, but even if I do, I want to have the K5 as well.

The K-5's sensor is very good which is why Pentax has been using variants of it many times since then.

However, at the moment I don't have any reason to believe that the K-3's sensor will show a significantly different performance. The differences you are seeing are most likely all due to inadequate comparisons. Once DxOMark results are available, we'll know for sure.

[mattb123]

Even if you save your files as DNG, LR4 will only give you a generic "non-profile" that doesn't reflect any advancements in the newer versions of ACR that is the engine of LR. I've always assumed that if one gets a new model camera, the latest and greatest software has to be used to ensure feature compatibility. It's kind of a racket that way. I would think that the next dot release of LR5 will integrate the K-3 and any new Pentax lenses. You may be able to roll your own K-3 profile for use in LR4 but that involves time, discipline, and maybe some money. I'd rather pay for the upgrade.

M

Yeah, I probably would get it eventually anyway. Once LR5 has a profile for the K-3 I imagine that will be the catalyst for me.

[Cynog Ap Brychan]

To take a more extreme example, I didn't see any loss of dynamic range when I changed from a D3 to a D800, which has three times as many MP. I'm hoping the K-3 will be on a par with the K-5 in that respect. But we shall discover all in the coming weeks.

[Class A]

Even if you save your files as DNG, LR4 will only give you a generic "non-profile" that doesn't reflect any advancements in the newer versions of ACR that is the engine of LR.

Of course the ACR version used by LR4 is not as up to date than the one used in LR5, but in terms of IQ there is no difference that I'm aware of. Both already support PV 2012 and I don't think Adobe does much in terms of supporting individual Pentax models. Adobe does support Canikon with more profiles, but for Pentax cameras you basically just get "Adobe Standard" and I don't think that there is a lot of tailoring to a specific Pentax model. They don't use a different demosaicing algorithm specifically tailored to the K-3, or something like that.

If all one wants is specific support for one's camera, it isn't necessary to upgrade to the latest LR version. One can just use the free Adobe DNG profile editor to create one's own tweaked camera profiles. Or use software such as from X-rite and a Color Passport to automatically generate the camera profile.

[Barry Pearson]

Yes, but you don't magnify to 100% and then paste that 100% crop to Facebook. ** You display the whole image in whatever medium you're using, and this is going to involve downsampling. When this downsampling happens, any more noise you see at the pixel level of the more magnified, higher-res image is compensated for. In other words, there's no IQ downside to more MP, assuming similar pixel QE - you'll either get the same noise performance with equal detail if radically downsized, or the same noise with more detail if slightly downsized. If printed or displayed at native size (really big,) a slight NR to the higher-MP image is going to clean up the noise, equaling it, while (probably) keeping more detail at the same time. Both sharpening and NR are easier to do, IMO, with more megapixels to work with - the results look more real when done sparingly and carefully.

My first impression comparing the K-3 with the K-5IIs at ISO 51200 is that, printed at A3+, both colour noise and luminance noise of the K-3 are worse than that of the K-5IIs. I find I need what I consider to be excessive Lightroom noise reduction parameters with the K-3 to get close to the K-5IIs. A friend examined my A3+ prints yesterday and agreed with me - it isn't just me.

Equivalent parameters (for ISO 51200) in Lightroom: K-5IIs - Luminance 50, Color 25; K-3 - Luminance 75, Color 50. The latter parameters are visually damaging the K-3 image. (The scenes were the same - the Imaging Resource tests, using DNGs).

I haven't performed my own side-by-side tests with these two cameras, but first results for similar but not identical photos are not promising. I hope I'm wrong, but until better results are in I suggest people don't assume that K-3 High ISO noise will be as a good as K-5IIs High ISO noise for a given scene shown or printed at the same size.

[Barry Pearson]

Of course the ACR version used by LR4 is not as up to date than the one used in LR5, but in terms of IQ there is no difference that I'm aware of. Both already support PV 2012 and I don't think Adobe does much in terms of supporting individual Pentax models. Adobe does support Canikon with more profiles, but for Pentax cameras you basically just get "Adobe Standard" and I don't think that there is a lot of tailoring to a specific Pentax model. They don't use a different demosaicing algorithm specifically tailored to the K-3, or something like that.

If all one wants is specific support for one's camera, it isn't necessary to upgrade to the latest LR version. One can just use the free Adobe DNG profile editor to create one's own tweaked camera profiles. Or use software such as from X-rite and a Color Passport to automatically generate the camera profile.

I would expect Adobe to fine-tune the colour matrices and a few other details for specific models, once they get hold of a camera.

However, I go with everything else you say. I never use Adobe profiles. At the moment I'm using the embedded profiles in the DNGs of my K-3, but soon I will use X-Rite's ColorChecker Passport and create my own profiles, as I did with previous cameras. (I will have a 2-luminant general purpose profile and a 1-luminant profile for my studio lights).

[Rondec]

My expectation with the K3 would be that there is more detail at low iso, allowing somewhat larger print sizes/cropping. At high iso, I would imagine the difference goes away. I seldom shoot above iso 3200, although sometimes I have to push those files in post up to 6400. I generally turn those files into B and W photos, because even if the K5 is "superior" at those isos, it isn't great. Not talking noise there. I really don't care about noise much -- mainly cause I shot film long enough that it doesn't bother me. It is the lack of dynamic range. Colors start looking washed out and don't retain their original hues.

Looks to me like I will be satisfied with the K3, when I finally get one...

[Rorschach]

My expectation with the K3 would be that there is more detail at low iso, allowing somewhat larger print sizes/cropping. At high iso, I would imagine the difference goes away. I seldom shoot above iso 3200, although sometimes I have to push those files in post up to 6400. I generally turn those files into B and W photos, because even if the K5 is "superior" at those isos, it isn't great. Not talking noise there. I really don't care about noise much -- mainly cause I shot film long enough that it doesn't bother me. It is the lack of dynamic range. Colors start looking washed out and don't retain their original hues.

Looks to me like I will be satisfied with the K3, when I finally get one...

Well, when there is very little light, that's what colours do: they gradually disappear and the world becomes greyscale. So I am not sure if that's a huge problem if one wants to preserve the atmosphere of the original scene.

[Samoht]

Starting to get confused now, reading this thread I get the impression that higher MP is a disadvantage and I'll be better off buying a K-5II rather than the K-3 that I was going to buy. Think I need to read a bit more about the K-3 before deciding which to buy.

[Winder]

My first impression comparing the K-3 with the K-5IIs at ISO 51200 is that, printed at A3+, both colour noise and luminance noise of the K-3 are worse than that of the K-5IIs. I find I need what I consider to be excessive Lightroom noise reduction parameters with the K-3 to get close to the K-5IIs. A friend examined my A3+ prints yesterday and agreed with me - it isn't just me.

We're talking about ISO 51200. What where the expectations for performance at this ISO? I have my auto ISO on the K-3 set to 6400 which is pretty high for most photography.