[dtmateojr]

what I meant by intended size is that given the same dpi (e.g. 240) you print as per image size. We expect the 36mp to have much larger print of course. Otherwise, what's the point of 36mp if you print at the same size that 12mp is capable of doing? To see what 36mp is really capable of, you print nearly twice as large as 12mp. And that's why the dpreview comparison makes sense because they view the images at the same magnification. When you make a comparison you want to be as fair as possible. If the comparison is print size, D800 wins by a significant margin. If the comparison is noise, D700 wins not by much but quite visible.

[jsherman999]

what I meant by intended size is that given the same dpi (e.g. 240) you print as per image size. We expect the 36mp to have much larger print of course. Otherwise, what's the point of 36mp if you print at the same size that 12mp is capable of doing?

Your point (unless I misunderstood you) was that when you print or display at the same size, the 12MP image will have less noise than the 36MP image - and that's not the case. Those two sensors will have about the same amount of noise at the same display sizes.

There is no 'noise' downside to the more MP here, in other words. There's no scenario where the D700 image can look 'better', because even if you print both very very large, you're up-sampling the D700 image more, highlighting the lack of resolution, making up data at a higher rate, etc.

I'm not badmouthing the D700, by the way, for my typical uses it was fantastic and I'd probably choose it today over any aps-c camera.. but it has no image noise advantage over the D600 or D800 or A7R.

Your original point was good, though, that with shadow-heavy images you'll see the delta in noise profiles more. It's hidden a lot of the time in bright, well-lit shots, and people falsely assume that it's similar, only to get a nasty surprise when they shoot in very low light and start processing the images.






.

[Eyewanders]

These convos are strange to me. Grain is only a problem if it's objectionable, which I don't believe the K30 or K-3 produce. Pick up a 77limited and a monopod and use it in place of that 50-135 and see what happens. I love my 50-135, but for many club scenes, it is just not quite fast enough. Hell, pick up an old pentax film body for nothing and a 77ltd with some fast film and see what happens. The 77 won't depreciate and a K1000 can be found for ridiculous cheap.

[dtmateojr]

Grain isn't much of a problem. Chroma noise is.

---------- Post added 03-06-14 at 12:22 AM ----------

Your point (unless I misunderstood you) was that when you print or display at the same size, the 12MP image will have less noise than the 36MP image - and that's not the case. Those two sensors will have about the same amount of noise at the same display sizes.

There is no 'noise' downside to the more MP here, in other words. There's no scenario where the D700 image can look 'better', because even if you print both very very large, you're up-sampling the D700 image more, highlighting the lack of resolution, making up data at a higher rate, etc.

I'm not badmouthing the D700, by the way, for my typical uses it was fantastic and I'd probably choose it today over any aps-c camera.. but it has no image noise advantage over the D600 or D800 or A7R.

Your original point was good, though, that with shadow-heavy images you'll see the delta in noise profiles more. It's hidden a lot of the time in bright, well-lit shots, and people falsely assume that it's similar, only to get a nasty surprise when they shoot in very low light and start processing the images.






.

At 240dpi, 12mp will print at 12x18 and 36mp at 20x30. At those print sizes, the 12Mp will definitely look cleaner. You can print the 36mp at 12x18 but the chroma noise will still be worse than the 12mp print although grain will be averaged out.

[jsherman999]

At 240dpi, 12mp will print at 12x18 and 36mp at 20x30. At those print sizes, the 12Mp will definitely look cleaner.

But it will also be smaller, and show less detail. Not really a fair comparison. Also you would be hard-pressed to say that the smaller, less detailed image really looks in any way 'better', because the finer-grained luminance noise on the higher-res image isn't really objectionable, unless you've accidentally underexposed and have a lot of chroma in there. (which you can also do with the smaller-res image.)

You can print the 36mp at 12x18 but the chroma noise will still be worse than the 12mp print although grain will be averaged out.

Not my experience. I think the downsampling algorithm Lightroom uses by default on export (adaptive bicubic) handles these things nicely:

D800 (ISO 6400, crop of 12MP downsample) :

D700 (ISO 6400, crop of native image, no enlargement) :


The noise pattern in particular is impressive, to me - look at the noise above the $10 bill, where the page meets shadow - blotchy in D700, fine-grained and almost acceptable in D800. Notice how the crease of the book in that shadow is defined differently. The better, more realistic definition in the noise pattern in the D800 shot translates directly to a slightly more refined, 'real-looking' image when downsampled or printed. It also tends to hold up to NR a bit better.

Taken from:

[dtmateojr]

Your "blotchy" D700 is not just affected by noise but by resolution. The text in the D700 is barely readable while the D800 looks sharp. Do you have another sample? It looks like you failed to focus properly or you cropped too tight. I don't see chroma noise in either of them so it looks like LR was set to it's default of +20 something for chroma noise reduction.

[jsherman999]

Your "blotchy" D700 is not just affected by noise but by resolution. The text in the D700 is barely readable while the D800 looks sharp. Do you have another sample? It looks like you failed to focus properly or you cropped too tight. I don't see chroma noise in either of them so it looks like LR was set to it's default of +20 something for chroma noise reduction.

No, I don't have another sample of that scene, but this was one of a series of several braced shots where I tried CDAF (live view), PDAF and MF just to make sure It wasn't a focus issue. Also, you can see from the objects in the image behind and in front of the plane of focus that there's no FF/BF. You're seeing one of the benefits of higher-resolution there (and, to be fair, a much newer sensor.)

This is typical of what I saw in owning both bodies for a year, and it's part of the reason places like DxOmark resample before they score noise.

Here's another example from Joseph James. He has an interesting way of stating things: "... Does it really make sense to call the 50D pics "more noisy"? Or, instead, does it make more sense to say the noise is basically the same, but the 40D pics are simply "more blurry"? ... Along these lines, I would argue that if the image with the higher pixel count can be downsampled (and/or NR can be applied) so that maximum frequency of noise matches that of the lower pixel count image, and the amplitude of the downsampled/NR image matches, or is less than, the amplitude of noise of the image with the lower pixel count, then the image with the greater pixel count is not "more noisy"."

.

---

Last edited by jsherman999; 03-06-2014 at 08:25 AM.

[dtmateojr]

No, I don't have another sample of that scene, but this was one of a series of several braced shots where I tried CDAF (live view), PDAF and MF just to make sure It wasn't a focus issue. Also, you can see from the objects in the image behind and in front of the plane of focus that there's no FF/BF. You're seeing one of the benefits of higher-resolution there (and, to be fair, a much newer sensor.)



This is typical of what I saw in owning both bodies for a year, and it's part of the reason places like DxOmark resample before they score noise.



Here's another example from Joseph James.



.

If your shot is perfectly in focus then this blotchyness (it's blurring really) is caused by cropping too tight and the resolution advantage of the D800 becomes pretty obvious. The noise profile is still debatable though. Another way to look at why this resampling "trick" is not fair is that you could stitch multiple D700 shots to equal 36Mp so that you do not have to downsample the D800. In this scenario you can totally remove the resolution mismatch from the equation and concentrate on noise profiles. SNR afterall is affected not by image size but by sensel size which is the whole point of this exercise. Expect the noise profile of the K5 to be the same as the D800 and D7000. Further to that, while downsampling may apparently reduce luminance noise, it won't affect chroma noise; the absence of the latter in both samples is the reason why I think your test is flawed (most likely LR NR). I expected ugly magenta and blue patches in both at ISO 6400.

Now upsampling is quite different. You don't upsample just so you could pixel peep deeper. There is no point in pixel peeping because no new data is added. You upsize to print larger so you can see it from farther so apparent noise really doesnt change

[Rondec]

If your shot is perfectly in focus then this blotchyness (it's blurring really) is caused by cropping too tight and the resolution advantage of the D800 becomes pretty obvious. The noise profile is still debatable though. Another way to look at why this resampling "trick" is not fair is that you could stitch multiple D700 shots to equal 36Mp so that you do not have to downsample the D800. In this scenario you can totally remove the resolution mismatch from the equation and concentrate on noise profiles. SNR afterall is affected not by image size but by sensel size which is the whole point of this exercise. Expect the noise profile of the K5 to be the same as the D800 and D7000. Further to that, while downsampling may apparently reduce luminance noise, it won't affect chroma noise; the absence of the latter in both samples is the reason why I think your test is flawed (most likely LR NR). I expected ugly magenta and blue patches in both at ISO 6400.

Now upsampling is quite different. You don't upsample just so you could pixel peep deeper. There is no point in pixel peeping because no new data is added. You upsize to print larger so you can see it from farther so apparent noise really doesnt change

I am not sure what you are arguing. But what I have seen, in general, is that sensors with more megapixels look better when printed at the same size. There is more detail there, noise is at worst equivalent (D800 tests better than D700 with regard to sports iso). There just isn't any down side, other than larger file sizes and slower processing as a result.

Obviously the D800 gives you two options -- print the same size with more detail and less noise than D700 or, print larger with more detail and equivalent noise to the D700. Either way, you are winner.

[dtmateojr]

I am not sure what you are arguing. But what I have seen, in general, is that sensors with more megapixels look better when printed at the same size. There is more detail there, noise is at worst equivalent (D800 tests better than D700 with regard to sports iso). There just isn't any down side, other than larger file sizes and slower processing as a result.



Obviously the D800 gives you two options -- print the same size with more detail and less noise than D700 or, print larger with more detail and equivalent noise to the D700. Either way, you are winner.

Not really. Depends on print size. In good light, a point and shoot camera can easily match a D800 print at 4x6 or even 5x7.

This illusion of more megapickles means less noise is a dream. Don't we all wish it's that simple? But we can't deny the results. My D700 shoots way cleaner than my K5II. My K5II shots look very similar to the D800 shots that I have seen (they are exactly the same sensor afterall). I own an E-M5 too which has a much higher resolution than the D800 and while it's noticeably noisier, I still find the ISO 6400 shots quite acceptable.

This argument on noise, in the end, does not really matter though. If you really have to shoot at insane ISOs it probably means that content is more important than aesthetics. If people noticed the noise in your shot, it's most likely because your shot was crap. Noise matters a lot in marketing a product but is immaterial in real photography.

[Rondec]

Not really. Depends on print size. In good light, a point and shoot camera can easily match a D800 print at 4x6 or even 5x7.

This illusion of more megapickles means less noise is a dream. Don't we all wish it's that simple? But we can't deny the results. My D700 shoots way cleaner than my K5II. My K5II shots look very similar to the D800 shots that I have seen (they are exactly the same sensor afterall). I own an E-M5 too which has a much higher resolution than the D800 and while it's noticeably noisier, I still find the ISO 6400 shots quite acceptable.

This argument on noise, in the end, does not really matter though. If you really have to shoot at insane ISOs it probably means that content is more important than aesthetics. If people noticed the noise in your shot, it's most likely because your shot was crap. Noise matters a lot in marketing a product but is immaterial in real photography.

My assumption is that you will compare sensors of the same size. A full frame sensor of even a generation ago will out perform an APS-C sensor, which should outperform a four thirds sensor, with regard to SNR, noise and dynamic range. This is shown across DXO Mark's testing. What is not shown is that if you keep the sensor size the same, but increase number of pixels, that you increase noise.

As to whether or not noise is important, it depends on the shot. If you are shooting landscapes or portraits, then absolutely, noise can be detrimental to the final result. On the other hand, if you are shooting in night clubs or shooting rock bands, then noise can easily just add to the gritty feel of your shots,

[fretlessdavis]

Not really. Depends on print size. In good light, a point and shoot camera can easily match a D800 print at 4x6 or even 5x7.

The human eye resolves between 5 and 10 lp/mm in a print at close inspection, meaning between 120 to 240 ppi on the print. Given the same optics, exposure, and processing, you wouldn't be able to tell at 8x10 between 12, 16, 24, or 36 megapixel images. It's just not humanly possible. For those with slightly less than ideal eyesight, you would likely not even be able to tell the difference. I'd venture to guess that people not trained as photographers wouldn't be able to tell the difference on the 5 lp/mm end of things. Also, without direct comparison, 150 ppi prints still look fantastic to the naked eye. Put them behind glass and up on a wall, and resolution really becomes a moot point.

Let's put it this way-- when printed at 150 ppi, a K-5 image will be 30" wide. How often will a 20"x30" print, be viewed in open air, with a nose all the way into it. How many people that would do that will have perfect vision, trained eyes, and a higher resolution print as a direct comparison?

Resolution is a non-issue these days.

[MJSfoto1956]

As clearly shown in analyzing DxOmark scores of bodies + lenses, a D800 + Sigma 24-70mm will be "out resolved" by a D7100 + Sigma 18-35mm. So saying that "sensor x" will out resolve "sensor y" is pretty much meaningless unless you also compare which lenses will be used in that comparison. One of the very nice things about the K-3 + Limited lenses is that not only does it provide a small, lightweight solution but that it offers "similar" IQ as a D800 used with certain highly-respected zoom lenses (Note that a D800 + primes wins the contest handily). Which is to say, the quality can be very very good indeed for both camera systems -- much of it depends on which lenses you intend to use.

YMMV

Michael

[dtmateojr]

My assumption is that you will compare sensors of the same size. A full frame sensor of even a generation ago will out perform an APS-C sensor, which should outperform a four thirds sensor, with regard to SNR, noise and dynamic range. This is shown across DXO Mark's testing. What is not shown is that if you keep the sensor size the same, but increase number of pixels, that you increase noise.



As to whether or not noise is important, it depends on the shot. If you are shooting landscapes or portraits, then absolutely, noise can be detrimental to the final result. On the other hand, if you are shooting in night clubs or shooting rock bands, then noise can easily just add to the gritty feel of your shots,

The E-M5 is actually cleaner than a Sony A900. It depends on the sensor as well. Older Sony sensors sucked. My A700 was already too noisy at base ISO after 15 seconds.

[jsherman999]

This illusion of more megapickles means less noise is a dream.

It doesn't mean more resolution means less noise, it means that you can't compare image noise by looking at things at the pixel level, and in many cases more MP actually results in the same image noise. The D800 vs. D700 is a great example of that. K3 vs. K5IIs is another, although there still may a very slight noise advantage to the K5IIs because of the K3 implementation (maybe imperceptible.)

My D700 shoots way cleaner than my K5II.

That's not because the D700 pixels are bigger (or better,) that's because the sensor area is larger

My K5II shots look very similar to the D800 shots that I have seen (they are exactly the same sensor afterall).

The K5 and D7000 share the same pixel tech as the D800, but the D800 has larger area - thus, for the same FOV and shutter speed and F-stop it's image will have about a stop less noise. It's sensor area that dictates that (along with the linear aperture of the lens being used to obtain that same FOV/F-stop.) I really hope you don't start claiming that a D7000 system has exactly the same low-light capability as the D800 or A7R now We'll need more lenses like the Sigma 18-35 f/1.8 before that could be generally true.

.

This argument on noise, in the end, does not really matter though. If you really have to shoot at insane ISOs it probably means that content is more important than aesthetics. If people noticed the noise in your shot, it's most likely because your shot was crap. Noise matters a lot in marketing a product but is immaterial in real photography.

I disagree. I shoot at ISO 6400 all the time, and I do end up cropping some of those shots to other dimensions, and often make prints for people. The less noise, the better, and personally I'd like to see another stop improvement in the next 5 years - I could use it.

Anyone who shoots in low-light venues would probably agree. I also think a lot of birders & sports shooters would like to be able to get really fast shutter speeds (negating the need for a tripod all the time) while maintaining minimal noise - better SNR brings that. I doubt a landscape shooter would need it as much though.

.