How much quality do you really need?

[myotis]

Hi,

My impression is that 180 PPI is the resolution where observers have difficulties telling prints apart.

Best regards
Erik

This is a great rule of thumb to have available for resolution, but in terms of print quality, can we then use another rule of thumb that there after quality will increase as sensor size increases (assuming sensors from similar generations, as I am aware of your Sony/Hasselblad comparison).

Cheers,

Graham

[ErikKaffehr]

Hi,

One way to see it would be to opt for a pixel size that avoids aliasing at your most used aperture. For f/8 I would guess something like 3.5 microns.

Smaller formats would be optimised for larger apertures. If you shoot f/5.6 on 4/3 you may need 2 microns to avoid aliasing.

That would be something like 70 MP on 24x36 and 54 MP on 4/3.

The best lenses, like Otus are optimal at f/4 (at least near axis), using f/4 the optimal resolution would be something like 140 MP on 24x36mm.

But, those image would be quite soft when viewed at actual pixels on screen.

Best regards
Erik
 

This is a great rule of thumb to have available for resolution, but in terms of print quality, can we then use another rule of thumb that there after quality will increase as sensor size increases (assuming sensors from similar generations, as I am aware of your Sony/Hasselblad comparison).

Cheers,

Graham

[myotis]

Hi,

That would be something like 70 MP on 24x36 and 54 MP on 4/3.

Best regards
Erik

Well that is much bigger than I expected! (and really interesting)

But this still seems to be centred on resolution, don't larger photo sites bring benefits in colour and tonal quality. 

Cheers,

Graham

[ErikKaffehr]

Hi,

Not really…

What affects colour is mostly the effort going into colour profiles and the CFA (Colour Filter Array).

Image noise is affecting both colour rendition and tonality, but image noise is depending on the number of photons captured. It doesn't really matter if they are captured by small pixels. Half size pixels can capture half the photons but you have twice the number pixels so the number of photons is the same.

This doesn't work really for read noise, so at very high ISO ratings larger pixels will be better. Another factor is that you want to keep the wiring to photodiode surface low.

That really means that there is an optimum pixel size for a given width of wiring and that optimal size is going down.

Phase One has "sensor +" that is hardware binning on some of it's sensors. That illustrates the issue pretty well. The effect of hardware binning shows as a bump in the DR curve in "screen" mode. It is also visible in the tonal range curve.

Switching to print mode the bump is pretty weak in the DR curve, but not all visible in tonal range curve.

DR is essentially defined by readout noise (it's definition is Full Well Capacity / read noise) while tonal range is dominated by full well capacity.

Screen mode is same as actual pixels on screen. Print mode is normalised to a small print size. The print size doesn't really matter, it would affect the numbers but not the ratios.

So, the essence is that you can reduce pixel size within reasonable limits without negative effects on image quality. What reasonable limits are varies with technology.

My guess is that what we have now, like 36-50 MP on 24x36 and 24MP on APS-C may be optimum today. But, to make best use of the best lenses we could have quite a bit more.

Best regards
Erik

But this still seems to be centred on resolution, don't larger photo sites bring benefits in colour and tonal quality. 

Cheers,

Graham

[myotis]

So, the essence is that you can reduce pixel size within reasonable limits without negative effects on image quality. What reasonable limits are varies with technology.

My guess is that what we have now, like 36-50 MP on 24x36 and 24MP on APS-C may be optimum today. But, to make best use of the best lenses we could have quite a bit more.

Best regards
Erik

I can't pretend to fully understand all of this, but I grasp the principal. I need to do a bit more homework on this it seems.

Thanks for taking the time to explain.

Cheers,

Graham

[Ray]

In the original question by Graham, the example of upgrading from a 12mp D700 to a 36mp D800 was used. Just out of curiosity, I checked the DXOMark website to compare these two cameras, selecting the later D810 model in preference to the D800.

What surprised me is that at base ISO, at the pixel level (screen mode), the much smaller D810 pixel is at least as good as the larger D700 pixel in all the parameters addressed by DXO, such as SNR, Tonal Range and Color Sensitivity. However, the Dynamic Range of the smaller pixel (which is really SNR in the shadows) is hugely better by almost 2 stops.

What this means is, if the subject is still and one can use a tripod, or if the lighting is bright enough to use the low base ISO of 64 with the D810 hand-held, a 12mp crop from the D810 sensor should have better noise characteristics than the full-frame shot from the D700, despite the larger pixels of the D700.

Unfortunately, this advantage of the D810 at the pixel level is not maintained at all higher ISOs. From ISO 200 to 6400 the D700 pixel has better SNR at 18% (skin tones), and marginally better tonal range and color sensitivity. However, the DR of the D700 pixel is marginally better only at ISO 800 and 1600.

http://www.dxomark.com/Cameras/Compare/Side-by-side/Nikon-D810-versus-Nikon-D700___963_441

[myotis]

In the original question by Graham, the example of upgrading from a 12mp D700 to a 36mp D800 was used. Just out of curiosity, I checked the DXOMark website to compare these two cameras, selecting the later D810 model in preference to the D800.

What this means is, if the subject is still and one can use a tripod, or if the lighting is bright enough to use the low base ISO of 64 with the D810 hand-held, a 12mp crop from the D810 sensor should have better noise characteristics than the full-frame shot from the D700, despite the larger pixels of the D700.

I have compared my D600 with APS crop against a D7000 for bird photography with a 300mm lens. The cropped D600 file was  noticeably better than the D7000. Given the better buffer of the D600, there seemed little point in using the D7000. This was with a range of ISOs up to 1600.

Cheers,

Graham

[Ray]

I have compared my D600 with APS crop against a D7000 for bird photography with a 300mm lens. The cropped D600 file was  noticeably better than the D7000. Given the better buffer of the D600, there seemed little point in using the D7000. This was with a range of ISOs up to 1600.

Cheers,

Graham

Hi Graham,
Better in what respect? If the D600 has a faster frame rate, larger buffer, more accurate auto-focussing etc, then those features alone would result in a greater percentage of better shots.

Whilst there's no doubt that the larger D600 pixel has a general noise advantage over the D7000 pixel, at most ISOs, (DR at base ISO being the exception - according to DXOMark), those noise advantages are reduced when the 16mp D7000 files are down-sampled to the 10.6mp file size of the D600 used in APS crop mode. When comparing image technical quality one should always compare equal size files.

I would guess that the slight advantage of the lower noise in the cropped D600 image, would be largely offset by the advantage of the higher resolution of the uncropped D7000 image. At least the D7000 provides those options. If noise is not a problem, one can take advantage of the higher resolution of the D7000 file. If noise is a problem, then one can reduce the D7000 file-size to the 10.6 mp of the cropped D600 image, which also results in a noise reduction, bringing the noise of the down-sampled D7000 image close to that of the cropped D600 image, I imagine. 

[myotis]

Hi Graham,
Better in what respect? If the D600 has a faster frame rate, larger buffer, more accurate auto-focussing etc, then those features alone would result in a greater percentage of better shots.

I mean better in image quality: sharpness and contrast, these were just tests so frame rate, larger buffer etc AF etc weren't being considered in the tests, but they were the the reason for doing the tests.  I only tried it because its a question raised fairly regularly in bird photography forums/blogs, and I just confirmed what others were saying.

I simply compared printed images or I exports as same size Jpegs to compare on screen, I have assumed that (maybe wrongly) that LR/C1/PS would resample for the appropriate print/file size.  At least well enough to allow a comparison. Is that wrong.


Cheers,

Graham

[razrblck]

The D7000 has very similar pixel size to the D800, and compared to a D600 it should have a stronger AA filter (giving a bit less sharpness overall). The buffer is atrociously small, especially if you shoot lossless RAW (that's like 7 shots before you have to wait, and even the faster SD cards don't help much but give you a few seconds less of waiting). On top of that, there are two years of sensor tech advancements between the two cameras. I'm not surprised you found differences in IQ.

[myotis]

The D7000 has very similar pixel size to the D800, and compared to a D600 it should have a stronger AA filter (giving a bit less sharpness overall). The buffer is atrociously small, especially if you shoot lossless RAW (that's like 7 shots before you have to wait, and even the faster SD cards don't help much but give you a few seconds less of waiting). On top of that, there are two years of sensor tech advancements between the two cameras. I'm not surprised you found differences in IQ.

Well, it didn't come as too much of a surprise as it mirrored other people's experiences, but before that,  I had simply assumed that the pixel advantage of the D7000, compared to the cropped  D600, would outweigh the buffer, etc advantages of the D600.

Cheers,

Graham

[johnwolf]

This is a timely subject for me. I've lately been printing extensively for a photo wall in my home. Lots of 12 x 18" prints, all black and white, of places we've been and people we've met. They are from a former Panasonic GX7 and my current cameras -- Ricoh GR, Fuji X-Pro1, and Leica M-Monochrom.

I realize my m4/3 is old technology, but the prints look obviously inferior to me. Still pretty good, but they seem to have less depth. And this print size seems about their reasonable limit.

Of prints from the other three, I see very little difference. Certainly nothing jumps out as a qualitative difference. I know this is not a rigorous test, but I don't care about that.

I did do a same scene comparison between the Fuji and Leica, and, again, neither print jumps out as superior, even with the Fuji color conversion. I am partial to the rendering of lighter grays, so I did a somewhat high key test. The cameras produce equally lovely and delicate upper tones. Maybe a test of the lower tones would be different. 

If anything can be said about this entirely unscientific test, it's that I'm amazed at the black and white rendering of the GR.

You can argue with my methodology or challenge my skills, but this is how I shoot and view prints, so it's what I care about and am able to produce. As a result of this, I'm considering selling my MM and lenses. There's a lot of money tied up in them and at my print size, I'm not sure they make sense.

To specifically answer the OP's question, had I done this printing before purchasing the MM, I would not have bought it. Well maybe I would because I like rangefinders, but not with print quality in mind.

John

[myotis]

This is a timely subject for me. I've lately been printing extensively for a photo wall in my home. Lots of 12 x 18" prints, all black and white, of places we've been and people we've met. They are from a former Panasonic GX7 and my current cameras -- Ricoh GR, Fuji X-Pro1, and Leica M-Monochrom.

I realize my m4/3 is old technology, but the prints look obviously inferior to me. Still pretty good, but they seem to have less depth. And this print size seems about their reasonable limit.

Of prints from the other three, I see very little difference. Certainly nothing jumps out as a qualitative difference. I know this is not a rigorous test, but I don't care about that.

To specifically answer the OP's question, had I done this printing before purchasing the MM, I would not have bought it. Well maybe I would because I like rangefinders, but not with print quality in mind.

John

You comments are very much to the point of my post in that "quality" once you get beyond a minimum acceptable resolution is about lots of different things, many of which are difficult to define.

But I think there is a problem  getting to grips with how you get that quality, I have a GX7, and compared to my Nikons (including my Nikon V1) and my Fuji, I just "don't like" the image quality, with prints having a two-dimensional feel  to them compared to a more 3d look of my other cameras. Some cameras like the Nikon D700 I mentioned have a reputation for giving especially nice image quality, and Fuji and  your Ricoh share that reputation, so not surprised they did well.

I have also found that using Capture 1, compared to lightroom, seems to give a more 3d look.  But I am still at the stage of trying to work out how much of these differences are real and how much are down to my lack of skill.

I was particularly interested in your comments on the GX7 as  it came out with very good reviews, but I have never been very happy with the image quality.

Cheers,

Graham

[johnwolf]

Graham, skill vs inherent quality is certainly an issue. Maybe someone with greater skill could coax more out of the MM. If so, they could probably do the same from the Fuji and GR. Where would it stop?

I think it's kind of a mute point, anyway, because our own skill is all we have. Assuming we're doing it all ourselves.

What I've really learned from my printing initiative is that content trumps pretty much everything else. And given the incredible capabilities today's lenses and sensors, the quality of the light is probably right behind.

Of course, we're talking about this print size.   

John

[myotis]

Graham, skill vs inherent quality is certainly an issue. Maybe someone with greater skill could coax more out of the MM. If so, they could probably do the same from the Fuji and GR. Where would it stop?

I think it's kind of a mute point, anyway, because our own skill is all we have. Assuming we're doing it all ourselves.

What I've really learned from my printing initiative is that content trumps pretty much everything else. And given the incredible capabilities today's lenses and sensors, the quality of the light is probably right behind.

Of course, we're talking about this print size.   

John

Yep, I fear I am going to be dead before I reach the level of skill I feel I need :-(

Post processing skills seem to have  a tremendous bearing on how impressed I am with a photograph. For example, having decided that Capture 1 is much better than Lightroom, most of  the photographs I am impressed by, are processed in Lightroom !  But, as you say its also about light, in fact I suspect light (and visualisation of how that light will work in the print) is more important than content, followed by printing skills, followed by content, followed by camera gear.

Or maybe they are all equally important, I think I take better photographs with camera gear I feel "happy/comfortable" with, even though i have it last in my list.

Cheers,

Graham

 

[razrblck]

Tools are just a mean to realize your vision. It doesn't matter how good the tools are, it matters how well you can use them to reach your goal.

[Quentin]

To determine which camera performed best, I tried the following test:

1.  Hasselblad H5D, various lenses, Tripod, filters, bag. 

2.  Sony A7R2, Loxia 21mm,  various other lenses, filters. 

In a direct comparison between shots taken with the two camera systems, these were my findings

The Sony performed far better than the Hasselblad.

At first, this was a surprise.   But then I realised that due to issues of weight, I hadn't taken the Hasselblad with me. 

Lesson learnt.  Every camera has its limitations.  YMMV.

[myotis]

Tools are just a mean to realize your vision. It doesn't matter how good the tools are, it matters how well you can use them to reach your goal.

This is one of these truisms that I don't thinks is really true.  The tools still need to be "good enough" to give you the capability of achieving your vision, and its easier to do this with tools you are familiar with.

Cheers,

Graham

[myotis]

To determine which camera performed best, I tried the following test:

1.  Hasselblad H5D, various lenses, Tripod, filters, bag. 

2.  Sony A7R2, Loxia 21mm,  various other lenses, filters. 

In a direct comparison between shots taken with the two camera systems, these were my findings

The Sony performed far better than the Hasselblad.

At first, this was a surprise.   But then I realised that due to issues of weight, I hadn't taken the Hasselblad with me. 

Lesson learnt.  Every camera has its limitations.  YMMV.

Very good :-)

 I now carry a Nikon V1 with the excellent 70-300 cx lens nearly everywhere I go.  it weighs less than 1kg, and sits in the palm of my hand. In good light it gives very good results, in poor light - not so good :-(

But for years, when doing ecological field work, I couldn't carry any camera gear, and missed out on many  good wildlife photographs, now I have the nikon, with the equivalent of an 810mm lens,  with me all the time.

Cheers,

Graham

[Quentin]

I've given up lugging heavy stuff everywhere.  Quality has come to smaller cameras.  An A7RII, my current weapon of choice, coupled with appropriate lenses, is very good and I can enjoy going places without compromise of image quality.

Photography is fun again!