[Pål Jensen]

we are talking about the difference between 36Mp (D800E) and 16Mp (K-5IIS)
- which is in theory 149% increase in linear resolution for the D800E over the K-5IIS
The K-5IIS is very good - but it is not miraculous.

Linear resolution is as meaningful as one-dimentional photographs. The difference in resolution between the 16mp and 36mp is ~20-30% (too lazy to do the math). If you want to double the K-5's resolution you need 64mp.

[UnknownVT]

Linear resolution is as meaningful as one dimentional photographs. The difference in resolution between the 16mp and 36mp is ~20-30% (too lazy to do the math). If you want to double the K-5's resolution you need 64mp.

The math is easy for area or 2 dimensional "resolution"
is it very simply 36/16 = 225%
or 2.25x area of the 16Mp -
so it is more than double the 2 dimensional area "resolution".

FWIW - traditional photographic/film resolution is measured linearly -
to get from area/2-dimensional "resolution" to the regular linear resolution
is then a matter of taking the square root
- area or 2-dimension is (linear)squared =
thus SqRt(225%) = 149%.

Wikipedia entry on Image Resolution

Basically, resolution quantifies how close lines can be to each other and still be visibly resolved. Resolution units can be tied to physical sizes (e.g. lines per mm, lines per inch), to the overall size of a picture (lines per picture height, also known simply as lines, TV lines, or TVL)

[normhead]

Hmmmm... VT I have to say thanks for putting the work into that... having done this a few times in the last few days.. and realizing that it's not rocket science but it is tedious.. thanks for those images. To me, being able to look at your image comparisons is worth more than any description or reference to spread sheets that purport to examine the issues doesn't tell me what I need to know. Because the issue is "what will this mean to me, and to decide that I need to see the image comparisons. I need to be able to look at the K-5 images and say.."OK I can live with that."

What I was hoping to establish at this point was what is the magic number, where there would be a technical advantage to FF. The answer surprised me. I was expecting that as long as I was shooting within 4900 pixel width, the K-5 would deliver to it's full potential. I came up with a number somewhat less than that. The K-5 images need to be cropped to somewhere between 3500 and 4000 pixels before they can compete with D-800 images also cropped to 3500-4000 pixels. D800 images cropped to 4928 are clearer then K-5 images at the same resolution. And that is because the lenses available for the K-5 are can only resolve about about 70% of what the K-5 sensor is capable of resolving. SO it doesn't surprise me at all that K-5 images upsized to D800 size is not as sharp. that's what I'd expect.

So the "big" conclusion here is if you really need the best resolution at the 4928 pixels the K-5 is capable of, you're going to have to buy a D800 and reduce the size to 4928, to get it. If on the other hand if you never need your images more than 3500 to 4000 pixels (I haven't done exact enough comparisons to even guess what the exact number might be) then you can get away with using a K-5.

All the points about cropping are noted, however, being a film guy, I always frame to avoid cropping. It's a "how much can I enlarge this image" issue. Maybe sometimes in the field I make a mistake and end up wanting a different crop. I can live with that. It teaches me to be careful not to make mistakes in the field. And to zoom the lens with my eye to the viewfinder looking for a field of view I might have missed.

To the guys that say "yes but I can shoot and then crop later", I'd say, do you want to be a person who has the eye to pick good images out of not so good one's, or do you want to develop an eye for what's there and the ability to frame it correctly and capture it? One is a photographer, I'm not sure what the other guy is.

However, why don't we try comparing a 18 or 20MP FF image vs the k-5IIs, if we're trying to claim FF advantage - or even better, a 28+MP APS-C sensor vs the D800.

I guess you missed the thread where I pointed out that where I upscaled the K-5 image to the size of the D3200 (24 MP APS-c) image to show they were identical. The limit on IQ in APS-c is imposed by the lens, not by the sensor. What all this messing around has taught me, if I've learned anything from it, is that if I'm going FF, I'm going to a D800. Nothing else makes sense. The D800 has the same theoretical relationship to lens resolution the K-5 does. It's overkill in terms of lens resolution, but in terms of reducing moire, being ready for improvements in lens technology and getting the most out of existing lenses, it's right where it needs to be.

Its crazy but, if you like the K-5, you should also like the d800 as the next format going up the scale. The Pentax 645, unlike the the D800 and K-5 is not at the theoretical max of it's sensor capacity. That would be 70 MP. So even though the format is capable or more the 645 is just marginally better than the D800 in IQ. If the K-5 and D800 have taught us anything, its that there are benefits to exceeding the practical limits of the resolving power of the lens. What the D3200 has taught us is that there is no benefit to exceeding the theoretical limits of the sensor (using current technology). Right now the D800 is the logical successor to the K-5 throne, the 645D needs to go to 70 MP to get back in the discussion.

Why I don't own a D800? I can't make a convincing case for needing an image more than 3500 pixels wide.

[normhead]

Basically, resolution quantifies how close lines can be to each other and still be visibly resolved. Resolution units can be tied to physical sizes (e.g. lines per mm, lines per inch), to the overall size of a picture (lines per picture height, also known simply as lines, TV lines, or TVL)

When you reduce the image size of the D800 images you reduce the resolution. SO say the D800 originally had a resolution of 4000 lines. Reduce the size of the image by half and you now have a resolution of 2000 lines. The K-5 might start with a resolution of 2500 lines. If you reduce it by 25% it also has a resolution of 2000 lines. YOu reduced the D800 image to a much larger degree, but that doesn't matter. You now have two images the same size with the same resolution. After that point it doesn't matter how much you reduce the images, the APS-c and FF images are the same. If that wasn't the case, you could reduce a D800 image to thumbnail size and still expect 4000 lines of resolution, even though the image is only 90 pixels big. That obviously can't be true.

Think of it this way... if you take a 35 mm picture of a contact print made by an 8x10 film camera. You don't get the detail you get in the 8x10, you get the detail of a 35mm negative. It only matters what resolution the 8x 10 image is, if you use the 8x10 negative to print. If you are printing at 300 dpi, and reduce your D800 image to 3600 pixels wide for a 12 inch print, you get nothing you wouldn't have gotten from a K-5 image reduced to 3600 pixels for a 12 inch print.

---

Last edited by normhead; 10-26-2012 at 10:41 AM.

[UnknownVT]

I guess you missed the thread where I pointed out that where I upscaled the K-5 image to the size of the D3200 (24 MP APS-c) image to show they were identical. The limit on IQ in APS-c is imposed by the lens, not by the sensor.

I am unlikely to have missed the rather "lively" discussion in that other thread.

My recall was that you tried to explain the resolution measurements from Imaging-Rosurce.com by their extinction figures - which is not really practical.

To help in this visually,
all I have done was to up-res/upsample the K-5IIS crop to the same size as the 100% crop of the Nikon D3200 (budget entry level but 24Mp).

I do believe the D3200 image does show clearer and more distinct details than that of the upsampled K-5IIS.

Again it is pretty simple maths, from Post #76 in that other thread:

Originally posted by falconeye
It is much easier to simply compare MP figures. And 24 is a 22% linear increase over 16.

That's area in pixels- 24/16 = 150% more pixels
gives per linear dimension SqRt(1.5) = 1.22 = 122%

[UnknownVT]

When you reduce the image size of the D800 images you reduce the resolution. SO say the D800 originally had a resolution of 4000 lines. Reduce the size of the image by half and you now have a resolution of 2000 lines. The K-5 might start with a resolution of 2500 lines. If you reduce it by 25% it also has a resolution of 2000 lines. YOu reduced the D800 image to a much larger degree, but that doesn't matter. You now have two images the same size with the same resolution. After that point it doesn't matter how much you reduce the images, the APS-c and FF images are the same. If that wasn't the case, you could reduce a D800 image to thumbnail size and still expect 4000 lines of resolution, even though the image is only 90 pixels big. That obviously can't be true.

Think of it this way... if you take a 35 mm picture of a contact print made by an 8x10 film camera. You don't get the detail you get in the 8x10, you get the detail of a 35mm negative. It only matters what resolution the 8x 10 image is, if you use the 8x10 negative to print. If you are printing at 300 dpi, and reduce your D800 image to 3600 pixels wide for a 12 inch print, you get nothing you wouldn't have gotten from a K-5 image reduced to 3600 pixels for a 12 inch print.

Sorry, no disrespect meant -
but that's a lot of text which tries to explain simple resolution -
but on images that are altered.

Once altered like printed or up/down-sampled -
the original "resolution" can no longer apply -

The Wikipedia quote in the correct context is supposed to be a measurement on the original image -
ie: 100% size, unaltered, with NO up or down-sampling, or any form of manipulation.

[normhead]

The Wikipedia quote in the correct context is supposed to be a measurement on the original image -
ie: 100% size, unaltered, with NO up or down-sampling, or any form of manipulation.

From PentaxForums.com: https://www.pentaxforums.com/forums/pentax-dslr-discussion/203193-last-post-h...#ixzz2AQm8Ip7w

exactly...

so if you don't use the original image, because your computer screen doesn't go to 7000 pixles wide it goes to at most 2840, then your image has to be downsized to be viewed on the computer screen, and the original resolution numbers are meaningless. And I've found nothing to indicate a K-5 image downsized 20 % is better or worse than a D800 image downsized 50% so they are the same size. The k-5 image is as good as a D800 image at that resolution, at least that's what my incredibly incomplete, cursory examination of the issue concluded.

[UnknownVT]

exactly...

so if you don't use the original image, because your computer screen doesn't go to 7000 pixles wide it goes to at most 2840, then your image has to be downsized to be viewed on the computer screen, and the original resolution numbers are meaningless. And I've found nothing to indicate a K-5 image downsized 20 % is better or worse than a D800 image downsized 50% so they are the same size. The k-5 image is as good as a D800 image at that resolution, at least that's what my incredibly incomplete, cursory examination of the issue concluded.

That's why we use unaltered crops from the original 100% images -
I stated clearly when they were as-is,
and when I did any up-res/sampling
(since I was responding to the way you were doing your comparisons.)

This is the image in Post #106 of that other thread -



straight 100% crops taken from the ISO100 "default" NR JPGs from the respective Imaging-Resource Studio shots.

By your logic above then most people don't need more than about 2Mp since the common max resolution screens are 1920x1080 = ~2Mp
and for those still on VGA - then they really only need 0.3Mp -
that obviously doesn't make any sense.

What we may perceive as a practical and financial limit,
only has bearing on our own individual needs -
this may not apply to anyone else -
a resolution measurement is a resolution measurement -
whether it is practical, or if anyone actually needs it or not
is not for us to say -
nor does it alter the resolution measurement.

[normhead]

On those above images... if the images were printed at 92 dpi, so they'd appear as they do on my computer screen, how big would those images be? Just guessing, the K-5 image (of the whole picture) would be between 50 and 60 inches across. And how critical would the sharpness of that part be to the enjoyability of the image? If you were looking at the whole image, and your eyes fell on that part of the image, would it make any difference to you if the K-5 image looked like it does on the screen or the D800 image looked like it does? Would anyone even care? Walking up to a 60 inch image... I think they'd both look great. And until someone shows me two 60 inch images where I prefer the D800 to the K-5 version.. I'll probably continue to think that way.

[UnknownVT]

On those above images... if the images were printed at 92 dpi, so they'd appear as they do on my computer screen, how big would those images be? Just guessing, the K-5 image (of the whole picture) would be between 50 and 60 inches across. And how critical would the sharpness of that part be to the enjoyability of the image? If you were looking at the whole image, and your eyes fell on that part of the image, would it make any difference to you if the K-5 image looked like it does on the screen or the D800 image looked like it does? Would anyone even care? Walking up to a 60 inch image... I think they'd both look great. And until someone shows me two 60 inch images where I prefer the D800 to the K-5 version.. I'll probably continue to think that way.

I have not argued about what you or I would prefer or find practical.

There clearly is a visible, and measurable difference between 16, 24 and 36 Mp -
whether or not anyone wants it, or needs it -
really is not my place to say.
But there clearly is a difference.

[normhead]

I have not argued about what you or I would prefer or find practical.

For me, it's pointless talking about anything else. The ruminations of the theoretical geniuses only become relevant when they impact my world.

[UnknownVT]

For me, it's pointless talking about anything else. The ruminations of the theoretical geniuses only become relevant when they impact my world.

That's fine as long as it is clear that "no difference" means for your usage.

I know it is not my place to speak for anyone else's usage or needs.

It is still a fact, and not theory, that there are visible and measurable differences between 16, 24, and 36Mp.

[dylansalt]

I let my eyes do the judging not number crunching.

For me there's quite a large difference when viewing on a 24" HD monitor even if it is at 72dpi when I view pics taken with a D800 or 600 compared to all the shots I have taken with my D7000 at normal ISO's.

For some odd reason though I can't see it with shots taken with the Canon 5D MK11 where I feel my D7000 actually resolves better?

Maybe it also has to to with sensor design as well?

[bossa]

This thread is absurd. I can't believe someone would go to such lengths to discredit an actual photographic format. I realize qualifications have been stated but they are only to avoid direct attacks from people who give a damn.

Norm, did you ever create a thread like this stating why nobody 'needed' a Pentax 645D or any other medium format digital camera? Based on your crusade against the 'full-frame" I assume there must be something I can read?

My only comment is this: Art is never dependent on 'detail' that is removed and separate from the work at hand and your reference to Cezanne is evidence of that. There is no reason to create volumes of 'proof' for why you, personally, don't need more than a K-5. That camera will never diminish the quality of your work or anybody else.

Great music can be created banging sticks together and great art can be created drawing with burnt twigs.

[Sankon]

I'm not sure what you are getting at.

Yes, I know you're not, as evidenced by your repeated posts on this subject

I'm "getting at" there is more to dismissing a FF than your reduction to image quality comparisons at K5 resolutions. Would you repeat the methodology reducing a K5 image to the resolution of a phone camera and declare the K5 pointless for your purposes ? Would the advantages of a full field of view and bigger, brighter viewfinder show up in your comparison ? The K5 is a GREAT camera, it simply doesn't need this kind of "justification". Strikes me as just a bit needy. Good luck though in your dedication to reducing evaluation of cameras to equivalent-resolution IQ.

IMHO, as ever.

Paul