Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
07-10-2016, 11:37 PM   #121
Pentaxian




Join Date: Nov 2013
Posts: 4,626
QuoteOriginally posted by clackers Quote
You've a lot of learning to do before talking so frequently and copiously about these matters, Nicolas.


Their only measurements are the screen tab. They are the contents of your RAW file on your SD card, it is the Real World.


You do not understand the methodology of the print tab - it's a postprocessing 'what if', a hypothetical, and they only take into account numbers of pixels - never sensor format size.


My source is DxoMark themselves - what is yours?


Detailed computation of DxOMark Sensor normalization - DxOMark
We don't need another one, let's use your own source:

QuoteQuote:
As can be seen, high-resolution sensors will gain more SNR, DR, TR and CS when reduced to a lower reference resolution. For DxOMark Sensor Overall Score and Metrics, we chose a reference resolution equal to 8 Megapixels, which is a bit less than a 12" x 8" print with a 300dpi printer. However, any other resolution can be chosen, as doing so only shifts the normalized values by a constant (because the reference resolution appears only as a logarithm in the formulas above).
What should be remembered is that doubling the resolution adds:
  • 3dB to the normalized SNR
  • 0.5 bit to the normalized DR
  • 0.5 bit to the normalized TR
  • 1.5 bit to the normalized CS.
Also note that in DxO Labs’ Measurement Database, one can look either at normalized or absolute results, yet in the context of the DxOMark Sensor Overall Score, only normalized results are relevant. (Read more in our Insights section.)
The guys provide you both results and explains the need for normalization. A bit like you would not compare temperation in C and F directly. They basically explain why this is the case with a mathetical proof and also explain that whatever the actual resolution choosed as reference this will not change the results.

And you what you do? You use the screen results, the non normalized result as the reference. You basically compare temperature in C with temperature in F...

Comparing the screen score of K5-IIs and D810 without normalization mean that you'll systematically display/print the D810 image 2.3 time bigger and yet will not change your viewing distance toward it.

07-11-2016, 01:18 AM   #122
Pentaxian
newmikey's Avatar

Join Date: May 2007
Posts: 1,557
One would almost read this discussion and run, tail between legs, for the woods! All we are talking about here is the camera body here and I have absolutely zero issue with acknowledging that, under ideal circumstances, a FF body probably delivers superior IQ when compared to an APS-C body especially at the edges of either camera's performance envelope and under challenging circumstances or alternatively in an image lab.

To me, that is not a real life discussion to begin with, to be honest. The IQ endgame is not solely determined by the camera body but by the combination of body, available lenses, software (or proficiency in using it) and available budget. With unlimited budget comes the luxury of getting into these types of discussions. Real life however for most of us, is a constant balancing of available budget and shooting- as well as processing capabilities.

Am I better of with a K-1 if availing myself of the K-1 means I will have to "make do" with lenses that weren't specifically designed for FF digital bodies or with lenses which are a grade down of what would be ideal?

Would it be very bold if I'd say that a K-3 II with a DA17-70mm/f4 would almost certainly be indistinguishable to a K-1 with the DA28-105/f3.5-5.6? Equal noise levels derived from being able to open up a stop or so at the long end (thereby lowering the required ISO), equal DOF (or rather lack of it) for the same reason? Only at the very extremes of performance would one be able to show any visible difference in noise or DOF. For 95% of shots, the difference would be nil.

Would it be equally bold to say that, with the suitable lenses and inside the performance envelope for both types of bodies, the difference can be overcome easily so as to be entirely unnoticable?

The whole point of this discussion escapes me, I have no need to establish that a Ferrari i a better car than a Ford Fiesta if both are capable of getting me from Paris to Munich within minutes of each other, seeing the speed limits and weather circumstances en-route. The Ferrari having of course cost me a multiple of what the Ford Fiesta did and also having a gas mileage to match. The fact that I could take the Ferrari out on a stretch of autobahn once in a while or on a circuit and show that it is faster wouldn't do much as an argument to why either car is "better" or "superior" in terms of the current discussion.

The K-1 is an extremely capable camera with a great sensor but the sum of the bits & pieces in the photobag is what eventually makes up the difference. For many, the K-1 would be an extremely expensive "have-too" which does not visibly impact the quality of their output.

That said: go ahead and tear me to shreds!
07-11-2016, 01:38 AM   #123
Pentaxian
noelpolar's Avatar

Join Date: Oct 2013
Location: Goolwa, SA
Posts: 3,065
QuoteOriginally posted by clackers Quote
Well, I've learnt so much good stuff these last few years from yourself and all the other members that it's a shame to see misinformation :-)
I wasn't picking on you or anything.... just liked your statement in general!
07-11-2016, 02:08 AM   #124
Site Supporter




Join Date: Mar 2014
Photos: Gallery
Posts: 962
QuoteOriginally posted by newmikey Quote
One would almost read this discussion and run, tail between legs, for the woods! All we are talking about here is the camera body here and I have absolutely zero issue with acknowledging that, under ideal circumstances, a FF body probably delivers superior IQ when compared to an APS-C body especially at the edges of either camera's performance envelope and under challenging circumstances or alternatively in an image lab.

To me, that is not a real life discussion to begin with, to be honest. The IQ endgame is not solely determined by the camera body but by the combination of body, available lenses, software (or proficiency in using it) and available budget. With unlimited budget comes the luxury of getting into these types of discussions. Real life however for most of us, is a constant balancing of available budget and shooting- as well as processing capabilities.

Am I better of with a K-1 if availing myself of the K-1 means I will have to "make do" with lenses that weren't specifically designed for FF digital bodies or with lenses which are a grade down of what would be ideal?

Would it be very bold if I'd say that a K-3 II with a DA17-70mm/f4 would almost certainly be indistinguishable to a K-1 with the DA28-105/f3.5-5.6? Equal noise levels derived from being able to open up a stop or so at the long end (thereby lowering the required ISO), equal DOF (or rather lack of it) for the same reason? Only at the very extremes of performance would one be able to show any visible difference in noise or DOF. For 95% of shots, the difference would be nil.

Would it be equally bold to say that, with the suitable lenses and inside the performance envelope for both types of bodies, the difference can be overcome easily so as to be entirely unnoticable?

The whole point of this discussion escapes me, I have no need to establish that a Ferrari i a better car than a Ford Fiesta if both are capable of getting me from Paris to Munich within minutes of each other, seeing the speed limits and weather circumstances en-route. The Ferrari having of course cost me a multiple of what the Ford Fiesta did and also having a gas mileage to match. The fact that I could take the Ferrari out on a stretch of autobahn once in a while or on a circuit and show that it is faster wouldn't do much as an argument to why either car is "better" or "superior" in terms of the current discussion.

The K-1 is an extremely capable camera with a great sensor but the sum of the bits & pieces in the photobag is what eventually makes up the difference. For many, the K-1 would be an extremely expensive "have-too" which does not visibly impact the quality of their output.

That said: go ahead and tear me to shreds!
I agree with you, in my book it is all about the LENS. The camera is a disposable thing at the end of a good LENS!!!

07-11-2016, 11:37 AM   #125
Pentaxian




Join Date: Nov 2013
Posts: 4,626
QuoteOriginally posted by newmikey Quote
One would almost read this discussion and run, tail between legs, for the woods! All we are talking about here is the camera body here and I have absolutely zero issue with acknowledging that, under ideal circumstances, a FF body probably delivers superior IQ when compared to an APS-C body especially at the edges of either camera's performance envelope and under challenging circumstances or alternatively in an image lab.

To me, that is not a real life discussion to begin with, to be honest. The IQ endgame is not solely determined by the camera body but by the combination of body, available lenses, software (or proficiency in using it) and available budget. With unlimited budget comes the luxury of getting into these types of discussions. Real life however for most of us, is a constant balancing of available budget and shooting- as well as processing capabilities.

Am I better of with a K-1 if availing myself of the K-1 means I will have to "make do" with lenses that weren't specifically designed for FF digital bodies or with lenses which are a grade down of what would be ideal?

Would it be very bold if I'd say that a K-3 II with a DA17-70mm/f4 would almost certainly be indistinguishable to a K-1 with the DA28-105/f3.5-5.6? Equal noise levels derived from being able to open up a stop or so at the long end (thereby lowering the required ISO), equal DOF (or rather lack of it) for the same reason? Only at the very extremes of performance would one be able to show any visible difference in noise or DOF. For 95% of shots, the difference would be nil.

Would it be equally bold to say that, with the suitable lenses and inside the performance envelope for both types of bodies, the difference can be overcome easily so as to be entirely unnoticable?

The whole point of this discussion escapes me, I have no need to establish that a Ferrari i a better car than a Ford Fiesta if both are capable of getting me from Paris to Munich within minutes of each other, seeing the speed limits and weather circumstances en-route. The Ferrari having of course cost me a multiple of what the Ford Fiesta did and also having a gas mileage to match. The fact that I could take the Ferrari out on a stretch of autobahn once in a while or on a circuit and show that it is faster wouldn't do much as an argument to why either car is "better" or "superior" in terms of the current discussion.

The K-1 is an extremely capable camera with a great sensor but the sum of the bits & pieces in the photobag is what eventually makes up the difference. For many, the K-1 would be an extremely expensive "have-too" which does not visibly impact the quality of their output.

That said: go ahead and tear me to shreds!

Would say for high iso purpose the 28-105 on the K1 should be "only" equivalent to a 18-70 f/2-3.5 on APSC, but the actual better handling of high iso on K1 for noise mean that is more like a 18-50 f/1.7-f/2.5.

As for the money, this is complex.

I like FA ltd primes so FF or APSC, I'll spend a lot.

But if I was to take a used D700 + a 28-75 + 70-200 both lenses from tamron and everything used I'd pay less than 1500€. Make it a brand new D610 and that would be 1800€. Theses price are bound to drop in the future.

We already entered an era where the entry level FF are less expensive than the APSC flagship. The cheap f/2.8 being at the same price that made the cost the cost the same.

But basically for sharpness/high iso an f/2.8 zoom on an FF give you as good performance as the best prime on APSC. Maybe not on rendering, but for sharpness, noise, deph of field control, yes.

Even if you just take indoor, family snapshots, go to any indoor/night event, the difference in performance will be immediately visible. No need to compare numbers.

Last edited by Nicolas06; 07-11-2016 at 11:51 AM.
07-11-2016, 04:37 PM   #126
Loyal Site Supporter
clackers's Avatar

Join Date: Jul 2013
Location: Melbourne
Photos: Albums
Posts: 8,522
QuoteOriginally posted by Nicolas06 Quote

Comparing the screen score of K5-IIs and D810 without normalization mean that you'll systematically display/print the D810 image 2.3 time bigger and yet will not change your viewing distance toward it.
You are now off the topic.

The topic is that the FF D810 RAW files are just as noisy as the K-5 IIs RAW files, because their pixel size and technology is the same.

You are now talking about postprocessing. That is not due to sensor size, but the numbers of pixels. There are some phones that have more pixels than some FF cameras.

Cropping or magnifying an image will not change which pixels are noise and which are not.

For the Print tab, DxOMark arbitrarily decide to project (again, they do NOT measure) noise remaining after re-sampling, not resizing.

The resampling loses detail but reduces noise by averaging ... looks at three black pixels and one white, for instance, and replaces them with two black. When noise reduction goes too far, we know about it ... the people in the pictures suddenly look like wax dummies.

The APS-C K-3 downsamples equally well in the DXO 'normalizing' process as the full frame Nikon D750 - it gets about the same 5dB increase. This is of course troubling to the Full Frame Fetishists who worship the 'Print' tab.

Nicolas, you have said many times on this forum that APS-C may well be good enough for you for a long time to come - please don't change your mind and spend money based on incorrect assumptions.

Last edited by clackers; 07-11-2016 at 04:45 PM.
07-11-2016, 10:56 PM   #127
Pentaxian




Join Date: Nov 2013
Posts: 4,626
Clackers, This is a photo. The signal is 2 dimensionnal. Your assumption the screen mode is the one to compare (despire explicit explanation from your source that you should not) consider that your interrested to a single pixel rather than the whole image. This is not how people see a picture. This is not how the theory work neither. If you take the same portion of the image, there either rougly same number of quite larger pixels (A7s, D610, A7, D5) or significantly more pixels all with the same size (D810, A7R-II). In the end, this make the signal for this portion of the image stronger.

Our eyes staring at the whole picture can't see more than 7-8MP worth of detail anyway and even if the print is made to 40"x60", one will not see the difference between 16, 24 or 36MP looking at the picture as a whole. The eyes will do the averaging and if taken at say 6400 iso the picture will look better taken with a A7s, D610, D810 or A7R-II than with K5-IIs. Some FF sensor have actually smaller or bigger photosites, in the grand scheme of things they still look better.

The normalization make sense, it match the real world usage.

Even if you were looking from near distance, while at low iso the 36MP image would look noticably sharper, at high iso, the FF image would still look noticably less noisy.

There no scenario were the APSC is as good, only scenario when it is good enough. Take a picture at iso 800 or less, print large if you want but look at it as the whole... You'll not see any noticable difference.

The reason I am to stay with an APSC is not because FF is not better. FF is better, it is because this is simply not relevant to me. APSC is good enough and also lighter smaller.

But ultimately, give me a cheap mirorless FF in K-mount, with good enough AF (as good as K3 is enough for me), and give me autoresize of the crop area in the EVF, then I get best of both world. I can get my small light ltd FF lenses when it is conveniant, or use my teles in the APSC crop when I need reach.

We are not there yet, so FF still bigger/heavier/expensive and we still miss some key lenses for me like a small/light 24mm and a mirrorless body too. It enough for me to look at pictures of LeRolls for example to see subject and technique will make more difference anyway than sensor format.

Last edited by Nicolas06; 07-11-2016 at 11:23 PM.
07-11-2016, 11:00 PM - 1 Like   #128
Pentaxian




Join Date: Nov 2013
Posts: 4,626
What I just brought is... A FA31, not a camera... Can't wait for it to come from the states by the way !

07-12-2016, 06:25 AM   #129
Pentaxian




Join Date: Feb 2015
Photos: Albums
Posts: 3,238
QuoteOriginally posted by Nicolas06 Quote
The normalization make sense, it match the real world usage.
Generally speaking, it works, but the theory assume that noise is signal independent, evenly distributed. According to DXO, the A7s downsampled 8M print is about as good as D810 downsampled 8M print, but I've compared A7s high iso normalized image with D810 high iso image and for some reasons (nature of noise shot noise instead of johnson noise, better dark current control on A7s?), the A7s image was cleaner. I'd have expected the D810 image would be cleaner due to averaging of the read noise. I suppose the A7s sensor cells are larger and allow more design freedom for the designers, that could explain the difference.

QuoteOriginally posted by Nicolas06 Quote
What I just brought is... A FA31, not a camera... Can't wait for it to come from the states by the way !
FA31... wow, that's a good one.
07-12-2016, 06:35 AM - 3 Likes   #130
Senior Member
macman24054's Avatar

Join Date: Aug 2013
Location: Axton, VA
Photos: Gallery | Albums
Posts: 229
i think the problem here is the big mistake of thinking a camera body with the latest and greatest specs is the answer to being a great photographer. That could not be farther from the truth. Photography is a combination of many things. Camera body, lens (more important than body), camera settings, lighting (more important than body), person behind the camera (most important factor), proficiency in post production (as important as body), knowledge of equipment and a few things I am sure I overlooked. With so many factors involved, exact reproduction of a given photo is almost impossible to happen taking photos in the field. In a scientific situation we can replicate. The last time i checked, photography is about art and ones vision of that moment in time. I do not get paid to photograph weddings and events because of the equipment I have. Not once has a client asked me what equipment I use. It is because I can capture the important moments and take decent artistic portraits with the style and vision my clients like. I do not photograph every wedding in my area. Some people are looking for a different style and vision of their wedding day photos.
07-12-2016, 01:03 PM   #131
Pentaxian




Join Date: Nov 2013
Posts: 4,626
QuoteOriginally posted by biz-engineer Quote
Generally speaking, it works, but the theory assume that noise is signal independent, evenly distributed. According to DXO, the A7s downsampled 8M print is about as good as D810 downsampled 8M print, but I've compared A7s high iso normalized image with D810 high iso image and for some reasons (nature of noise shot noise instead of johnson noise, better dark current control on A7s?), the A7s image was cleaner. I'd have expected the D810 image would be cleaner due to averaging of the read noise. I suppose the A7s sensor cells are larger and allow more design freedom for the designers, that could explain the difference.

FA31... wow, that's a good one.
This first and foremost assume the technology is the same. And it is not. A7S sensor has different design than D810. Sony explain it. D810 is optimized for low iso quality with even a iso 50 mode. A7s is optimized for high iso...

And it is correctly show by print scores on DxO, it not like you would not see the difference in the DxO print graphs.

There 1EV difference for SnR at high iso and 2EV difference for dynamic range. The gain is also visible for tonal range and color sensitivity.

The properties that allow the noise to be reduced by merging small photosites is the same properties that allow a bigger photosite to have lower noise. A bigger photosite just hope the signal is the same all over the (bigger) surface of the photosite and average the values.

But there more to a photosize than just pure size. There the quantity of signal you can store, how much time and with how much precision the information can be stored. There the quality of the colors filters, the design of the micro lenses, the low pass filter (there one on A7s), the percentage of the surface really used to grab light and the part used for circuitry and other stuff like that.

It is true that the bigger the photosites, the less citcuitery you have, and the easier it is to make efficiant micro lenses and all. On the contrary, the smaller, the more likely you are to find yourself impacted by current leaks. There many factors.

It is true that the noise is not uniformely random. That impact the bigger photosites significantly: they can't reproduce the fine details, are more prone to moire and have more need for a low pass filter as a mitigation that futher decrease sharpness.


It is easier

Last edited by Nicolas06; 01-31-2017 at 02:03 PM.
07-13-2016, 12:33 AM   #132
Loyal Site Supporter
clackers's Avatar

Join Date: Jul 2013
Location: Melbourne
Photos: Albums
Posts: 8,522
QuoteOriginally posted by Nicolas06 Quote
Our eyes staring at the whole picture can't see more than 7-8MP worth of detail anyway
Again, we're not talking about the level of detail, or viewing distances, or resolution, we're talking about noise.

Norm Head was right, that when you are incorrect on some fact, you start talking about something else - for many posts - that was never in dispute.

If one in twenty pixels in a D810 RAW file is noise, it's also one in twenty pixels in a K5-II file.

That's what DxOMark measured, and reported in the so-called Screen tab. It is actually what you get when you open the files up in Lightroom or Photoshop or whatever. The noise performance is virtually identical.

The Print tab is *not* the real world. They never measured the noise again after downsampling, they just made some projection from the Screen tab. They would estimate my Samsung Galaxy S6 phone camera would downsample the same as a flagship FF Nikon D4 according to them, because they both have 16Mp.


QuoteOriginally posted by Nicolas06 Quote
subject and technique will make more difference anyway than sensor format.
You have (as usual) changed the topic again, but ... of course I agree totally with what you just said.

Your K-3 and FA31 together will be a fantastic combination.

Last edited by clackers; 07-13-2016 at 12:42 AM.
07-13-2016, 12:49 AM   #133
Loyal Site Supporter
clackers's Avatar

Join Date: Jul 2013
Location: Melbourne
Photos: Albums
Posts: 8,522
QuoteOriginally posted by biz-engineer Quote
Generally speaking, it works, but the theory assume that noise is signal independent, evenly distributed. According to DXO, the A7s downsampled 8M print is about as good as D810 downsampled 8M print, but I've compared A7s high iso normalized image with D810 high iso image and for some reasons (nature of noise shot noise instead of johnson noise, better dark current control on A7s?), the A7s image was cleaner.
That's exactly right, Biz-Engineer.

Have a look at the Screen tab, not Nicolas' misleading Print tab estimate, and it's clear to see the 5dB difference thanks to the pixel size and better technology.

Lesson? Have a low amount of noise in your original image, and you don't have to do much post processing, which of course destroys detail too.
07-13-2016, 05:22 AM   #134
Pentaxian




Join Date: Nov 2013
Posts: 4,626
QuoteOriginally posted by clackers Quote
Again, we're not talking about the level of detail, or viewing distances, or resolution, we're talking about noise.
Man you care of 1 single photosite, you theory say if a sensor has a given single pixel of the whole sensor surface, it will be wonderfull very high quality. You signal would be like 110db and not 43, but nobody would ever interrested. You'd say look the quality is much betterbBecause your definition of the signal would be perfectly uniform color and no more.

This is not like that. If you use the K5-IIs vs K1 with proper lenses at same apperture and look at the star, its the K1 that will be able to actually detect the dimmest star, not the K5-IIs. At the limit, some feint data would not be only considered as noise or filtered by the color filter because it hit only one photosite while on the FF the information would go through 2 photosite and the algorithmss will be able to something out of it. You find a brigher point on the final image where the APSC would have only noise.

Last edited by Nicolas06; 07-13-2016 at 05:30 AM.
07-13-2016, 10:55 AM   #135
Pentaxian




Join Date: Feb 2015
Photos: Albums
Posts: 3,238
QuoteOriginally posted by Nicolas06 Quote
Man you care of 1 single photosite, you theory say if a sensor has a given single pixel of the whole sensor surface, it will be wonderfull very high quality. You signal would be like 110db and not 43, but nobody would ever interrested. You'd say look the quality is much betterbBecause your definition of the signal would be perfectly uniform color and no more. This is not like that. If you use the K5-IIs vs K1 with proper lenses at same apperture and look at the star, its the K1 that will be able to actually detect the dimmest star, not the K5-IIs. At the limit, some feint data would not be only considered as noise or filtered by the color filter because it hit only one photosite while on the FF the information would go through 2 photosite and the algorithmss will be able to something out of it. You find a brigher point on the final image where the APSC would have only noise.
I like how you turn things around. Nicolas, the only different of view point between you and clackers is that you use a theoretical model that generally works, and Clackers says that general models that generally work may fail in particular.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
aperture, aps-c, background, blur, camera, dof, dslr, feet, ff, ff over aps-c, field, focus, frame, full frame, full-frame, k-1, k1, lens, paper, pentax k-1, people, picture, posters, stitching
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
FF vs APS-C Field of View revisited Ole Pentax K-1 2 05-07-2016 02:13 PM
Confused about Angle of View of Lenses on FF vs APS-C? Kath Pentax SLR Lens Discussion 3 10-01-2015 09:55 AM
Does FF vs APS-C affect amount of light? windhorse General Photography 46 03-02-2015 07:07 PM
Quick question regarding field of view - FF vs APS-C glass? Julie Pentax SLR Lens Discussion 5 12-23-2012 05:33 PM
APS-C does not increase focal length over FF, it decreases field of view. TomTextura Photographic Technique 135 06-09-2012 04:58 PM



All times are GMT -7. The time now is 03:37 PM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top