Luminous Landscape Forum
Site & Board Matters => About This Site => Topic started by: theguywitha645d on March 15, 2012, 09:13:48 pm
-
Great article Michael.
-
The problem with writing anything on the web is that there's always some smart guy eager to belittle by pointing out an exception.
The 645D is a $10,000 camera. The D800 is a $3,000 camera. What part of it's much more expensive isn't clear?
As it is I find myself adding more caveats and explanations than I'd prefer for clear communication, but apparently not as much as I should. Do I have to start adding footnotes with exceptions and caveats? :(
Michael
-
Michael, what strikes me most about the D800 is that to get the most out of it, hand held photography needs to be at 1 over either 2 or 3 times the focal length. Given that the pixel density of the NEX 7 is higher and I believe that it lacks sensor based image stabilisation, has this presented any real world problems when using the NEX 7?
And how does the weight of the NEX 7 system add or detract from that?
Whilst there has been a lot written about "We don't need more megapixels than X", in an age of 100% pixel peeping on the Internet, I can't help but wonder if we'll run into a megapixel limit that's lower than we might have expected because the camera becomes too unusable at full resolution. And by that I mean that we'll need to use 1/100 with a 50mm/1.4 (no IS) (for example.)
What's curious about the released images thus far is that whilst the dynamic range at low ISO (100 - 200) seems to have been increased dramatically, there has been a much smaller improvement over ISO 200. With the propensity for image blur due to smaller pixels, it would seem that the faster shutter speeds are going to drive up the need to use higher ISO values.
I'm curious as to where the cutoff points are here. For example, are you better off using a D700 indoors because you can more easily (and reliably) get sharp pictures at ISO 400 with 1/50 with a 50mm than you are using a D800 that requires ISO 800 (or more) so that you can shoot at 1/100 (or faster) with the same lens?
Or is the future is one where either all of the lens/camera combinations that we use involve some sort of stabilisation technology in order to make the camera more usable?
-
I'm curious as to where the cutoff points are here. For example, are you better off using a D700 indoors because you can more easily (and reliably) get sharp pictures at ISO 400 with 1/50 with a 50mm than you are using a D800 that requires ISO 800 (or more) so that you can shoot at 1/100 (or faster) with the same lens?
All things being equal, more pixels will never deliver less absolute details. In the very worst case, the D800 would not deliver any additional details over the lower resolution D700. In most cases, there will be some advantage. Indeed, individual pixels may look more blurred, but the image will look identical.
Now, all things are not being equal because the high ISO behavior of high density sensors may be worse. In the present case though, it seems that after down res to 12 mp, the D800 is about one stop better than the D700/D3 at high ISOs, which puts it roughly at the level as or slightly below the D3s in terms of high ISO noise capability.
This looks like the closest thing to the universal camera. MF like DR/detail at low ISO, D3s like image quality at high ISOs...
Cheers,
Bernard
-
Great article Michael.
Good answer! never be the nail that is standing up you might get hit on the head with a hammer! :)
Marc
-
I know I'll get jumped on for this, but I think removing a well-designed a.a. filter rarely results in much more real detail. Instead, it allows the generation of false detail beyond the nyquist limit, which gives the impression of sharpness. Aliasing looks very sharp. And at the resolutions we use now, nobody can tell the difference in many landscape images. But that "extra" detail just may not be in the original scene at all. This article explains the issue quite well, I think: http://www.dvxuser.com/articles/article.php/20.
Right now, even the highest resolution bayer sensors fail to fully exploit the abilities of decent lenses. Ctein has pointed out that it will take sensors with resolution in the hundreds of megapixels to actually accomplish that. In the meantime, there are two strategies people use to get the look of ultimate sharpness--both of them relying on digital artifacts and processing.
One is to remove the a.a. filter, and accept some aliased edges and false details that look sharp. These artifacts are often indistinguishable from actual detail in, say, landscape photography. The second strategy is to cut off most detail beyond nyquist with an a.a. filter, and use advanced sharpening techniques that try to recreate what the lens saw. If pushed far enough, deconvolution sharpening not only restores sharpness, but introduces a mist of grain and other digital artifacts that make the file look super sharp. (Just being honest, here.)
Personally, I prefer the second strategy. Deconvolution sharpening gets better all the time. And I would rather have control over the process than accept whatever artifacts the sensor gives me. Not just moire, but distorted edges and other inventions of the sensor and camera processing.
Pretty much every digital file made without an a.a. filter has aliasing, baked in. Just check the resolution charts of your favorite lenses. They all show false color and false detail--even those taken on cameras with a weak a.a. filter. It's fine if you like how the file looks, but we shouldn't confuse this aliasing with actual resolution.
I think that in ten years people will look back on the whole a.a. debate as a quaint vestige of digital growing pains. In the meantime, we should keep it real. Digital cameras without a.a. filters record a lot of false data. If we want to incorporate that into the look of our images, fine. But it's still there. For the most part non-a.a. cameras don't reveal more resolution. They create it.
Now I'll go hide in my bunker to wait for the incoming flames...
-
Everyone seems to know Michael wrote this article but me. Am I missing something that's on the screen, or do we just know that if another writer isn't mentioned it's Michael writing?
Nice article, an organized synopsis useful to those who haven't yet made their decision.
-
I know I'll get jumped on for this, but I think removing a well-designed a.a. filter rarely results in much more real detail. Instead, it allows the generation of false detail beyond the nyquist limit, which gives the impression of sharpness. Aliasing looks very sharp. And at the resolutions we use now, nobody can tell the difference in many landscape images. But that "extra" detail just may not be in the original scene at all. This article explains the issue quite well, I think: http://www.dvxuser.com/articles/article.php/20.
Right now, even the highest resolution bayer sensors fail to fully exploit the abilities of decent lenses. Ctein has pointed out that it will take sensors with resolution in the hundreds of megapixels to actually accomplish that. In the meantime, there are two strategies people use to get the look of ultimate sharpness--both of them relying on digital artifacts and processing.
One is to remove the a.a. filter, and accept some aliased edges and false details that look sharp. These artifacts are often indistinguishable from actual detail in, say, landscape photography. The second strategy is to cut off most detail beyond nyquist with an a.a. filter, and use advanced sharpening techniques that try to recreate what the lens saw. If pushed far enough, deconvolution sharpening not only restores sharpness, but introduces a mist of grain and other digital artifacts that make the file look super sharp. (Just being honest, here.)
Personally, I prefer the second strategy. Deconvolution sharpening gets better all the time. And I would rather have control over the process than accept whatever artifacts the sensor gives me. Not just moire, but distorted edges and other inventions of the sensor and camera processing.
Pretty much every digital file made without an a.a. filter has aliasing, baked in. Just check the resolution charts of your favorite lenses. They all show false color and false detail--even those taken on cameras with a weak a.a. filter. It's fine if you like how the file looks, but we shouldn't confuse this aliasing with actual resolution.
I think that in ten years people will look back on the whole a.a. debate as a quaint vestige of digital growing pains. In the meantime, we should keep it real. Digital cameras without a.a. filters record a lot of false data. If we want to incorporate that into the look of our images, fine. But it's still there. For the most part non-a.a. cameras don't reveal more resolution. They create it.
Now I'll go hide in my bunker to wait for the incoming flames...
I've tried to get my head around this in other threads but my understanding is false information is only present if the subject is a repeating pattern and a high frequency pattern? It is not present all the time ie random patterns. Please correct me if I'm wrong.
Marc
-
This is an interesting issue that does need resolution (pardon the pun).
Looking forward to the debate.
Regards
Tony Jay
-
Well, there are people on this forum who know a lot more about the science than I do. But my understanding is that this kind of aliasing is always present. It just becomes more obvious when there are repeating patterns over a large area of the photograph that interact with the sensor to create a large moire pattern. Or hot highlights that stress the sensor into creating obvious false color. In a random landscape scene, we would have to look more closely to see it expressed--say, on the edge of branches or telephone lines, in wood detail, in color speckles and jaggies and thickened details.
-
Well, there are people on this forum who know a lot more about the science than I do. But my understanding is that this kind of aliasing is always present. It just becomes more obvious when there are repeating patterns over a large area of the photograph that interact with the sensor to create a large moire pattern. Or hot highlights that stress the sensor into creating obvious false color. In a random landscape scene, we would have to look more closely to see it expressed--say, on the edge of branches or telephone lines, in wood detail, in color speckles and jaggies and thickened details.
I'm under the opposite understanding so let me start another thread on this subject in Cameras, lenses etc
let's get the smart guys here to clarify this important issue
Marc
-
Hi,
No, you always get fake detail and or fake contrast. Repeating patterns just makes this obvious.
Best regards
Erik
I've tried to get my head around this in other threads but my understanding is false information is only present if the subject is a repeating pattern and a high frequency pattern? It is not present all the time ie random patterns. Please correct me if I'm wrong.
Marc
-
I agree with Bernard here. For same size images or prints, any increase in sensor pixel numbers will never deliver worse results using the same techniques.
However, if a sensor has a significant increase in pixel-count, then to realize the full potential of the increased sharpness and detail that such a sensor is capable of, one may have to use a tripod, or increase shutter speeds for hand-held shots.
The 1/FL rule (or guide) for shutter speed without a tripod, used to apply to 35mm film for a reasonably sharp 8"x10" print. If one considers such a guide useful and valid, then it would also apply to the D800 in the same circumstances, provided the resulting image or print were also 8"x10" (or 8x12).
However, those who buy the D800 will likely want to make prints considerably larger than 8"x12", or will want to use the cropping potential of the high pixel numbers to provide an effectively longer focal length, in which case the 1/FL guide becomes more like 1/3FL or 1/4FL in the absense of VR.
I'm sure glad I have one good Nikkor lens with VR, the 24-120/F4. The 14-24/2.8 lacks VR but that's not really a problem. 1/4FL at 14mm is still only 1/60th, although I've seen it reported that at very short focal lengths the 1/FL guide breaks down. But I'm not sure about this. Maybe I should do some test comparisons ;D .
-
Quote
The problem with writing anything on the web is that there's always some smart guy eager to belittle by pointing out an exception.
Unquote
If it is logic they are trying to use to contradict you then that is reasonable and expected. But if they are merely trying to wind you up then there isn't a chance in hell of winning unless it is face to face. If it is on the internet then, if possible, you ignore them. Ironically all of us have probably indulged in the art of wind ups? :)
-
I recently got a 24" printer and now am starting to see my images different than before. I now am happy to have used MF film for my serious shooting. I am not totally sure how much real resolution my (now sold for a Mamiya 7ii) Mamiya Press had - shoots of a sector star sugggested, that I was near the resolution limit of my scanner (with appropriate high res film). That'd be about 100 Megapixels (I guess real resolution is much less, but still great). The corners were not that good. Old glass, cheap.
Now that I have a Mamiya 7ii which much better glass (43, 80 and 150 mm) I hope to get something nearer to the scanner limit.
Cost:
1. A whole shitload of work (shooting, development, scanning, etc).
2. Film and development cost
3. Exchangeable sensor camera with 3 lenses: 3500.- €
4. Scanner: I think it was about 3000.- €
Why do I tell that ?
For an amateur with low production and limited budget constraints who wants to print big MF film still is maybe the best way to get awesome technical quality if one likes the process and is willing to deal with the hassle(sheet film even better, but with even more hassle). For day to day images I use my S95 which lives in the pocket of my trousers.
If you need fast production, less hassle, convenience and more multi purpose applicability, of course digital definitely is the way to go.
But I know pros (architecture) who still shoot 4x5" film for a living and who are happy with that.
Whats this english proverb again - "Horses for courses" or so ....?
I'm curious to see the first shootouts between MF drum scans of a Mamiya 7 and the new Nikons.
-
That ship sailed a long time ago.
http://www.luminous-landscape.com/essays/back-testing.shtml (http://www.luminous-landscape.com/essays/back-testing.shtml)
I haven't shot film in a long time but I have a huge library. Occasionally I am asked to teach scanning to someone and so I dust off my Imacon Flextight and do some scans. When I do I am always bowled-over by how poor medium format film performs compared to contemporary digital. 35mm? Not even worth discussing.
Too bad. I sometimes miss the darkroom.
Michael
-
If I hadn't infrared dust removal in scanning, I think I'd stop using film, there'd be a borderline of trouble for me. I just looked at the sector star image I took with the old Mamiya Press. 96 sectors = 48 cycles up to a diameter of 57 Pixels from a 4000 DPI LS-9000 scan ... depending how exactly I judge it I come to something like about 25 effective megapixels for a 6x9 cm negative, but these are good pixels then. And this was a camera I had already 20 years ago. All my images from that time will transform to 25 good megapixels today. I'm curious how the 24" printing will go in the next months with this old stuff and I'm looking forward to it.
The Mamiya 7ii I have now is better (sharper):These guys (http://www.onlandscape.co.uk/2011/12/big-camera-comparison/) claim something like 50 MP for the Mamiya 7 kit they used with a 4000 DPI scan. They claim 80 MP with a 8000 DPI scan. With a microscope (not really practical for day to day use though) they achive something over 100 MP. Whatever we believe in - If I wanted something near that today (non stitched) in the digital world, I'd have to pay a real lot of money.
If I had some 50-60 k Euro to waste, maybe I'd jump on to the high end digital train, but as a non pro -and this was my constraint- I'll wait a bit more. These 2 professional guys who still use 4x5" film I learned to know the last months (one an architecture photographer, the other doing landscape and calendars) really gave me something to think about.
I'm not against digital at all. Actually I don't have darkroom anymore since the digital processing allows things I could not even dream of in the past.
For me, I think it will not take too much time (maybe some more years - no idea) until the cost/quality ratio for high end digital is at a point where I'll change.
But not yet now.
With all due respect
Cheers
~Chris
-
Great article,
One point of contention:
Think about it. If there was no noticeable difference would Nikon have gone to the bother of creating a separate product? I think not.
The simple answer is that Nikon would have created a separate product if they believe people will pay more money irregardless of if it actually has better quality.
That being said, I expect the D800E to have better resolution.
We will soon find out.
Jonathan
-
I know I'll get jumped on for this, but I think removing a well-designed a.a. filter rarely results in much more real detail. Instead, it allows the generation of false detail beyond the nyquist limit, which gives the impression of sharpness. Aliasing looks very sharp. And at the resolutions we use now, nobody can tell the difference in many landscape images. But that "extra" detail just may not be in the original scene at all. This article explains the issue quite well, I think: http://www.dvxuser.com/articles/article.php/20.
Right now, even the highest resolution bayer sensors fail to fully exploit the abilities of decent lenses. Ctein has pointed out that it will take sensors with resolution in the hundreds of megapixels to actually accomplish that. In the meantime, there are two strategies people use to get the look of ultimate sharpness--both of them relying on digital artifacts and processing.
One is to remove the a.a. filter, and accept some aliased edges and false details that look sharp. These artifacts are often indistinguishable from actual detail in, say, landscape photography. The second strategy is to cut off most detail beyond nyquist with an a.a. filter, and use advanced sharpening techniques that try to recreate what the lens saw. If pushed far enough, deconvolution sharpening not only restores sharpness, but introduces a mist of grain and other digital artifacts that make the file look super sharp. (Just being honest, here.)
Personally, I prefer the second strategy. Deconvolution sharpening gets better all the time. And I would rather have control over the process than accept whatever artifacts the sensor gives me. Not just moire, but distorted edges and other inventions of the sensor and camera processing.
Pretty much every digital file made without an a.a. filter has aliasing, baked in. Just check the resolution charts of your favorite lenses. They all show false color and false detail--even those taken on cameras with a weak a.a. filter. It's fine if you like how the file looks, but we shouldn't confuse this aliasing with actual resolution.
I think that in ten years people will look back on the whole a.a. debate as a quaint vestige of digital growing pains. In the meantime, we should keep it real. Digital cameras without a.a. filters record a lot of false data. If we want to incorporate that into the look of our images, fine. But it's still there. For the most part non-a.a. cameras don't reveal more resolution. They create it.
Now I'll go hide in my bunker to wait for the incoming flames...
No reason to hide, you're pretty much right on the money.
-
Digital cameras without a.a. filters record a lot of false data. If we want to incorporate that into the look of our images, fine.
When I create a photo, I don't record data, I try to bring into existence an impression of my experience. That's the difference between art and science.
Quite often the artistically challenged forget that, and obsess about the faults of their gear, rather than their own lack of imagination. :o
-
"The simple answer is that Nikon would have created a separate product if they believe people will pay more money irregardless of if it actually has better quality."
I think that your assumption is based on the vanity factor, such as a gold plated thingamajig to hang around your neck, not on a perceived desire for quality. $300 dollars is not a lot separating the two models so I would eliminate the vanity factor. A company like NIKON frequently makes decisions based on what they perceive the customer wants, not vanity, and there us the pride factor in producing a possible breakthrough product. Face is still big in Japan even though Seppuku is out of fashion.
It definitely be an interesting summer watching the DSLR and MF crowds duke it out. Science doesn't always eliminate stubborn pride and desire.
I've always wanted one of those status symbol Leica's with the built in quality to boot. My boss won't let me buy everything my heart desires.
-
"[...] My boss won't let me buy everything my heart desires.
Stupid boss ... :P
-
Can I quote you? Your safe since she can't user her teacher look on you ::) ::).
-
What should I say?
I'm self-employed.
Draw your conclusions ... ???
-
Hi,
If you like the result and the film approach works for you than it is just fine.
On the other hand I'd say that Michael is also right. Much depends on weather you like the film looks and if you are really good at driving your scanner.
Tim Parkins article is interesting in that it shows how good results you can get from film with adequate processing. It also seems that the Mamiya 7 you have is a very competent camera, having very good lenses.
I have done a lot of testing of film versus digital and I essentially share Michaels view, but I don't miss the dark room, especially not the chemicals.
Some of my experiment are summarized here http://echophoto.dnsalias.net/ekr/index.php/photoarticles/59-sony-alpha-900-vs-67-analogue-round-2 and here http://echophoto.dnsalias.net/ekr/index.php/photoarticles/16-pentax67velvia-vs-sony-alpha-900
Best regards
Erik
If I hadn't infrared dust removal in scanning, I think I'd stop using film, there'd be a borderline of trouble for me. I just looked at the sector star image I took with the old Mamiya Press. 96 sectors = 48 cycles up to a diameter of 57 Pixels from a 4000 DPI LS-9000 scan ... depending how exactly I judge it I come to something like about 25 effective megapixels for a 6x9 cm negative, but these are good pixels then. And this was a camera I had already 20 years ago. All my images from that time will transform to 25 good megapixels today. I'm curious how the 24" printing will go in the next months with this old stuff and I'm looking forward to it.
The Mamiya 7ii I have now is better (sharper):These guys (http://www.onlandscape.co.uk/2011/12/big-camera-comparison/) claim something like 50 MP for the Mamiya 7 kit they used with a 4000 DPI scan. They claim 80 MP with a 8000 DPI scan. With a microscope (not really practical for day to day use though) they achive something over 100 MP. Whatever we believe in - If I wanted something near that today (non stitched) in the digital world, I'd have to pay a real lot of money.
If I had some 50-60 k Euro to waste, maybe I'd jump on to the high end digital train, but as a non pro -and this was my constraint- I'll wait a bit more. These 2 professional guys who still use 4x5" film I learned to know the last months (one an architecture photographer, the other doing landscape and calendars) really gave me something to think about.
I'm not against digital at all. Actually I don't have darkroom anymore since the digital processing allows things I could not even dream of in the past.
For me, I think it will not take too much time (maybe some more years - no idea) until the cost/quality ratio for high end digital is at a point where I'll change.
But not yet now.
With all due respect
Cheers
~Chris
-
I really don't want to start a principal debate on film vs. digital and I don't remember myself saying Michael was wrong at some point.
For me it is clear the days of film are counted, apart from very few niches, like Majiks thread on the Inuit life.
That is pretty clear.
But there still is that not so small niche where a certain amount of high quality can be achived by film with much less investment than with digital, especially if one already has the anolog MF equipment. (Actually I'm working hybrid).
What would I have to pay for a 50 MP system with 3 lenses again ?
I'm sure there will be the day when I switch, because of the simple constraints of the cost/quality ratio.
Digital quality is getting cheaper rapidly. Really fast.
I also believe there is still some time left - but the clock is ticking.
If I hadn't had the Mamiya Press when I returned to photography I most likely would have bought a DSLR and kept craving for a MFDB.
I would have had the feeling of a bad compromise.
With MF film and the equipment I already have right now I'm in a pretty good situation.
If I had to start from scratch today I definitely would go digital, but I didn't start from scratch two years ago and I had old stuff to scan (still not finished doing this).
Honestly - I'm curious to meet the day when I will sell my Mamiya 7ii and my LS-9000.
And I'm even more curious what the equipment will be then.
But the day is not yet today.
Cheers
~Chris
-
Re: D800 hyperbole
« Reply #19 on: March 16, 2012, 03:10:56 PM » Reply Quote
--------------------------------------------------------------------------------
from: dds on March 15, 2012, 11:51:46 PM
Digital cameras without a.a. filters record a lot of false data. If we want to incorporate that into the look of our images, fine.
When I create a photo, I don't record data, I try to bring into existence an impression of my experience. That's the difference between art and science.
Quite often the artistically challenged forget that, and obsess about the faults of their gear, rather than their own lack of imagination. :o
Good answer and to the point of photography for most of us. It is worth mentioning that the AA filter removes real data, which is in a sense, creating false data. Which "falseness" do you prefer is the real question.
Tom
-
It is worth mentioning that the AA filter removes real data, which is in a sense, creating false data. Which "falseness" do you prefer is the real question.
Well ... it is also worth mentioning that it is usually easier to reverse the blurring of the image caused by an AA filter than it is to remove the artifacts caused by detail beyond the sensor's limit ...
I'm gonna get the D800, not the D800e.
-
Everyone seems to know Michael wrote this article but me. Am I missing something that's on the screen, or do we just know that if another writer isn't mentioned it's Michael writing?
Nice article, an organized synopsis useful to those who haven't yet made their decision.
Bump. An answer would be appreciated.
-
The style is recognizable and
"This coincides with my return to Toronto from Mexico"
confirms it.
-
Well ... it is also worth mentioning that it is usually easier to reverse the blurring of the image caused by an AA filter than it is to remove the artifacts caused by detail beyond the sensor's limit ...
I'm gonna get the D800, not the D800e.
remember with the 800E you can shoot at f8 creating an AA filter or f5 and no AA filter your decision. With the 800 you have no choice........ although I like the idea of vibrating the sensor as an AA filter that way you can turn it on or off at your discretion
Marc
-
Well ... it is also worth mentioning that it is usually easier to reverse the blurring of the image caused by an AA filter than it is to remove the artifacts caused by detail beyond the sensor's limit ...
I'm gonna get the D800, not the D800e.
That’s true, but my point is that both the presence or absence of an AA filter introduces artifacts or false information, you simply have to choose which.
-
The style is recognizable and
"This coincides with my return to Toronto from Mexico"
confirms it.
This is in response to my question?
Let me be more clear. I'm not arguing if Michael wrote this piece or that his style isn't recognizable. In any case, you usually don't have to do any investigative work to determine the author of an article.
What I'm asking is am I missing something that would normally be there. Like a place where authors names usually go that I'm just missing? Or maybe it's just a habit where if he doesn't list another author, then he just assumes we'll know it's him? Either way I'm fine with, it's the clarification I'm after. I'd rather not guess.
-
Unless an article on this site is attributed to someone else, I'm the guilty party.
Michael
-
Quote
The problem with writing anything on the web is that there's always some smart guy eager to belittle by pointing out an exception.
Unquote
If it is logic they are trying to use to contradict you then that is reasonable and expected. But if they are merely trying to wind you up then there isn't a chance in hell of winning unless it is face to face. If it is on the internet then, if possible, you ignore them. Ironically all of us have probably indulged in the art of wind ups? :)
I actually think that this is a good thing about the 'net. No longer have journalists, authors and officials the monopoly on the "one, right truth". People will always critizise what you write. Some critique may be perceived as irrelevant, naiive, etc, but I think one must have a fairly big ego to think that one can never learn from critique.
Part of this democratization is that it is becoming harder to judge what is a fair and balanced comment based on "best practice", "references" or "experience", and what is not. My guess (hope?) is that our children will be a lot better than ourselves at judging this than we will ever be.
So what if a guy found an exception in your text. Hopefully the text will still have value for its target readers. Writing "bullet proof" is , as you say, very hard, and tends to make the text more similar to legal contracts than anything anyone would willingly read in their spare time.
-h
-
I wish Michael would simply go back to writing subjective experience articles instead of trying-to-be-as-objective-as-humanly-possible-and-still-be-condemned-by-the-iTroll-community… Trying to be objective equates to trying to be boring. I like to read about his enthusiasm, in fact, I usually look for the enthusiasm in any story, no matter if it's appropriate enthusiasm or misplaced.
Till now, unless one has been able to outlay $25,000 to $50,000 for an MF back, camera and lenses, this type of resolving power has not be accessible.
Whatever happened to renting a MF back for a fraction of the cost?
Keep in mind that a 36 Megapixel camera isn't for everyone. If you just shot for the web and electronic media; if you rarely make prints larger than Super-A3 (13X19"), if you don't own high-end glass and know how to get the most from it, then neither the D800 nor the D800e are going to make a visible difference to your photography – at least not in terms of their higher resolving power.
Uh oh, that sort of flies right into the face of the "everything matters" article! But I don't care, because I understand the gist of what he's trying to say.
-
with the 800E you can shoot at f8 creating an AA filter or f5 and no AA filter your decision. With the 800 you have no choice
On the other hand, with the D800 you do have the choice of using f/5 or even f/2.8 for shallow DOF while putting focus on a subject like a tiled roof or article of clothing and yet not suffer from moiré (or less of it, anyway). With the D800E, you do not have that particular choice.
As you suggest, this is a choice between two ”slightly imperfect but pretty darned close" approximations of the image striking the sensor, and people will have to choose which set of imperfections they would rather have to deal with. I for one completely understand why Nikon expects the majority to go with the option of "less moiré, slightly more sharpening sometimes required".
-
As you suggest, this is a choice between two ”slightly imperfect but pretty darned close" approximations of the image striking the sensor,
A significant difference is that an aa-filter will only (negatively) affect very high spatial frequencies (close to the Nyquist frequency). For a 36MP camera, those are really minute details.
Not having an aa-filter can affect any and all spatial frequencies, meaning that even for a 36 MP camera and moderate print sizes/viewing distances the errors can sometimes be seen.
-h
-
Personally, I think it was perfectly stated as written, stating facts and offering solid advice, leaving nothing really to be argued about. However I believe the debate that will continue to rage will be the D800 v (comparable) MF DB at the $3000 entry point.
FWIW, I have the AA D800 on order, because my need for it is primarily people (wearing clothes ;)) in a non-studio environment -- and I believe it will be a superior tool to my MF kit for that purpose. But in no way do I see it replacing my MF kit...
Cheers,
-
I wish Michael would simply go back to writing subjective experience articles instead of trying-to-be-as-objective-as-humanly-possible-and-still-be-condemned-by-the-iTroll-community… Trying to be objective equates to trying to be boring...
Amen, brother!
-
There are some test pictures here http://www.fotopolis.pl/index.php?gora=1&lewa=2
Use Google translate and find the D800 and D800E and the Leica S2 and compare. You can download RAW files.
Here is one screen shot from Lightroom 4 of D800 and D800E and it should be obvious which one is the D800E :)
I would have attached it but it was too big.Here is a link to the screen shot http://www.pbase.com/hkruse/image/142193651/original
-
Here is one screen shot from Lightroom 4 of D800 and D800E and it should be obvious which one is the D800E :)
I would have attached it but it was too big.Here is a link to the screen shot http://www.pbase.com/hkruse/image/142193651/original
While it is "obvious," I personally don't think the difference is "significant."
Flame-suit donned
:)
-
Unless an article on this site is attributed to someone else, I'm the guilty party.
Michael
Thanks Michael.
-
Hi,
I tried some pretty hefty sharpening on both, not easy to say what sharpening is optimal. I still think that the D800E is sharper, D800E on the right.
Best regards
Erik
There are some test pictures here http://www.fotopolis.pl/index.php?gora=1&lewa=2
Use Google translate and find the D800 and D800E and the Leica S2 and compare. You can download RAW files.
Here is one screen shot from Lightroom 4 of D800 and D800E and it should be obvious which one is the D800E :)
I would have attached it but it was too big.Here is a link to the screen shot http://www.pbase.com/hkruse/image/142193651/original
-
So you found the files after all. Did you look at the S2 files also?
I find the D800E to look a little sharper but it's not obvious that there is (much) more real detail.
What sharpening parameters did you find optimal on each? I did tweak them individually and used my normal default low ISO sharpening for fine detail for my 1Ds mkIII and the params are a=50,r=0.6,d=70,m=30
-
All things being equal, more pixels will never deliver less absolute details. In the very worst case, the D800 would not deliver any additional details over the lower resolution D700. In most cases, there will be some advantage. Indeed, individual pixels may look more blurred, but the image will look identical.
Now, all things are not being equal because the high ISO behavior of high density sensors may be worse. In the present case though, it seems that after down res to 12 mp, the D800 is about one stop better than the D700/D3 at high ISOs, which puts it roughly at the level as or slightly below the D3s in terms of high ISO noise capability.
This looks like the closest thing to the universal camera. MF like DR/detail at low ISO, D3s like image quality at high ISOs..
If this looks to be a universal camera then my concerns are as follows:
- what are the requirements for shooting hand held - does the average person need 1/2xfocal-length or 1/3x-focal-length or..?
- how do the various lenses with Nikon's current VR technology help with the above?
- do I need to spend $3000 on the camera and then how many more thousands on lenses to make use of 36MP if I don't have any Nikon equipment currently and I want to cover wide angle (somewhere at or below 19mm) to medium telephoto (300mm)?
- I suppose the last question is really asking the question, for how many people is this a $3000 camera and for how many will it be a much more expensive camera if they want good quality 36MP images?
I'm content to wait for DxO and other websites to find out how good/bad the sensor is but similarly, DxO and those other websites generally never tell me what it really means to own/use cameras nor what it really costs me to get into them.
-
I'm content to wait for DxO and other websites to find out how good/bad the sensor is but similarly, DxO and those other websites generally never tell me what it really means to own/use cameras nor what it really costs me to get into them.
Well, then wait for my review in late April. ;)
Michael
-
If this looks to be a universal camera then my concerns are as follows:
- what are the requirements for shooting hand held - does the average person need 1/2xfocal-length or 1/3x-focal-length or..?
That depends a lot on your requirements and hand-steadiness, but let us not be dazzled by the exaggeration (or at least amplification) of resolution differences cause by counting pixels instead of good old lines per mm, or lines per picture height. The 36MP of the D800 promises about twice the linear resolution of 9MP, which is roughly comparable to what good low to moderate ISO speed 35mm format color film gives. So if that leads you to want to print at about twice the linear size and view from equally close, then you probably need to about double the shutter speed for hand-holding.
So if you used 1/(focal length) with 35mm film and a target of sharp 8"x10" prints, you might try 1/(twice focal length) with the D800 and a target of 16x20" prints that are as sharp as you old 8x10s when viewed from equally close.
P. S. As I have said elsewhere, even the D800 probably only matches the resolution of the finest grained monochrome films like TMAX 100, so maybe one could experiment with a film like that to get a hint of what is needed to get the most out of the D800, as fas as technique, lens choices, f-stops, etc.
Another option would be to look at the focal length to shutter speed rule that works for you with a 16MP DX format camera, which is like examining a crop from the D800.
-
I even put my 10 MP compact on a tripod on a trip with clearly better results than handheld.
-
Hi,
Yes I found the files after some digging.
My sharpening was:
D800: 51,0.5,100,9
D800E: 38,0.5,100,9
Both in LR4
Thanks for the tip on the S2, didn't have the time to check out, yet.
Best regards
Erik
So you found the files after all. Did you look at the S2 files also?
I find the D800E to look a little sharper but it's not obvious that there is (much) more real detail.
What sharpening parameters did you find optimal on each? I did tweak them individually and used my normal default low ISO sharpening for fine detail for my 1Ds mkIII and the params are a=50,r=0.6,d=70,m=30
-
I suppose the last question is really asking the question, for how many people is this a $3000 camera and for how many will it be a much more expensive camera if they want good quality 36MP images?
That's a good practical question.
And how would that 36MP FF image compare with a MF Pentax 645 or an older 31MP or 39MP Hassy?
-
Hi,
What I feel is that the Leica needs a sensor with higher resolution. With the present size of pixels I feel the image has to much staircase effects.
Anyway, the Leica S2 is imply to expensive for my budget. Nikon owners/buyers have now a really affordable high res camera at an affordable price.
Best regards
Erik
So you found the files after all. Did you look at the S2 files also?
I find the D800E to look a little sharper but it's not obvious that there is (much) more real detail.
What sharpening parameters did you find optimal on each? I did tweak them individually and used my normal default low ISO sharpening for fine detail for my 1Ds mkIII and the params are a=50,r=0.6,d=70,m=30
-
If this looks to be a universal camera then my concerns are as follows:
- what are the requirements for shooting hand held - does the average person need 1/2xfocal-length or 1/3x-focal-length or..?
- how do the various lenses with Nikon's current VR technology help with the above?
- do I need to spend $3000 on the camera and then how many more thousands on lenses to make use of 36MP if I don't have any Nikon equipment currently and I want to cover wide angle (somewhere at or below 19mm) to medium telephoto (300mm)?
- I suppose the last question is really asking the question, for how many people is this a $3000 camera and for how many will it be a much more expensive camera if they want good quality 36MP images?
Over on Nikongear there are D800 shots with legacy lens which indicate that the older glass will be quite good on the new body. No need to rush out and buy "G" lenses.
-
what does ....quite good.....mean? It isn't a technical term that I am aware of. :)
-
Over on Nikongear there are D800 shots with legacy lens which indicate that the older glass will be quite good on the new body. No need to rush out and buy "G" lenses.
The pixel density of the D800 is the same as the D7000 and did people suggest that new glass was needed for that camera? These comments come every time a new camera with more megapixels comes around and especially full frame. I remember how the Canon 1Ds mkIII would stress even L-glass according to some reviewers and having in mind that the pixel density is like the old 20D APS-C sensor :) And nobody suggested new glass for that one. Of course there is always the question about sharpness in corners and edges but the lack of sharpness on some glass in these areas will be stressed also by lesser resolving sensors like e.g. the D700 or 5D.
-
If this looks to be a universal camera then my concerns are as follows:
- what are the requirements for shooting hand held - does the average person need 1/2xfocal-length or 1/3x-focal-length or..?
- how do the various lenses with Nikon's current VR technology help with the above?
Being able to use a sufficiently fast shutter speed for hand-held shots at low ISO has always been a problem in photography when the subject is not brightly illuminated. It's why Canon invented its Image Stabilisation system, and it's why Nikon followed later with its VR system. I find these technologies tremendously useful. They remove the need for a tripod much of the time.
One should not forget that neither a tripod nor VR technology will serve much purpose when the subject is moving. However, for static subjects, modern VR and IS lenses can offer a 3-stop shutter speed advantage which actually translates to 1/8FL when using a non-VR lens.
For example, if a non-VR 100mm lens requires a shutter speed of 1/100th for a reasonable chance of a sharp shot, then a 100mm lens with the latest VR technology should provided similarly sharp results at 1/13th sec, but not a guaranteed sharp result of course, because of the random nature of hand-shake.
I've never bothered to test this, but the methodology would be to take several shots of a static target, using say a 100mm lens at 1/13th with VR on, then the same number of shots at 1/100th with VR off.
After examining the results, one should find that a few of the shots in both cases are tack-sharp, some moderately sharp, and a few, perhaps most, below standard.
In order to increase the chances of getting a sharp result, I would prefer to use 1/2FL without VR, say 1/200th with a 100mm lens, and 1/25th with VR switched on.
In a situation where a shutter speed of 1/13th or even 1/25th of a sec is required with a 100mm lens, I think a tripod would be more reliable than VR. It should produce sharp results every time, provided MLU is enabled.
On the other hand, if lighting conditions permit a shutter speed of 1/FL, I think that could reasonably be considered sufficient to get the best out of a D800 if the lens has VR technology. Perhaps this is why most of the lenses recommended by Nikon for use with the D800 are VR lenses.
-
I even put my 10 MP compact on a tripod on a trip with clearly better results than handheld.
I even put my 14MP compact on a mini-tripod strapped to ski-poles, tent-poles, fences, road-signs; or simply used body-weight to jam the closed mini-tripod against walls, trees,... with clearly better results than handheld.
Now with a heavier camera and lens those ad hoc approaches don't work so well - but using a bean bag support works better, the additional weight helps.
-
Say...I looked at those samples in the link...eh, wasn't that "cloth"?
Yes, it was. So, where's my moiré on the D800e sample?? I DEMAND to
see moiré! Ok, maybe it wasn’t the right frequency, Kenneth. Or perhaps
the photographer cleaned it up before posting. Regardless…the D800e
has a slightly sharper, cleaner look.
In deference to my fellow D800 shooters, I doubt anyone will see a diff between the
D800 and D800e in most human-scale prints. Nevertheless, I’m ok with ordering
the “e”, at least until I have a chance to shoot a variety of subjects myself. If Michael
gets his D800e before I do, and his review can convince me the “e” is a worthless
paperweight, then I’ll cancel my order, or send it back to Amazon, as the case
may be. Then I’ll order a D800 and wait till the year 2015 to get it. Or, I could order a
Fuji X-Pro 1, an Nex-7, or a 5d Mark III, or just keep shooting my D7000.
It’s all good, these days. 8)
-
For what it's worth, Amazon lists D800 as #1 selling camera currently. Where is Canon 5Dm3?... 16th place!
-
Being able to use a sufficiently fast shutter speed for hand-held shots at low ISO has always been a problem in photography when the subject is not brightly illuminated. It's why Canon invented its Image Stabilisation system, and it's why Nikon followed later with its VR system. I find these technologies tremendously useful. They remove the need for a tripod much of the time.
I know it is not the point, but for the sake of historical correctnes, VR was in fact released in a Nikon compact camera first and was later followed by Canon in DSLR lenses.
Cheers,
Bernard
-
For what it's worth, Amazon lists D800 as #1 selling camera currently. Where is Canon 5Dm3?... 16th place!
Many of us were hoping for a 5DIII with a sensor that would compete with the best of Sony's. Because of this disappointment many are switching to Nikon arguably a better camera and definitely a better sensor. In 3 years with a 5DIII I'd be shooting with a sensor using 6 year old technology, with the D800 it will only be 3 year old technology. I'm willing to pay $1000/year for the latest sensor technology but Canon wants me to pay $1,167/ year for what? No wonder the sales for the D800 are so good.
Marc
-
For what it's worth, Amazon lists D800 as #1 selling camera currently. Where is Canon 5Dm3?... 16th place!
Those Amazon sales numbers can show some curious spikes ... Maybe a burst of pre-orders that are easy to cancel has been driven by multiple recent favorable stories about the D800 on a little photographic site owned by Amazon: DPreview. The Olympus OM-D E-M5 had some stunning Amazon rankings after its recent favorable DPreview coverage too.
Still, I am not disputing the widespread sentiment that Nikon is producing impressive, inovative new products while Canon has become a complacent market leader, just "punching the clock" on some its recent products.
-
Amazon or not, I will say this, the same file of the ubiquitous DPREVIEW still-life had me sit up and pay attention....
-
Many of us were hoping for a 5DIII with a sensor that would compete with the best of Sony's. Because of this disappointment many are switching to Nikon arguably a better camera and definitely a better sensor. In 3 years with a 5DIII I'd be shooting with a sensor using 6 year old technology, with the D800 it will only be 3 year old technology. I'm willing to pay $1000/year for the latest sensor technology but Canon wants me to pay $1,167/ year for what? No wonder the sales for the D800 are so good.
Marc
This pretty much covers me but what I'd like to know more about the D800 and its lens requirements before I start splashing around with cash. $3000 for a camera, ok. But if it's $10,000 for camera+stuff, I'll think again. I suspect that it won't be $10,000 but that it will be more than $3000.
-
I know it is not the point, but for the sake of historical correctnes, VR was in fact released in a Nikon compact camera first and was later followed by Canon in DSLR lenses.
Cheers,
Bernard
Was this the Nikon Zoom 700VR QD 35mm film compact camera released in 1994, Bernard?
If so, Canon gave us the first image-stabilised zoom lens for the SLR just a year later, the EF 70-300/F4.5-5.6 IS USM. It took Nikon another 5 years to produce their first VR zoom for an SLR, the 80-400/F4.5-5.6 D, in 2000. That was rather slack of Nikon, don't you think? ;D
They lagged behind Canon to the same degree regarding full-frame DSLRs. Canon produced the 11mp 1Ds in 2002. 5 years later Nikon announced its first full-frame DSLR, the D3 with only 12mp.
But that's history. I think Nikon have truly caught up with Canon; even streaked ahead.
I don't know if VR II is better than the latest Canon IS, but the claimed 4 stop shutter speed advantage is really impressive. If there's no movement in the scene, one should be able to use a shutter speed as slow as 1/6th sec with a 100mm VR lens as opposed to 1/100th without VR. That's a huge advantage. Who needs a tripod! (Okay! I know you still need one for really slow shutter speeds on a moonlit night ;D ).
-
Hi,
I now have a Sony Alpha 77 SLT a 24 MP APS-C camera. In full frame this would correspond to 54 MP. With that camera I can see that some lenses don't keep up, but some do quite fine. The 16-80/3.5-4.5 normal zoom seems to keep up fine but 70-300/3.5-4.5 APO zoom seems to "fall apart" at longer focal lengths, but that lens didn't really impress on the other cameras either.
I general, I would presume that good lenses are OK at high pixel densities. Indeed some experiments I have may indicate that resolution (LP/mm at 50% MTF as measured by Imatest) scales perfectly with pixel pitch. This test was done on the optical axis, corner figures are probably much worse.
I'd aslo suggest that extracting all available resolution from a sensor/lens combination requires careful and precise work. Exact focusing, camera on solid tripod, mirror lock up or mirrorless camera and electronic camera release.
The final point may be that the sensor outresolving the lens is not a bad thing. If that happens the sensor simple oversamples the image. We get smoother edges with no jaggies, no aliasing and so on.
Best regards
Erik
The pixel density of the D800 is the same as the D7000 and did people suggest that new glass was needed for that camera? These comments come every time a new camera with more megapixels comes around and especially full frame. I remember how the Canon 1Ds mkIII would stress even L-glass according to some reviewers and having in mind that the pixel density is like the old 20D APS-C sensor :) And nobody suggested new glass for that one. Of course there is always the question about sharpness in corners and edges but the lack of sharpness on some glass in these areas will be stressed also by lesser resolving sensors like e.g. the D700 or 5D.
-
This pretty much covers me but what I'd like to know more about the D800 and its lens requirements before I start splashing around with cash. $3000 for a camera, ok. But if it's $10,000 for camera+stuff, I'll think again. I suspect that it won't be $10,000 but that it will be more than $3000.
Just a few days ago, walking along a beach in Thailand, I came across a few Rhesus monkeys splashing around in their favourite swimming pool.
I was carrying my D7000 with 24-120/F4 zoom attached. I really needed a zoom with a reach of at least 400mm, which would have been 600mm equivalent on the D7000. Nevertheless, I took a few shots with zoom lens set at 120mm, F8 and 1/500th exposure.
Curious as to how sharp such distant images would appear when severely cropped, considering that this lens is not at its sharpest fully extended and not at its sharpest at F8, I cropped the following image to 2.4MB, which is the equivalent of an image from a 0.8mp sensor of dimensions 5.4mmx3.6mm, using the equivalent of an 800mm lens in 35mm format terms.
As mentioned by others, the D7000 has the same pixel density as the D800, so I can presume that this crop would be similarly as sharp had I been using a D800 with this lens instead of the D7000.
In relation to the D800, I calculate the crop factor of this 2.4MB image to be about 6.7x. (6.7 x 120mm = 804mm).
The crop looks reasonably sharp to me. What do you think?
-
I forgot to mention the ISO of the above image. Before some of you comment that there is noticeable noise in that 100% crop representing detail at the pixel level similar to what I deduce the D800 would produce, the shot was considerably underexposed at ISO 100.
If I had increased ISO to get an ETTR shot at that shutter speed and aperture, I estimate from the histogram in ACR it would be at least ISO 800. What do you think?
So what we are looking at here is a simulation of a small P&S sensor, smaller than average, with fairly massive pixels for a P&S (just 0.8mp covering a sensor area of about 5.4mmx3.6mm) and a massive telephoto reach of 800mm in 35mm format terms. Furthermore, to achieve the shutter speed of 1/500th necessary to freeze motion, and to get the benefit of a good DoF that F8 provides at the 800mm FL equivalence, an ISO of 800 was required.
-
This pretty much covers me but what I'd like to know more about the D800 and its lens requirements before I start splashing around with cash. $3000 for a camera, ok. But if it's $10,000 for camera+stuff, I'll think again. I suspect that it won't be $10,000 but that it will be more than $3000.
Well I invested in 2 Leica R's for the anticipated increased resolution in the 5DIII so just a couple of adapters and I'm set (got lucky on this decision), I already have a Nikkor 14-24 for the wide stuff I'll just remove the adapter!
saved myself $200 ordering the D800E instead of the 5DIII and got 2 extra stops of DR and 14,000,000 more photosites!
Marc
-
saved myself $200 ordering the D800E instead of the 5DIII and got 2 extra stops of DR and 14,000,000 more photosites!
Assuming that Nikon gets any close to being able to produce the tremendous number of bodies the market seems to be expecting and that the retailers play fair with us... ;)
Cheers,
Bernard
-
As a very long time Canon user I think the so called upgrade to the 5D2 to the 5D3 is pathetic and I am seriously considering switching over to Nikon... What in the world was Canon thinking??? Eleanor
Many of us were hoping for a 5DIII with a sensor that would compete with the best of Sony's. Because of this disappointment many are switching to Nikon arguably a better camera and definitely a better sensor. In 3 years with a 5DIII I'd be shooting with a sensor using 6 year old technology, with the D800 it will only be 3 year old technology. I'm willing to pay $1000/year for the latest sensor technology but Canon wants me to pay $1,167/ year for what? No wonder the sales for the D800 are so good.
Marc
-
… and got 2 extra stops of DR...
?
-
Canon is so asleep at the switch right now that it's not even funny. While Nikon effects a paradigm shift of photography with the D800, Canon is still trying to convince their customers that they've finally fixed three year-old focusing problems. Ouch.
Maybe the have something up their sleaves. But given their corporate attitude, I wouldn't count on it.
- N.
-
For what it's worth (or not worth??)...just noticed that the Nikon D800 now stands at the top of all sensor ratings (incl. above the Phase IQ180) at DXOMark. Eleanor
-
For what it's worth (or not worth??)...just noticed that the Nikon D800 now stands at the top of all sensor ratings (incl. above the Phase IQ180) at DXOMark. Eleanor
Yes, and the dxomark website seems to have slowed to a crawl as every man and his dog goes to check it out ;)
But a score of 14.1 EV for landscape... wow. If I'd been silly enough to pre-order the 5DIII, I'd be cancelling.
-
"Canon is so asleep at the switch right now that it's not even funny. While Nikon effects a paradigm shift of photography with the D800, Canon is still trying to convince their customers that they've finally fixed three year-old focusing problems. Ouch."
I don't know, it seems to me that both Canon and Nikon, with these two cameras, tried to address the desires of their customers. It seems to me that Canon 5DII users rarely complained about resolution and frequently complained about autofocus, and fps speed. Many D700 users, and pretty much all other Nikon users have a seriously pent up demand for a high resolution camera since the 1dsII was announced (was this 2004?), and for an affordable one since the 5dII announcement. I suspect this is the real reason there is so much interest in the D800 at the moment. Many of the 5dII users that were content with their camera purely for it's resolution perhaps still are. Hence the apparent lower demand to upgrade.
I use Nikon and have ordered a D800e because I print and sell large landscape prints and currently use a D3. But I live in and sell my prints in a ski town and so I shoot a fair bit of what I'll call "sportscapes" where both high resolution and shooting speed are handy. Too me, although the D800 will be nice, I could have been very happy with a more all round camera like the 5dIII seems to be.
-
I forgot to mention the ISO of the above image. Before some of you comment that there is noticeable noise in that 100% crop representing detail at the pixel level similar to what I deduce the D800 would produce, the shot was considerably underexposed at ISO 100...
Ray, that is incredibly impressive, i.e., the sensor ability to pull out so much detail in so underexposed areas (eyes and mouth in particular)!
On a related note, seems to me that in the future (and not so distant) it won't really matter if we exposed correctly (bar blown highlights) or even focused correctly (Lytro technology). It won't even matter if we composed correctly (the CS6 content-aware move etc.). The future will turn the century old "f/8 and be there" into 10% of that and 90 % of a new paradigm: "look ma what I made from this crappy file!" In other words, photographers will become photoshoppographers™ :-\
-
Forgive my ignorance here.... but how can this new Nikon CMOS sensor out perform a medium format CCD sensor like that in the IQ180 which is supposed to have "better ?" pixels?? I'm really curious, being the owner of the Phase P65+, which it also beat in the new DXO tests... I'm wondering if the D800 and D800e will have creamy smooth files at base iso like medium format sensors too.....thanks, Eleanor
Yes, and the dxomark website seems to have slowed to a crawl as every man and his dog goes to check it out ;)
But a score of 14.1 EV for landscape... wow. If I'd been silly enough to pre-order the 5DIII, I'd be cancelling.
-
For the benefit of those who are having trouble connecting to the DXO results, or who are not familiar with the site, I'll provide a brief summary of what I've found so far.
(1) The D800 pixel has very similar performance to the D7000 pixel in terms of SNR at 18% grey (skin tones), Dynamic Range, Tonal Range, and Color Sensitivity, as predicted and as assumed in my arguments relating to my D7000 images above.
At equal print size, the D800 is of course better in all parameters.
(2) Compared with the D3s at equal print size, the D3s still retains the low-noise-at-high-ISO title. Above ISO 1600, the D3s has better DR by amounts ranging from 0.4EV to almost 1EV at ISO 12,800 and 25,600, at equal print size.
However, the 0.4EV DR advantage of the D3s at ISO 3200 and 6400 is not nearly as significant as the 2 &1/3rd stop DR advantage of the D800 at its base ISO of 100, compared with the DR of the D3s at its base of ISO 200.
(3) At equal print size, the D800 outstrips the D700 in every respect. SNR is over one stop better at base ISO, and about one stop better at ISO 12,800 and 25,600.
-
Ray, that is incredibly impressive, i.e., the sensor ability to pull out so much detail in so underexposed areas (eyes and mouth in particular)!
Well, thank you, Slobodan. I sometimes wonder how these images might appear on a well-calibrated monitor, because I'm processing them on a Dell notebook, because I'm currently travelling.
On a related note, seems to me that in the future (and so distant) it won't really matter if we exposed correctly (bar blown highlights) or even focused correctly (Lytro technology). It won't even matter if we composed correctly (the CS6 content-aware move etc.). The future will turn the century old "f/8 and be there" into 10% of that and 90 % of a new paradigm: "look ma what I made from this crappy file!" In other words, photographers will become photoshoppographers™ :-\
I think you might be right. I'm reminded that certain great painters of the past would sometimes make notes of what they saw when travelling, with regard to their impressions of color, detail and shape etc, then on return to their studio would refer to such notes when painting the scenes from memory.
I consider the RAW image to be an extremely detailed set of notes, and Photoshop a terrific paint brush, figuratively speaking. ;D
-
... I consider the RAW image to be an extremely detailed set of notes, and Photoshop a terrific paint brush, figuratively speaking. ;D
+1
The file or negative scan is just raw material.
The raw must be cooked, means the information which is in it and which you have seen in the scene must be unveiled.
As we know grandmaster AA did a lot of magic stuff in the darkroom - which is photoshop today.
-
?
Over the 5DIII see DxO results
Marc
-
Well for next Christmas I hope the IQ180+ has one of theses sensors in it! Larger of course
Marc
-
Awhile back I heard via the "grapevine" that Phase "might" be coming out with more "inexpensive" (whatever that means) CMOS sensor cameras or backs (?) in the future along with their backs with CCD sensors...while this is just hearsay and speculation...I wouldn't be surprised given the advances in CMOS sensors. Eleanor
Well for next Christmas I hope the IQ180+ has one of theses sensors in it! Larger of course
Marc
-
Just set up this quick comparison between the D2x, D3x and D800:
http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/792%7C0/(brand)/Nikon/(appareil2)/485%7C0/(brand2)/Nikon/(appareil3)/207%7C0/(brand3)/Nikon
You've got to wonder how some of us managed to shoot with a D2x as little as 5 years ago... ;) As an example, the following Flickr set Chinese images were shot with a D2x with only 10 stops DR:
http://www.flickr.com/photos/bernardlanguillier/sets/72157594364221841/
So OK, we now have incredibly good sensors, but I feel that we should not over stress the value for our photography more today than we did back then.
For Canon shooters also, what really matters is the abilities of a camera relative to the intended usage. I am sure that the 5DIII will fare very well for most actual usage and deliver results aligned with the needs, just like the D2x was able to deliver excellent images 5 years ago.
Nobody looking at amazing prints shot from a 5DIII file will care about the fact that it was shot with a body that is a bit behind the curve.
If I were a Canon shooter I am not sure I would really consider switching.
Cheers,
Bernard
-
Bernard -- Thanks for another reminder of your fine images from Jiuzhaigou and elsewhere in Sichuan.
And I'll second your thoughts about not over stressing the role of the new hardware.
While I am looking forward to using a modern body for my lenses left over from F-mount Kodaks,
there is still plenty of mileage left in the image archives from the 14n, Fuji S2, Canon 10D and original 5D,
not to mention more transparencies than I'll ever get around to scanning.
Reprocessing a lot of old files with the new capabilities in LR4 and ACR7 may be as much fun as a new capture device.
Regards,
Rick
-
Hi,
My understanding is that there is little difference between CMOS and CCD regarding image quality, it is just different circuit design. The main differences are:
- CMOS has additional electronics in each pixel, making the sensel smaller (advantage CCD). This also means that a CMOS sensor without microlenses has worse fill factor than CCD.
- CMOS can be read out nondestructively, allowing for "correlated double sampling" (advantage CMOS)
- CMOS can have much lower read out noise, D800 advantage in DR comes from this (advantage CMOS)
- CMOS now has useful live view
Color is only affected by the CGA (Color Grid Array) and not by circuit technology.
Most of the advantage Nikon has in DxO mark probably comes from the low readout noise, that really helps dynamic range but also high ISO.
Canon seems to have some issues with their CMOS sensors, readout noise is quite high and they seem to have worse noise characteristic than Nkon. It seems that Nikon simply cuts away some noise, while Canon is less aggressive, making Canon better for some astronomic work.
The Nikon D800 sensor is developed and made by Sony. Most of the advantage is coming from Sony's design with an ADC (Analog Digital Converter) for each column of the sensor.
In my view and experience DR may be a bit overrated. I never had problems with lack of DR on any of my cameras. Also, all the advantage in DR is in the shadows. Highlight DR is same on all cameras, except if they under expose intentionally. Fujifilm has some technologies to extend highlight DR, by having extra pixels assigned for highlight.
My guess is that MF digital has advantages in three areas:
- A larger sensor can collect more photons, always an advantage. Shot noise is reduced. Helps with DR but not that much (around one stop).
- The lenses are less stressed for MTF. MTF for an ideal lens drops almost linearly with pixel size, so larger pixels mean better MTF
- The CGA may have better characteristics, especially if it is designed without much concern for high ISO
Best regards
Erik
Forgive my ignorance here.... but how can this new Nikon CMOS sensor out perform a medium format CCD sensor like that in the IQ180 which is supposed to have "better ?" pixels?? I'm really curious, being the owner of the Phase P65+, which it also beat in the new DXO tests... I'm wondering if the D800 and D800e will have creamy smooth files at base iso like medium format sensors too.....thanks, Eleanor
-
Thanks Erik...and one more question...when comparing my Leica M9 and Sony NEX7 both at base iso and both with same Leica lens...50 lux....the M9 files have that smooth look all over the file, including mid tones. The Sony files have a slightly " grainy" look at 100 percent. Because of this I would choose to shoot landscape with my 18 megapixel Leica over my NEX7 with 24 mp ...both using Leica lenses. Is this difference a ccd vs. CMOS issue! Thanks! Eleanor
Hi,
My understanding is that there is little difference between CMOS and CCD regarding image quality, it is just different circuit design. The main differences are:
- CMOS has additional electronics in each pixel, making the sensel smaller (advantage CCD). This also means that a CMOS sensor without microlenses has worse fill factor than CCD.
- CMOS can be read out nondestructively, allowing for "correlated double sampling" (advantage CMOS)
- CMOS can have much lower read out noise, D800 advantage in DR comes from this (advantage CMOS)
- CMOS now has useful live view
Color is only affected by the CGA (Color Grid Array) and not by circuit technology.
Most of the advantage Nikon has in DxO mark probably comes from the low readout noise, that really helps dynamic range but also high ISO.
Canon seems to have some issues with their CMOS sensors, readout noise is quite high and they seem to have worse noise characteristic than Nkon. It seems that Nikon simply cuts away some noise, while Canon is less aggressive, making Canon better for some astronomic work.
The Nikon D800 sensor is developed and made by Sony. Most of the advantage is coming from Sony's design with an ADC (Analog Digital Converter) for each column of the sensor.
In my view and experience DR may be a bit overrated. I never had problems with lack of DR on any of my cameras. Also, all the advantage in DR is in the shadows. Highlight DR is same on all cameras, except if they under expose intentionally. Fujifilm has some technologies to extend highlight DR, by having extra pixels assigned for highlight.
My guess is that MF digital has advantages in three areas:
- A larger sensor can collect more photons, always an advantage. Shot noise is reduced. Helps with DR but not that much (around one stop).
- The lenses are less stressed for MTF. MTF for an ideal lens drops almost linearly with pixel size, so larger pixels mean better MTF
- The CGA may have better characteristics, especially if it is designed without much concern for high ISO
Best regards
Erik
-
D800-800E comparison
Roll your cursor over the image to change from D800 to D800E
http://julian.marmotandweasel.com/photos/d800-d800e_test.html
-
Wow. It's going to be so nice to not feel like I need to rub my eyes when processing files anymore.
-
Scenario:
"Somewhere, deep inside Canon "skunkworks", a place so secret, even the
people who work there don't know it exists, a 48mp, no AA filter Ultra-Sensor
is being readied to reclaim "Top Dog" status in the full frame dslr world."
Well, probably. Anyway, for now, this is the Nikon shooter's day in the sun. We
shall enjoy it, as "all glory is fleeting", especially in the technology world that
is a never-ending creative destruction force. Besides, increasingly, the
line between "Nikon/Canon/Sony/Fuji/etc/etc shooter" is becoming faded,
as photographers are buying into more than one system. It's all good...really
good, even if somewhat disconcerting as to where value and necessity lay.
But yes, undeniably Fun ;)
-
Eleanor,
I don't know. Two observations I have is that the Leica has a larger sensor, so it will collect more photons. Most of the noise we see is coming from statistical distribution of noise, it's normally called shot noise. Doubling sensor area should reduce noise by about 40 percent. My guess is that shot noise on Leica at 100 ISO would correspond to shot noise on NEX-7 at 50 ISO.
The other issue is that Leica doesn't have AA-filter while the NEX-7 has AA-filter. So the NEX needs more sharpening, that also increases noise.
Not much, but this is the explanations I can come up with.
The enclosed diagram from DXO-mark shows about one stop advantage of M9 over NEX-7.
Best regards
Erik
Thanks Erik...and one more question...when comparing my Leica M9 and Sony NEX7 both at base iso and both with same Leica lens...50 lux....the M9 files have that smooth look all over the file, including mid tones. The Sony files have a slightly " grainy" look at 100 percent. Because of this I would choose to shoot landscape with my 18 megapixel Leica over my NEX7 with 24 mp ...both using Leica lenses. Is this difference a ccd vs. CMOS issue! Thanks! Eleanor
-
Thanks for this very helpful explanation Erik. All makes sense! Now I'm wondering if the new Nikon D800e will have creamy smooth files at the base iso...similar to the M9?? Guess time will tell! Eleanor
Eleanor,
I don't know. Two observations I have is that the Leica has a larger sensor, so it will collect more photons. Most of the noise we see is coming from statistical distribution of noise, it's normally called shot noise. Doubling sensor area should reduce noise by about 40 percent. My guess is that shot noise on Leica at 100 ISO would correspond to shot noise on NEX-7 at 50 ISO.
The other issue is that Leica doesn't have AA-filter while the NEX-7 has AA-filter. So the NEX needs more sharpening, that also increases noise.
Not much, but this is the explanations I can come up with.
The enclosed diagram from DXO-mark shows about one stop advantage of M9 over NEX-7.
Best regards
Erik
-
I'm not completely sure, but I could imagine the "game changer" properties of these new high-res DSLRs might on the long run even change the mind of a filmosaurus like me. But its not only about resolution, noise, color depth and low ISO or overall technical IQ.
Its also about the overall style of the rendering of the world by such systems and the change in shooting style. My biggest concern is if and if yes -how- digital will change or even corrupt me as the kind of photographer I want to be.
I think I'll loan a D800 some time this year along with some Zeiss glass and do the testing for myself. I'll also definitely try some MFDB system before I'll do the switch to full digital.
Its pretty much clear that we are living in terrific times with a lot of choice and possibilities of how we want to work. But with the freedom of choice comes the burden of responsibility and the endless possibilities to screw up.
Exciting!
-
Over on Nikongear, Sten Rasmussen has had the opportunity to shoot the Nikon D800 versus the 5DMkIII using the same Nikon lens. While the Nikon seems better, the Canon is no slouch. He is shooting under conditions which really push a sensor - at night and sunrise. It's an on-going test so check it out.
-
Digital made me a better film photographer ;)
Back to the Nikon interesting no doubt and surely a big seller for them. I've gone past being tired of the megapixel discussions they are tedious and almost irrelevant to most photographers. You certainly had reason to complain back in ye early digital days gasp all of 2 megapixels which was a bit of a joke really. You can be certain that for some people out there 36 megapixels won't be enough ::)
Clearly there is a sales pitch that continues to this day and I'm surprised how many get suckered into it, no doubt the camera makers are loving every minute of it!
For the rest of us it's like using a flame thrower to "take out" a house fly, it's a tad over the top to say the least. I'm sure someone will take a fisheye lens shot and try to crop it to a 500mm one, or compare pixel peeping v various MF cameras etc. In a few years time folks will declare the mp wars are over when we hit 50mp, then someone will tell us we're missing out on all that resolution that 100mp offers.
Sorry to be the super sceptic that I am well known for. I used to get asked once in a while why I shoot 35mm film or APS-C digital for landscapes and not MF or FF DSLR's it's simply a case of "what you needs/want/want to carry/price etc" Great images happen even if they're not super high res. These discussions are merely distractions from what we should be doing, which is worrying about what we take pictures of, and a lot less about what we use to take pictures!
I'm guilty of wasting time in forums as much as most are, it's threads like this that remind me it's time to just "go out" and actually take some pictures, and I'm going to use my cheapest junky kit lens so that when I run into an American tourist with a big Canon full frame DSLR and L lenses :)
-
.....In a few years time folks will declare the mp wars are over when we hit 50mp, then someone will tell us we're missing out on all that resolution that 100mp offers....
Until a man can count the hairs on a gnat's ass at 900 yards in the middle of a moonless night, we will never have enough resolution or low light performance. ;D
-
Until a man can count the hairs on a gnat's ass at 900 yards in the middle of a moonless night, we will never have enough resolution or low light performance. ;D
Two words: shaved gnats.
-
Two words: shaved gnats.
+1
-
I mentioned above a comparison between the DxOMark scores of the D800 and D2x and marveled at the fact that we were already able to take exhibition grade images with the D2x back then although it did have clear DR and high ISO shortcomings.
I did another pass, comparing this time the D2x with the J1... to find out that they basically deliver near identical levels of performance... with the J1 providing a lot more DoF everything else being equal. If I were to stick to my 2007 views of the world, I would have no choice but to call the J1 the ultimate landscape camera. ;)
http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/(appareil1)/744%7C0/(brand)/Nikon/(appareil2)/207%7C0/(brand2)/Nikon
We do live in a wonderful period of photographic evolution, do we not?
Cheers,
Bernard
-
We do live in a wonderful period of photographic evolution, do we not?
Yes, even with bottom-of-the-range entry-level cameras ;-)
Nikon 1 J1 :: Sony SLT A-35 :: Nikon D2X (http://www.dxomark.com/index.php/Cameras/Compare-Camera-Sensors/Compare-cameras-side-by-side/%28appareil1%29/744|0/%28brand%29/Nikon/%28appareil2%29/716|0/%28brand2%29/Sony/%28appareil3%29/207|0/%28brand3%29/Nikon)
-
... we were already able to take exhibition grade images with the D2x back then although it did have clear DR and high ISO shortcomings....
True, because we took those shortcomings into account and worked around them, just like we did with slide film. If we couldn't pull out four-stop underexposed shadow detail, well, so be it, we let it in the dark and worked with composition and esthetics that did not rely on it. Many Kodachrome/Velvia shots produced deep black shadows, resulting in strong contrast and saturated colors, that we then used for almost graphical representation of the subject. Photographers like Pete Turner and Eric Meola come to mind.
-
Digital made me a better film photographer ;)
Back to the Nikon interesting no doubt and surely a big seller for them. I've gone past being tired of the megapixel discussions they are tedious and almost irrelevant to most photographers. You certainly had reason to complain back in ye early digital days gasp all of 2 megapixels which was a bit of a joke really. You can be certain that for some people out there 36 megapixels won't be enough ::)
Clearly there is a sales pitch that continues to this day and I'm surprised how many get suckered into it, no doubt the camera makers are loving every minute of it!
Those of us who used to shoot film were surely aware that the larger formats had clear advantages over smaller formats in part because the same film types were available for most formats.
If we had had a situation where the finest grain films were only available for the APS-C format, then the next finest grain film was only available for 35mm format, and the larger formats like MF could only use film with the coarse grain typical of ISO 800 film and above, then the advantages of the larger format would not have been so apparent and clear-cut.
However, for the past 10 years this has effectively been the situation with digital cameras; the smaller the format, the finer the grain, ie. the smaller the pixel.
The usual pattern with Canon has been to provide us with a new, finer-grained cropped-format one year, then make us wait several years for their larger full-frame models to catch up, in terms of pixel density (ie. film grain).
How many years between the Canon 6mp D60 and the 16mp 1Ds2? How many years between the 8mp 20D and the 21mp 5D2?
Nikon (or Sony) seem to have broken this pattern and given us within just a couple of years, a full-frame DSLR with the same pixel density (ie. virtually the same pixels) as their next smaller format, the D7000.
This is how it should be, considering that both of these formats can use the same lenses.
The difference between the 5D3 and the D800 from this perspective is that the D800 can use the equivalent of a slightly finer grain film type but of similar ISO and similar push-processing characteristics.
If I were in charge of Canon, I'd be pulling out all stops to get a full-frame 7D out on the market as quickly as possible, and at an attractive price. That would be approximately a 50mp camera since Canon's cropped-format is slightly smaller than Nikon's and also has slightly greater pixel density.
-
Not much, but this is the explanations I can come up with.
I am under the impression that some Sony cameras have prioritized "accurate" colors over SNR. I.e. choosing a color filter that lets the color correction matrix be closer to all diagonal terms, at the cost of discarding more light. I dont know about Leica.
-h
-
I'm not going to argue film formats or that bigger is better. I will argue that for many "good enough" has been reached with digital some time ago. I suppose for many 35mm film was good enough, APS-C wasn't, nor was 110 disc and other smaller formats. Some liked MF and LF but that was a small group
There are more interesting things to talk about, and the 36mp users will surely crave more pixels you jump on this tech love affair bandwagon you'll never get off it.
And personally I think Bayer technology is going to be replaced with multi layered colour aware sensors offerings real benefits for end users, and not just more resolution. I would not wish to invest in current Bayer technology too much.
-
And personally I think Bayer technology is going to be replaced with multi layered colour aware sensors offerings real benefits for end users, and not just more resolution. I would not wish to invest in current Bayer technology too much.
Barry, you may be correct. Personally I don't care as long as it comes in an F mount :)
-
For what it's worth, Amazon lists D800 as #1 selling camera currently. Where is Canon 5Dm3?... 16th place!
Further to my earlier post, here is a little anecdotal evidence just how bad the 5Dm3 yawn-inducing character is, tweets from my local Calumet:
-
I will argue that for many "good enough" has been reached with digital some time ago.
Quite so. No argument there. For the vast majority of people who take photos, the P&S camera or the mobile phone is sufficient. As smartphones become more sophisticated, incorporating better quality cameras like the Nokia Pure View 808, even the latest P&S will be more than most people require. Their mobile phone will be sufficient for all photographic purposes.
Only those who are passionate about photography, even obsessed, find the leap forward of the D800 interesting. Those who are hardly interested at all in photography will find it all a big yawn of course.
There are more interesting things to talk about, and the 36mp users will surely crave more pixels you jump on this tech love affair bandwagon you'll never get off it.
All prosperous societies are on a technological bandwagon. Jump off it and you're back in the Stone Age. Cameras and lenses are technological devices that have an understandable fascination for many people. The lens, whether camera lens, binocular or telescope, is like an extension of our eyesight, and the recorded image is like an extension of our powers of memory.
Who would not crave to have better eyesight and better memory! Without the benefits of the higher resolution of the telescope, we'd probably still be wondering if the universe revolved around the earth.
-
– Use the optimum aperture. Apertures above about f/11 introduce diffraction and effectively act as an unintentional AA filter
– Use a really solid tripod and head
– Use Live View, or mirror lock up with a remote release or self timer.
– Use critical focusing, using single point AF and LV focus magnification (up to 23X)
– When shooting hand-held use lenses with VR when possible, and also a high shutter speed... 2 or 3 over the reciprocal of the focal length, not the 1/focal length of olden days
– Use the lowest possible ISO, and if shooting JPGs turn off high ISO noise reduction even at low ISO
There are some trade offs for higher resolution. Are you prepared for a MF shooting style?
Cheers,
-
– Use the optimum aperture. Apertures above about f/11 introduce diffraction and effectively act as an unintentional AA filter
Better still, use a D800E instead of a D800. At F11, aliasing artifacts will be minimised and you will probably get the same resolution at F11 as the D800 provides at F8.
My own tests comparing the Canon 40D (pixel density equivalent to full-frame 25.6mp) with the Canon 50D (pixel density equivalent to a 38.4mp full-frame) show that the 50D at F16, with 50mm prime lens, delivers the same resolution as the 40D at F11. Also, the 50D at F11 delivers the same resolution as the 40D at F8.
– Use a really solid tripod and head
It may not help if the subject is moving.
– When shooting hand-held use lenses with VR when possible, and also a high shutter speed... 2 or 3 over the reciprocal of the focal length, not the 1/focal length of olden days
The reciprocal of the focal length would be more than adequate with modern VR lenses, especially with VRII lenses that claim a 4 stop advantage. A 2 stop advantage is equivalent to 1/4FL. A 3 stop advantage, 1/8FL, and a 4 stop advantage 1/16FL.
– Use the lowest possible ISO, and if shooting JPGs turn off high ISO noise reduction even at low ISO
There are some trade offs for higher resolution. Are you prepared for a MF shooting style?
One should always try to use the lowest ISO consistent with the appropritate shutter speed required to freeze subject movement and/or camera shake, and the appropriate aperture for the desired DoF, whatever the resolution of the sensor. However, if one fails to achieve this due to poor lighting and/or miscalculation, the results will not be worse than they would have been using the same settings with a lower resolution camera of similar design and quality.
However, it may be the case that a higher resolution sensor of a different design, such as the IQ180, may achieve better SNR, Tonal Range and Color Sensitivity than the lower resolution camera, such as the D800, only when used at its significantly lower base ISO, which in the case of the IQ180 is only 29, as opposed to 74 for the D800.
I don't believe D800 users will need to be prepared for an MF shooting style when (1) base IS is more than a stop higher than most MFDBs, (2) VRII lenses allowing a 4 stop shutter speed advantage are now available, and (3) the same DoF, whether shallow or extended, can be achieved with at least a 1 stop wider aperture with the D800, giving it more than a 2 stop shutter speed advantage at base ISO without VR. ;D
-
Glad you know more than Michael.
Cheers,
-
Glad you know more than Michael.
Cheers,
It was from Michael I first learned a number of years ago that the 1/FL rule as it applies to the cropped format should be 1/FL 35mm equivalent. In other words, in circumstances when a shutter speed of 1/80th is sufficient with an 80mm lens on full frame, that same shutter speed of 1/80th is required with a 50mm lens on a cropped format camera, in the same circumstances, because a 50mm lens on the cropped format provides the same FoV as an 80mm lens on full-frame.
The techniques required to maximize image quality using a Canon 7D should be even more stringent than those required for the D800 because the 7D has the greater pixel density of a 46mp full-frame.
-
An experiment that anyone can try, and which only takes a few minutes, is a follows.
Put your favourite lens on you favourite camera. Step outdoors, find a detailed subject, and do a series of exposures at a range of shutter speeds, from silly slow to ultra fast. Use a variable ND filter so that the ISO and aperture remain the same for each shot.
Load the files on your screen and step though them at 100%. You will quickly see what hand-held speeds allow critically sharp images, and which don't.
Add image stabilization, caffeine, and lack of sleep can be added as variables. Season to taste.
You'll be amazed.
Michael
-
An experiment that anyone can try...
Outrageous! Substitute specific information for opinion! Agghhh!
-
Outrageous! Substitute specific information for opinion! Agghhh!
Yep... That reminds me of an anecdote about medieval scholastic philosophers who beat the crap out of the guy who happened to be overhearing their attempts to determine, theoretically of course, the number of teeth in a horse, and dared to suggest to just open the horse's mouth and count ;D
-
An experiment that anyone can try, and which only takes a few minutes, is a follows.
... do a series of exposures at a range of shutter speeds, from silly slow to ultra fast. ...
Indeed, doesn't every basic photography class do something like that? Scenes including writing on signs or clock-faces are good subjects. In my basic class, we did it with slides so we could all judge each other "at great magnification on screen", but in the old sense. There was a variation by at least two stops, from those who could get quite good results at down to half or less the 1/f shutter speed, to those who needed to double that venerable speed guideline to be safe.
-
just open the horse's mouth and count
"Aristotle maintained that women have fewer teeth than men; although he was twice married, it never occurred to him to verify this statement by examining his wives' mouths."
Bertrand Russell, Impact of Science on Society (1952) ch. 1
(As you might guess, much has been said about why Aristotle held that opinion - but whatever his reason I think I might have been squeamish about poking around in someone's mouth in an age not celebrated for dental hygiene.)
-
An experiment that anyone can try, and which only takes a few minutes, is a follows.
Put your favourite lens on you favourite camera. Step outdoors, find a detailed subject, and do a series of exposures at a range of shutter speeds, from silly slow to ultra fast. Use a variable ND filter so that the ISO and aperture remain the same for each shot.
Load the files on your screen and step though them at 100%. You will quickly see what hand-held speeds allow critically sharp images, and which don't.
Add image stabilization, caffeine, and lack of sleep can be added as variables. Season to taste.
You'll be amazed.
Michael
I've just done that Michael and the results tend to confirm that a shutter speed of 1/FL in conjunction with VR is about right, although maybe not as reliable as a faster shutter speed, or the use of a tripod, in producing consistently sharp results
However, I don't have a variable ND filter at hand. I used to use those mostly during my waterfall phase a few years ago. In any case, the reality is, when shooting hand-held, one may have to contend with the negative effects of noise and image degradation due to underexposure or use of a higher ISO if the shutter speed requires it.
If the sharper results from a shutter speed faster than 1/FL are negated by greater noise, then so be it. That's reality.
Because memory and storage is now so affordable, I often bracket exposure +/- one stop. The problems I tend to find with the greater exposures which may be close to a 1/FL exposure (in conjunction with VR), is unsharpness due to subject movement rather than camera shake. For this reason I may prefer the least exposure at 4x the shutter speed of the greatest exposure, even though it's far from being an ETTR.
Although I wasn't able to use a variable ND filter for my tests, I was able to use a brick wall as my test target, and a very nice brick wall it is too. ;D
In order not to appear biased I've included two pairs of comparisons of 100% crops that have received no more than the default sharpening of 25 in ACR.
One comparison shows an exposure at 100th with a 120mm lens being at least as sharp as the same scene shot at 1/400th, with VR. The other shows an exposure at 1/125th being very slightly less sharp than another shot at 1/400th. Nevertheless, the 1/400th shot is noticeably noisier, and a slightly greater amount of sharpening applied to the 1/125th shot more or less equalizes both sharpness and noise.
My favourite VR lens at the moment is the Nikkor 24-120/F4. In fact it's my only Nikkor lens with VR, so it has to be my favourite. ;D
I see no reason why the test images below would not be relevant to a D800 situation, the difference being that one would simply get a wider FoV with the D800 at the same focal lengths used.
-
I know I'll get jumped on for this, but I think removing a well-designed a.a. filter rarely results in much more real detail. Instead, it allows the generation of false detail beyond the nyquist limit, which gives the impression of sharpness. Aliasing looks very sharp. And at the resolutions we use now, nobody can tell the difference in many landscape images. But that "extra" detail just may not be in the original scene at all. This article explains the issue quite well, I think: http://www.dvxuser.com/articles/article.php/20.
Right now, even the highest resolution bayer sensors fail to fully exploit the abilities of decent lenses. Ctein has pointed out that it will take sensors with resolution in the hundreds of megapixels to actually accomplish that. In the meantime, there are two strategies people use to get the look of ultimate sharpness--both of them relying on digital artifacts and processing.
One is to remove the a.a. filter, and accept some aliased edges and false details that look sharp. These artifacts are often indistinguishable from actual detail in, say, landscape photography. The second strategy is to cut off most detail beyond nyquist with an a.a. filter, and use advanced sharpening techniques that try to recreate what the lens saw. If pushed far enough, deconvolution sharpening not only restores sharpness, but introduces a mist of grain and other digital artifacts that make the file look super sharp. (Just being honest, here.)
Personally, I prefer the second strategy. Deconvolution sharpening gets better all the time. And I would rather have control over the process than accept whatever artifacts the sensor gives me. Not just moire, but distorted edges and other inventions of the sensor and camera processing.
Pretty much every digital file made without an a.a. filter has aliasing, baked in. Just check the resolution charts of your favorite lenses. They all show false color and false detail--even those taken on cameras with a weak a.a. filter. It's fine if you like how the file looks, but we shouldn't confuse this aliasing with actual resolution.
I think that in ten years people will look back on the whole a.a. debate as a quaint vestige of digital growing pains. In the meantime, we should keep it real. Digital cameras without a.a. filters record a lot of false data. If we want to incorporate that into the look of our images, fine. But it's still there. For the most part non-a.a. cameras don't reveal more resolution. They create it.
Now I'll go hide in my bunker to wait for the incoming flames...
I wonder if this even matters, unless you are a forensic or other scientific photographer who requires 'real' details. By the time we have run an image through photoshop even with only a few tweaks it is not a 'real' representation of reality anyway.
I wonder if this even matters to most clients/customers. I've exhibited A3+ prints from 6-12MP cameras and listening to people talking about the prints (without making them aware I am the photographer) people have often volunteered how 'sharp' they look, and that's on matte papers.
Whilst technically I think the D800(E) is amazing, I just can't see that many people actually need 36MP.
I agree with what you say, the linked article was really informative too, I just don't understand the relevance of the discussion in most photographic contexts. Sampling, aliasing, quantisation etc are relevant in many fields and in creative audio I can see how the artefacts will be potentially problematic, but as a genuine question, why does this even matter to most photographers? Except as a fun discussion:-)
As a second question, does aliasing introduce detail or just smear what is there? (I think I understand the concepts of sampling, aliasing and bayer matrix construction and understand how sampling is an issue for example in audio or scientific measurements, but can't get my head round how it applies at the camera sensor/RAW conversion level.) Any explanation or links would be good thanks.
-
The keyword is angular velocity.
If you have a horizontal resolution of lets say 4000 px for a 12 MP camera and a lens with a viewing angle of 20° (FF 100 mm equivalent) you will get problems at least when camera movements of 20°/4000px = 0.005° (one pixel wide) get recorded, probably earlier. With a shutter speed of 1/100 s, which would be in the traditional "safe" zone a horizontal camera movement of 0.005°/(1/100s) = 0.5°/s would cause one pixel wide movement blur, which would result in half the resolution or a fourth the MP, here 12MP-> 3MP. To me this appears to be a pretty slow angular speed to cause movement blur. If the resolution gets higher the problem gets worse with the square root of the resolution gain in MP. So 4 times the resolution (here from 12 to 48 MP) would only "need" the 0.25°/s angular speed to quarter the MP.
-
Sampling, aliasing, quantisation etc are relevant in many fields and in creative audio I can see how the artefacts will be potentially problematic, but as a genuine question, why does this even matter to most photographers? Except as a fun discussion:-)
Aliasing at capture usually does not affect audio recordings in a perceptable way, at least not for sensible equipment constructed the last 20 years.
Aliasing at capture can affect image recordings in a perceptable way, especially for aa-less cameras.
As a second question, does aliasing introduce detail or just smear what is there? (I think I understand the concepts of sampling, aliasing and bayer matrix construction and understand how sampling is an issue for example in audio or scientific measurements, but can't get my head round how it applies at the camera sensor/RAW conversion level.) Any explanation or links would be good thanks.
When sampling with significant aliasing, you capture "something". This something is a function of the scene and the camera, but it is ambiguous: two (or very many) quite different scenes can generate the exact same raw files. As the raw developer have no other information than the raw files, which of those scenes would you like it to render?
I think that map-making is a good analogy. Imagine making a topology map that is to be represented as 1km x 1km squares ("pixels"). How would you like to calculate each squares value? Measure the elevation above sea level by placing you GPS or something similar at the exact mid of each square? Or would you rather do it by averaging the elevation of all points inside the square? Or would you want to smooth even more?
-h
-
Aliasing at capture usually does not affect audio recordings in a perceptable way, at least not for sensible equipment constructed the last 20 years.
Aliasing at capture can affect image recordings in a perceptable way, especially for aa-less cameras.When sampling with significant aliasing, you capture "something". This something is a function of the scene and the camera, but it is ambiguous: two (or very many) quite different scenes can generate the exact same raw files. As the raw developer have no other information than the raw files, which of those scenes would you like it to render?
I think that map-making is a good analogy. Imagine making a topology map that is to be represented as 1km x 1km squares ("pixels"). How would you like to calculate each squares value? Measure the elevation above sea level by placing you GPS or something similar at the exact mid of each square? Or would you rather do it by averaging the elevation of all points inside the square? Or would you want to smooth even more?
-h
Thanks for reply, I need to think about this a bit more:-)
-
I think that map-making is a good analogy. Imagine making a topology map that is to be represented as 1km x 1km squares ("pixels"). How would you like to calculate each squares value? Measure the elevation above sea level by placing you GPS or something similar at the exact mid of each square? Or would you rather do it by averaging the elevation of all points inside the square? Or would you want to smooth even more?
This is maybe a difference between the theoretical case of aliasing when sampling is measurement at single discrete instants in time, or single points in space [your "midpoint of the square"] vs the case with photography, which is more like averaging light levels over each photosite [your "averaging the elevation of all points inside the square"]. Isn't that some kind of low pass filtering in itself?
But color filter arrays mess the simple view up, and maybe that sampling of each color over only 1/2 to 1/4 of the area is the main villain. Could examples of luminosity aliasing in nearly monochrome subjects be due mainly to the luminosity values given by demosaicing being based mainly on data from green pixels, so that there are gaps in the spatial coverage of those "luminosity" measurements?
-
This is maybe a difference between the theoretical case of aliasing when sampling is measurement at single discrete instants in time, or single points in space [your "midpoint of the square"] vs the case with photography, which is more like averaging light levels over each photosite [your "averaging the elevation of all points inside the square"]. Isn't that some kind of low pass filtering in itself?
But color filter arrays mess the simple view up, and maybe that sampling of each color over only 1/2 to 1/4 of the area is the main villain. Could examples of luminosity aliasing in nearly monochrome subjects be due mainly to the luminosity values given by demosaicing being based mainly on data from green pixels, so that there are gaps in the spatial coverage of those "luminosity" measurements?
Yes, integrating the signal over a square is lowpass filtering the signal. It is not a very efficient lowpass filter, though. For a monochrome sensor with 100% fill-rate or perfect microlenses, the "integrate all light within a pixel" idea might be right. For a color-filtered sensor with non-perfect micro-lenses and <100% fill-factor, it is not right.
It is a question of degree: how much aliasing and how much passband blurring will there be if I do operation "X" on my camera/scene-combination. Removing the AA-filter will (everything else equal) tend to increase aliasing, and increase passband sharpness.
-h
-
Thanks "h", that seems to confirm my rough reasoning that the use of a color filter array makes aliasing far worse. This is supported I suppose by the worse aliasing hat happens in video made from still sensors where only a selection of the photosites are read at all, so that it is closer to the "point samples" considered in the simplest mathematical models of sampling and aliasing.
So it would be nice if someone could produce an "X3" technology (all color information measured at each spatial location) which works better that the somewhat flawed, noise prone, Foveon implementation. Then the AA filter would be far less needed, or could at least have a lighter touch. I have read of patents on several alternative approaches to X3 from several major sensor makers, using stack of color filters and such, but at most these have been deployed in small, special purpose sensors.
-
Thanks "h", that seems to confirm my rough reasoning that the use of a color filter array makes aliasing far worse. This is supported I suppose by the worse aliasing hat happens in video made from still sensors where only a selection of the photosites are read at all, so that it is closer to the "point samples" considered in the simplest mathematical models of sampling and aliasing.
So it would be nice if someone could produce an "X3" technology (all color information measured at each spatial location) which works better that the somewhat flawed, noise prone, Foveon implementation. Then the AA filter would be far less needed, or could at least have a lighter touch. I have read of patents on several alternative approaches to X3 from several major sensor makers, using stack of color filters and such, but at most these have been deployed in small, special purpose sensors.
I agree that a Foveon-type sensor would be less prone to aliasing-induced artifacts at a given sensel-pitch, but it will still have luminance aliasing.
I believe that spatially, if sensel density can be some factor X higher, much the same characteristic can be achieved using traditional Bayer technology.
I speculate that this is the reason why we don't see these exotic designs available; it is simply easier and less expensive to shrink current methods, than it is to do revolutionary things at sufficiently small spatial scale and modest cost. Perhaps this trend will continue until we hit some hard quantum law?
I am only adressing spatial behaviour here, noise, saturation, color response etc are also interesting.
-h
(edit: fix my quotes)
-
I agree that a Foveon-type sensor would be less prone to aliasing-induced artifacts at a given sensel-pitch, but it will still have luminance aliasing.
I believe that spatially, if sensel density can be some factor X higher, much the same characteristic can be achieved using traditional Bayer technology.
I speculate that this is the reason why we don't see these exotic designs available; it is simply easier and less expensive to shrink current methods, than it is to do revolutionary things at sufficiently small spatial scale and modest cost. Perhaps this trend will continue until we hit some hard quantum law?
I am only adressing spatial behaviour here, noise, saturation, color response etc are also interesting.
-h
Is this why some of the older MF backs used to have a 3 shot mode? I think that was the name, 3 shots were taken with a different colour filter in front of the sensor, then the shots combined.
-
I agree that a Foveon-type sensor would be less prone to aliasing-induced artifacts at a given sensel-pitch, but it will still have luminance aliasing.
I believe that spatially, if sensel density can be some factor X higher, much the same characteristic can be achieved using traditional Bayer technology.
I speculate that this is the reason why we don't see these exotic designs available; it is simply easier and less expensive to shrink current methods, than it is to do revolutionary things at sufficiently small spatial scale and modest cost. Perhaps this trend will continue until we hit some hard quantum law?
I am only adressing spatial behaviour here, noise, saturation, color response etc are also interesting.
-h
I hope some day we can not only count photons at a photosite but also measure the frequency of the light doing away with color filters, years ago there was an idea of small piezoelectric "spikes" for photosites these would vibrate at the frequency of the light that was hitting them. If you could read out the frequency and voltage for each "piezo-spike" the problem would be solved.
Marc