Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: EgillBjarki on January 17, 2009, 09:46:24 pm
-
According to Phase One my P30+ digital back has 12 f stops dynamic range.
Dxomark measures Nikon D3x dynamic range at 12.84 f stops.
The current top back from Phase One, the P65+ has 12.5 f stops according to them selfs.
I ask my self... Why do I have a big, slow Hasselblad H2 with a P30+ on it when the Nikon D3x is out there?
I know the P30+ has a slightly bigger sensor, little more resolution and the images are a little sharper straight out of the back but thats about it?
D3x has fast focus, really good ISO, grate screen, more robust built and MORE DYNAMIC RANGE?
One could argue that the lenses for medium format cameras have more resolution power, I'm sure thats true...
But why should I stay with medium format? What do you guys think?
-
According to Phase One my P30+ digital back has 12 f stops dynamic range.
Dxomark measures Nikon D3x dynamic range at 12.84 f stops.
The current top back from Phase One, the P65+ has 12.5 f stops according to them selfs.
I ask my self... Why do I have a big, slow Hasselblad H2 with a P30+ on it when the Nikon D3x is out there?
I know the P30+ has a slightly bigger sensor, little more resolution and the images are a little sharper straight out of the back but thats about it?
D3x has fast focus, really good ISO, grate screen, more robust built and MORE DYNAMIC RANGE?
One could argue that the lenses for medium format cameras have more resolution power, I'm sure thats true...
But why should I stay with medium format? What do you guys think?
Did you measure the dynamic range of your P30+ with the same tool you used on the D3x (if possible?) That would be the correct comparison in my opinion. I find this difficult to believe frankly.
-
Egill,
apart from other aspects, there are no objective comparisons between MFDBs and DSLRs, AFAIK. The DR published by PhaseOne is useful for exactly nothing.
Why don't you make some shots of some graduated grey wedge (or whatever color), or something like a color checker card, but underexposed four, five, six stops and publish the raw files to make comparisons possible? (VERY even illumination and CLEAN card or whatever is essential.)
-
For what it's worth, I think DxOmark is consistently about two stops optimistic (across a number of DSLRs I know well). I have no idea how realistic Phase's number is, but I've seen numbers in the 11.5-12 stop range for MFDBs in enough places that I'd tend to believe them. I would call the D3x (which I own and think is a wonderful camera) approximately an 11 stop camera. It wouldn't surprise me at all if an MFDB had somewhere between 1/2 and 1 stop of extra dynamic range over a D3x. On the other hand, you list all the reasons I bought my D3x (when the other camera in the running was an H3DII/31). The resolution of a D3x will make a heck of a 24-inch wide print - mine has inspired me to be in the market for a 24-inch printer!
I think the market for MF has retreated to the rarefied air above 50 MP - the D3x (and, I'm sure, forthcoming competitors from Canon and maybe others) is so good that it's hard to justify buying a 31 or even 39 MP MF system for more than twice the price any more. Fortunately, from the MF makers' viewpoint, "35mm" DSLRs are running out of room to cram more pixels on their sensors - a number of physical limits converge in the neighborhood of 25-35 mp, while those same limits will let MF have between 50-70 mp on a 36x48 mm sensor, and somewhat more than that on a P65+ size sensor...
-Dan
-
For what it's worth, I think DxOmark is consistently about two stops optimistic (across a number of DSLRs I know well).
Consistent is the key word. It's the relative difference and not the absolute number that counts in this respect of DR where subjective elements play a part.
If we compare 3 cameras, say the Canon 5D, the Nikon D700 and the D3x, we find the DXO results for maximum DR are 10.77EV for the 5D, 11.85EV for the D700 and 12.84EV for the D3X.
The dynamic range of the D3X is a whole 2 stops better than that of the 5D. If we make an 8x12" print from both images, the D3X print will have 2 & 1/2 stops more DR than the print from the 5D. Wow!! Phwoar!! That's substantial. That's progress! Hurry up the D700X! .
-
Specs don't mean anything; I'd test both and see what you like better. If the D3x works out better for you, why not go with it?
John
-
Specs don't mean anything; I'd test both and see what you like better. If the D3x works out better for you, why not go with it?
John
Nonsense! Manufacturers' specs may be suspect, but tested results from a reputable company mean something. However, it's always advisable to do one's own tests, but unless you own the equipment or are able to hire or borrow the equipment, it's not possible. In addition, there's always the factor that many photographers are not necessarily competent to do precise comparisons, as is evidenced by the total confusion that's reigned so far in respect of the capabilities of the D3X.
-
My point is that if you do a test and your eyes don't see the difference, who cares what the specs say? We can argue the numbers til we're blue in the face, but in the end it's all about the actual results that you see on paper (or on screen if that's your final result). Besides, there's other limiting factors to consider like the DR of your output device or the accuracy of the output profile or the fact clients won't notice the difference even with a magnifying glass.
Sometimes I read this forum and wonder whether anyone does any actual photography instead of just endlessly debating technoblabber.
John
-
The problem with DR measurements is that the ISO standard or definition isn't all that useful for photographers. Most of the earlier canon's seemed to have about 11.5 or 12 stops DR but when you looked at what DR was useful with something like Imatest (scientific image testing and analysis software) the DR in the useful range was something like 8 or 9 stops (at base ISO). The MFDB's on the other hand really do get like 11-12 stops useful DR at base ISO. I tested the canon 1D3 with Imatest and my Leica DMR and the leica had a 1.5 stop advantage in DR over the 1D3 and though I haven't tested my p20 it has what appears to be the same jump over my DMR. The gap is closing but the MFDB's still have more.
-
That's interesting!!!
I want to check DR of my Sinarback, and Black/White film, and a Slide film, to better understand the real difference between these medias!
I don't have the software that you use to check DR.
Can I test DR using, suppose, a Gray Card and calibrated NEC Wide Gamut monitor, and calibrated Nikon 9000 scanner?
Thanks!
-
The easiest way to have a look is to shoot a scene with shadows with the same crop. Expose to get the highlights at right without cropping and then look at the shadows. How much detail can you see in each? That works okay for film too but if you are comparing digital you can lift the shadows and see how much you can lift them without getting blobs, banding, or whatever. You can also check where the shadows are on the histogram.
You can't test high DR cameras with a grey step wedge as they DR of those is not high enough. You need to use a transmissive step wedge (like negative film) to get enough DR. I bought mine from Stoufer and it's rated at something like 13.5 stops. I think the printed grey step wedges are only good for 9 stops.
A good monitor has a contrast ratio of 700::1 or even now I have seen 1000:1 so that's 9 or almost 10 stops. 2^9=512 and 2^10=1024 I don't know how the translation works from camera to monitor - someone will hopefully step in here....
-
Feel free to unload your P30+ to me!
-
Eric, thank!
-
So the point being, "Numbers are not everything, see for your self".
I guess the digital back is still as good as it was when I bought it... But I'm sure you understand my thoughts.
I have the Nikon D700 as a secondary camera, maybe the best solution is to wait a year for the D700x and be happy with my gear.
All very helpful thoughts, thanks for pinching in yours!
Best regards from Denmark,
Egill Bjarki
-
The problem with DR measurements is that the ISO standard or definition isn't all that useful for photographers
Measuring the DR has nothing to do with ISO. The ISO is relevant only so far as higher ISO reduces the DR. However, if the question is the maximum DR, for example for a landscape, then it does not matter which camera at which ISO has the highest DR; the only thing matters is, how high that DR is.
Most of the earlier canon's seemed to have about 11.5 or 12 stops DR
What does it mean "seemed to have"? I have never seen claimed this nonsense.
The MFDB's on the other hand really do get like 11-12 stops useful DR at base ISO
In your imagination. Although the concept of "useful DR" is subjective, I can't imagine any criteria leading to 12 stop useful DR with any present MFDB.
-
Sometimes I read this forum and wonder whether anyone does any actual photography instead of just endlessly debating technoblabber
Sometimes I read this forum and wonder if luddites do any actual photography or are only complaining about topics they don't understand.
-
Hi,
My point would be is that you should keep the gear you have. Lot of people on these forums think that MF has real benefits. Many users are overly enthusiastic when confronted with a new technology. Keep your stuff, shoot with Hassy and Nikon, if you discover that you don't really need the Hassy than you may sell it off. No one but you can make that decision.
BTW, I agree on the numbers and measurement stuff. I'm an engineer by occupation and also a bit scientifically minded, so lab data is interesting for me. Measurements are interesting and relevant, but we need to put them in perspective. In my view the best slide film, ever, was Velvia 50. Velvia had in my experience a DR of about 5 steps and was absurdly hard to scan, I loved it nevertheless. DR of ten plus steps is impressive, but I'm not really sure it matters.
I also feel there is some contradiction in the conventional wisdom that MFDBs have high DR but get noisy at high ISOs, it simply doesn't make sense to me. In my simple view DR and high ISO capability should go hand in hand.
Best regards
Erik
My point is that if you do a test and your eyes don't see the difference, who cares what the specs say? We can argue the numbers til we're blue in the face, but in the end it's all about the actual results that you see on paper (or on screen if that's your final result). Besides, there's other limiting factors to consider like the DR of your output device or the accuracy of the output profile or the fact clients won't notice the difference even with a magnifying glass.
Sometimes I read this forum and wonder whether anyone does any actual photography instead of just endlessly debating technoblabber.
John
-
I want to check DR of my Sinarback, and Black/White film, and a Slide film, to better understand the real difference between these medias!.
Can I test DR using, suppose, a Gray Card and calibrated NEC Wide Gamut monitor, and calibrated Nikon 9000 scanner?
1. You can not measure the DR of the film on existing shots, because it is not linear.
2. Re your Sinarback: the grey card is not good (you would need lots of shots). However, a grey (or whatever color) wedge, like a Kodak or Stouffer wedge, or a clean color checker card are very suitable. Underexposed three stops, very evenly illuminated, they give the different levels to measure on.
Here is how the measurements look like (this is actually a demo showing that the noise is raw channel independent). The red is at the intensity -5.17 EV (from saturation), which is not really dark, but this is at ISO 3200. The noise is 14.4%. The green sample is at -5.15 EV, the noise is 14.34%. The blue sample is at -5.13 EV, the noise is 14.07%.
(http://www.panopeeper.com/Demo/Noise_ColorIndependence_Red.GIF)
(http://www.panopeeper.com/Demo/Noise_ColorIndependence_Green.GIF)
(http://www.panopeeper.com/Demo/Noise_ColorIndependence_Blue.GIF)
-
I understand that you already have P1. In my opinion , in general do not compare this equipment. If you compare two shots , from nikon and from p1. You will see that Nikon files are thought blurred. Dynamic range... well the only theory.
Nikon obviously is a gra camera, but not replace MFDB. Sorry...
If you want to change the equipment I recommend the new H3DII 50
Best regards,
Design Freak
-
Measuring the DR has nothing to do with ISO. The ISO is relevant only so far as higher ISO reduces the DR. However, if the question is the maximum DR, for example for a landscape, then it does not matter which camera at which ISO has the highest DR; the only thing matters is, how high that DR is.
Panopeeper, Eric is not referring to the camera's sensor gain ISO control, he's referring to the International Standards Organization standard on noise measurements in photography/electronic still-picture imaging, or ISO standard 15739:2003 (http://www.iso.org/iso/iso_catalogue/catalogue_tc/catalogue_detail.htm?csnumber=28835). This standard does contain a definition for dynamic range (in addition to definition and measurements of noise) for digital still cameras.
And Eric is correct--the standard's definition for dynamic range for digital still cameras has signficant flaws. In fact, I just googled the standard and found an excellent paper on both the standard, and its flaws relative to digital still cameras. You can read it at: http://doug.kerr.home.att.net/pumpkin/ISO_Dynamic_range.pdf (http://doug.kerr.home.att.net/pumpkin/ISO_Dynamic_range.pdf). Doug does a superb job covering this, so I'll leave it at that.
Egill, in the end, I'll second both Juanito's and Eric's advice (third their advice?). The specs are only meaningful if you have a lot more information than Phase is providing. As Eric suggested, test with a transmissive wedge or even a contrasty scene (much easier) and trust your eyes.
Remember, by the time you see an image from your raw file, the image may have gone through any number of gamma conversions, level adjustment and subjective (by your raw developer manufacturer) tweaks. Reconstructing the imaging chain to get back to the raw data is not trivial, if you want to understand how the raw data maps to real-world quality of your images.
Just doing a side-by-side will save you more than a little time, and tell you what you really want to know, particularly if you begin lifting the shadows and looking for artifacts, again as Eric suggests.
Hope that helps,
Brad
-
Feel free to unload your P30+ to me!
The interesting thing about open forums is you have so many different types of photographers. Some shoot locked down on a tripod, some are amateurs, so shoot professionally on deadline.
The interesting thing about all of these tests is they don't really equate to real world work.
I've seen (and must admit have done) iso tests where I set up studio strobes and work 200 to 800 iso comparing noise. The thing is I would never use 800 iso in the real world this way. The same with testing the dynamic range of the capture.
When I need high dynamic range and high iso it's because the conditions aren't perfect. The light is usually a mix of kelvin temperatures, backlit, or window light, the glow from a restaurant, and in those conditions any camera I've used at 800 iso has a lot of stuff happening in shaodws, but then again so did any film I used in those conditions, so I guess it's a wash.
You can't open the internet and click on d3x or H3d or any camera and not get reviews that run the gamut. "The D3x replaces medium format, or noting replaces medium format and usually there is some 200% pixel crop of a wall, or eye to prove it.
But obviously we all have different needs and circumstances, which is why one person says "my p25 shoots great 800 iso and the next person says they wouldn't dare move a medium format back of 200 iso.
Once again, if your a professional working on deadline it just doesn't work that way these tests show. When you have 45 seconds of backlit direct sunlight left in the day, the production brief calls for interactivity and movement then the camera that is useful in that situation is the one that will hold focus and allow you to get the shot.
This doesn't mean that a D3x is better or worse than an H3d, or a P45 is not a good camera, it just means that the only way you can compare is under pressure in the actual conditions you shoot and at the end of the day, nobody you shoot for is usually looking at dynamic range that is 4% bettter, or if there is some chorma noise in the shadows. They look to see if you caught the emotion, got the frame needed and if it's in focus.
I know deep in my heart that the cameras are only about 15% of the equation on shooting a pretty photograph. We can all talk this to death, shoot 200 color charts, compare noise and DR until our eyes bleed, but when you sit down at the computer to build that first web gallery I can promise you image format, file size, DR are the last thing you will be looking at.
Medium format does have a place, though in reality the dslrs are getting very close to covering their territory, but for most of us we have and always will shoot multiple formats.
A medium format camera with a ccd is not going to focus as quickly as a dslr, offer the same versatility, or be as cost effective. A dslr just doesn't tether as easily as a most medium format backs, doesn't have that ultra crisp sharpness out of the can (as long as it's in focus).
Still, once you get down to the post production of a major campaign, the camera gets forgotten very quickly.
Go to this site, http://www.portusimaging.com/ (http://www.portusimaging.com/) and click on portfolio, then case studies and tell me which image was shot with which camera and once you get to final, which camera made a difference.
-
My point is that if you do a test and your eyes don't see the difference, who cares what the specs say? We can argue the numbers til we're blue in the face, but in the end it's all about the actual results that you see on paper (or on screen if that's your final result). Besides, there's other limiting factors to consider like the DR of your output device or the accuracy of the output profile or the fact clients won't notice the difference even with a magnifying glass.
Sometimes I read this forum and wonder whether anyone does any actual photography instead of just endlessly debating technoblabber.
You heretic, fancy suggesting people take photos to see which camera is best! The cheek of it.
Sometimes I read this forum and wonder whether anyone does any actual photography instead of just endlessly debating technoblabber.
Sometimes I read this forum and wonder if luddites do any actual photography or are only complaining about topics they don't understand.
I'd say that suggesting people actually look at real images to test a camera is simply commonsense.
-
Sometimes I read this forum and wonder if luddites do any actual photography or are only complaining about topics they don't understand.
Fair enough: I don't have an iphone nor am I Facebook friendly. But I've been a working pro for many years now and I shoot with everything from small format digital to MFDB to large format film. My goal as a photographer is to create images that meet the various needs of my clients and to speak to my own artistic needs. I really don't care about the published specs so long as the camera does what I need it to.
I don't view the MFDB process or images as somehow sacrosanct to the point that they can't be replaced by small format. I'm sure there's all kinds of technical data to support whatever position you want, but if your eyes don't see the difference, it doesn't really matter. Seems to me that you can debate the topic all you want or just rent the dumb thing and see for yourself. I guess I'm still analog in that way...
John
-
My goal as a photographer is to create images that meet the various needs of my clients and to speak to my own artistic needs. I really don't care about the published specs so long as the camera does what I need it to.
It's not necessarily the published specs that count but the tested specs in relation to another camera using the same methodology. Without test results using the same methodology, one has to rely upon subjective opinions which tend to vary considerably simply because they are subjective.
If you are perfectly happy with the performance of your camera, why change it? If you are not perfectly happy and see room for improvement, whether in respect of resolution, noise at high ISO or dynamic range, then competent and thorough test results of various models of camera should give you a clear indication as to which upgrade will suit your purposes, provided you are capable of interpreting the results.
If I were able to hire any piece of equipment at a reasonable price, whether camera body or lens, I would do so and test the equipment myself. Unfortunately, often it simply isn't possible to hire new technology until it's been on the market for many months or even years, and certain lenses and equipment might never be available for hire in one's area.
The other issue, of course, is that thorough, careful and meticulous testing can take a lot of time if you are not already set up and prepared for such testing. You might spend many hours or even days to discover the sort of information that is already available on the DXOmark site, such as camera A has a better DR than camera B.
-
I made a test with a Mamiya with Phase P30 and 150mm AF (not the last generation lens) next to a Sony a900 with Zeiss 135mm 1.8 both shots at about f8 with strobes.
Post-processed both files to my liking on order to get what I think is the best result for both. I printed then on A3+ size.
I showed the prints to a photographer that never used a DB on his life and she could not tell which was the better one. I showed the same prints to two art directors that normally are very anal and works with photographers that use DB only, and they picked the P30 prints, without needing to put their nose on it, in a matter of few seconds.
I could tell the difference pretty easily, maybe because I was looking at colors and gradients more than detail. though difference of crispiness is quite tangible. Colors were generally more spot on for the Phase back, though 1 or 2 colors were closer to the original on the a900 file.
The Phase also had less CA, thought the CA on the a900 was so small that showed only on the monitor and not on the print. Gradient were better on the Phase as well. The Phase file obviously could take more abuse on the curves or exposure slider.
The a900 had the best lens it can fit at this time, probably using the new mamiya 150mm (the 3.400$ one) the difference would have been more visible.
In any case the differences where very minimal and I guess even most of the clients will not spot the better one. In any case that is not a good enough reason to ditch the Phase back for a DSLR and not being able to please some clients that have higher expectations.
Two different tools for different kind of photography. If someone can afford both, better keep both of the systems.
-
I made a test with a Mamiya with Phase P30 and 150mm AF (not the last generation lens) next to a Sony a900 with Zeiss 135mm 1.8 both shots at about f8 with strobes.
Post-processed both files to my liking on order to get what I think is the best result for both. I printed then on A3+ size.
I showed the prints to a photographer that never used a DB on his life and she could not tell which was the better one. I showed the same prints to two art directors that normally are very anal and works with photographers that use DB only, and they picked the P30 prints, without needing to put their nose on it, in a matter of few seconds.
I could tell the difference pretty easily, maybe because I was looking at colors and gradients more than detail. though difference of crispiness is quite tangible. Colors were generally more spot on for the Phase back, though 1 or 2 colors were closer to the original on the a900 file.
The Phase also had less CA, thought the CA on the a900 was so small that showed only on the monitor and not on the print. Gradient were better on the Phase as well. The Phase file obviously could take more abuse on the curves or exposure slider.
The a900 had the best lens it can fit at this time, probably using the new mamiya 150mm (the 3.400$ one) the difference would have been more visible.
In any case the differences where very minimal and I guess even most of the clients will not spot the better one. In any case that is not a good enough reason to ditch the Phase back for a DSLR and not being able to please some clients that have higher expectations.
Two different tools for different kind of photography. If someone can afford both, better keep both of the systems.
Very helpful input and highly relevant!
I could also imagine that it takes a pro to spot the diff. between high image quality in high end medium format digital cameras and high end 35mm DSLR.
Is it worth it? Different from person to person I guess?
I'm keeping my Hasselblad H2 and Phase One P30+ AND the Nikon D700 and stop thinking about it for some time
-
Panopeeper, Eric is not referring to the camera's sensor gain ISO control, he's referring to the International Standards Organization standard on noise measurements in photography/electronic still-picture imaging
Well, if he meant it that way, that's different, of course.
I find the way of measuring of noise as the standard deviation on pixel values useful (actually, that's how I am doing it). However, the ISO defines SNR=1 as the limit of the useful dynamic range, and that is plainly ridiculous; I guess some Microsoft employees have participated in the respective committee.
The acceptable level of noise varies in at least two dimensions: personal preference and limits dictated by the situation. I do not accept any perceivable noise in a daylight landscape shot in sunshine from tripod. On the other hand, those shooting in night concerts without flash light and tripod will put up with much noise.
Back to the SNR=1: I don't know if anyone here has an idea, how much noise that means. I certainly did not, so I created a small demo; everyone can decide for himself/herself if this criterion of the ISO specification is useful.
The first capture shows an area of an image with noise = 97.29%, as close as I could find something to SNR=1. I isolated the red raw channel, for the blue channel exhibits much higher level of noise. The selected area, the cedar shingle roof of a house (marked by an orangy rectangle) does not show any details. So much to SNR=1.
The second capture is from a much higher esposed shot of the same scenery; still very noisy, the red channel is at 36.46% (SNR~3), but it shows, that the roof does contain details, which are totally lost with SNR=1.
-
Go to this site, http://www.portusimaging.com/ (http://www.portusimaging.com/) and click on portfolio, then case studies and tell me which image was shot with which camera and once you get to final, which camera made a difference.
Very interesting, thanks for sharing this!
Cheers,
Bernard
-
deleted - thanks Brad for getting there first and explaining that confusing terminology for ISO (International Standards) vs ISO/ASA.
-
... the ISO defines SNR=1 as the limit of the useful dynamic range, and that is plainly ridiculous ...
I would say that it is not ridiculous, but simply a standard engineering meaning of "dynamic range" that is different from (always more than) what photographers mean these days when they talk about "dynamic range". It might be better if photographers had stuck with more traditional wording like "stops of subject brightness range that can be handled with adequate IQ in the shadows" or maybe "exposure latitude".
Anyway, for an engineering standard, dynamic range relates to the ratio between maximum signal and noise floor, and the latter is where SNR=1.
P. S. added later. The ISO standard does not use the word "useful" which appears in Panopeepers's comment above. It simply defines dynamic range (and not using measurements at the level where SNR=1, but at a level about 1% of maximum signal strength.)
The problems seem to arise when an objective factual measurement of a technical characteristic, intended mainly for designers of cameras not users of them, gets misinterpreted as being a direct measure of something for which a precise objective standard is not [Edit: added "not"!] possible, like how many stops of the output range from highlights down is "useful". Useful to whom, in what circumstances? My rule of thumb for artistic photography (not technical work like astronomy) would be to set a lower limit at about SNR=5 to 10, so declare the bottom two or three stops to be of little or no use.
-
It might be better if photographers had stuck with more traditional wording like "stops of subject brightness range that can be handled with adequate IQ in the shadows" or maybe "exposure latitude".
Unfortunately, terms like 'adequate IQ in the shadows' are still too subjective to be completely meaningful, so perhaps the engineering standard for DR should be the one which is always used. I don't see that it matters if the figures seem inflated in relation to usable stops of subject brightness range, as long as they are all inflated to the same degree.
Perhaps the terminology translated into practical figures for the photographer could then be something like, (1) For the minimum standard of 'almost accpetable' shadow noise for the 'rough and ready', reduce the engineer DR figure by the equivalent of one stop, (2) For an everage acceptable standard, according to market research in which photos containing shadow noise were exhibited to 100,000 participants, reduce the engineering DR figure by the equivalent of 2 stops. (3) For the fussy, meticulous and anally retentive, reduce the engineering DR figure by 3 stops.
One might think of better descriptions for those three categories, but you get the idea .
-
I would say that it is not ridiculous, but simply a standard engineering meaning of "dynamic range" that is different from (always more than) what photographers mean these days when they talk about "dynamic range
he's referring to the International Standards Organization standard on noise measurements in photography/electronic still-picture imaging, or ISO standard 15739:2003. This standard does contain a definition for dynamic range (in addition to definition and measurements of noise) for digital still cameras
This standard is about digital imaging, not about designing sensors. Plus, I have not mentioned it in the previous post: in fact the SNR=1 case is non-existent, because at that level many pixels are zero, representing negative pixel values as well; this renders the calculation of standard deviation a mockery.
-
I don't see that it matters if the figures seem inflated in relation to usable stops of subject brightness range, as long as they are all inflated to the same degree.
See my previous post: the "inflated to the same degree" is an illusion, for different cameras and different raw processors handle these low pixel values differently.
Therefor I stop measuring the noise at levels, where zero pixels appear in such number that the result is significantly affected by them. The max. noise level in praxis is 50% (SNR = 2) in my charts (and that is horrendeously noisy already; for example your 5D shows strong horizontal banding at that level).
-
See my previous post: the "inflated to the same degree" is an illusion, for different cameras and different raw processors handle these low pixel values differently.
Therefor I stop measuring the noise at levels, where zero pixels appear in such number that the result is significantly affected by them. The max. noise level in praxis is 50% (SNR = 2) in my charts (and that is horrendeously noisy already; for example your 5D shows strong horizontal banding at that level).
See my edited post above.
-
Very helpful input and highly relevant!
I could also imagine that it takes a pro to spot the diff. between high image quality in high end medium format digital cameras and high end 35mm DSLR.
Is it worth it? Different from person to person I guess?
I'm keeping my Hasselblad H2 and Phase One P30+ AND the Nikon D700 and stop thinking about it for some time
You better. Use it as long as you can afford it and enjoy it, which is what I do and I think that is the only way to do it. No time to endlessly worry over decisions that already have been made.
No use in selling it now anyway, it is next to worthless
-
I would say that it is not ridiculous, but simply a standard engineering meaning of "dynamic range" that is different from (always more than) what photographers mean...
Agreed.
Engineering dynamic range is great measurement. It's easy to understand. It's an easy number to compare over a wide variety of cameras easily. However, it is only a single number, and it only describes a very narrow set of conditions. Despite the simplicity of the number, mistakes often arise due to the failure to account for differences of scale (such as comparing 12-bit ADU to 14-bit, or comparing 10 MP to 15 MP). Even when scale is properly accounted for, people often draw sweeping conclusions that are not supported by this one number.
I suggest that photographers stop drawing conclusions from a single number and at least begin to use a chart. Here is an analogy.
MTF-50 is a great measurement. It's easy to understand. It's an easy number to compare over a wide variety of lenses easily. However, it is only a single number, and it only describes a very narrow set of conditions. Despite the simplicity of the number, mistakes often arise due to the failure to account for differences of scale (such as comparing lp/mm with l/mm, or 50 lp/mm for an APS-C system vs 50 lp/mm for a FF35 system). Even when scale is properly accounted for, people often draw sweeping conclusions that are not supported by this one number.
In both cases, there are better solutions than using a simple number. A picture speaks a thousand words.
MTF, plotted on a chart, says a lot more about a lens than any single number can. Using MTF 50 may misguide you to a lens that drops off sharply after MTF 50, while another coasts smoothly at MTF-45 out to very high resolution.
S/N, plotted on a chart of exposure, says a lot more about dynamic range than any single number can. Using engineering dynamic range may misguide you to a camera that has worse S/N in the midtones and shadows that you will actually use (say, S/N 1:8+).
Emil Martinec uses such a chart here:
http://theory.uchicago.edu/~ejm/pix/20d/te...-p2.html#SNR-DR (http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p2.html#SNR-DR)
-
Yeah,
But how do you say that my curve is better than your curve? Anyone can say that 85.0 is more and therefore better than 84.9, but is a curve better than another? What if the curves cross? Two many questions and simple answers.
Erik
Don't trust complex things to a single number. Use a chart.
-
Engineering dynamic range is great measurement. It's easy to understand. It's an easy number to compare over a wide variety of cameras easily. However, it is only a single number, and it only describes a very narrow set of conditions.
Yes it does. Typically the DR at a particular ISO is a single number expressed in dB or EV or F stops. It relates only to a specific ISO and all comparisons should be made in relation to that specific ISO.
The problems occur when we do not have a well-defined, internationally accepted standard that all testers of cameras adhere to, whether it's ISO sensitivity, DR or S/N.
I'm taking a few liberties here, with respect to the following examples, and I hope Phil Askey does not object (Sorry Phil. I'm British, but I'm also Australian ).
I've been using the 5D for about 3 & 1/2 years and I have great respect for that camera. But I can't deny that the D3 and D700 are better cameras, in most respects. Perhaps the mistake I've made, due to misinformation on the internet, is that these cameras, the D700 and D3, excel to their greatest extent in very high ISO performance. They do excel in very high ISO performance, but not to the extent that they excel at lower ISOs.
However, before comparing cameras' DR or S/N at a specific ISO, it is only sensible to test what the ISO actually is. Here again, one can't trust the manufacturers' rating. They tend to exaggerate.
This tendency for manufacturers (or their marketing departments) to exaggerate the specifications is the reason why indpendent and objective reviewers and testers provide a useful role, hopefully to the benefit of the consumer.
But what happens when the so-called independent reviewers provide results which are also inconsistent just as the manufacturers published specs. How does the consumer handle this situation?
Following are two quite different results of an ISO (exposure) accuracy test of the 5D. Having read the Dpreview review of the 5D a few years ago, I was always under the impression that Canon had understated the actual and real ISO of the 5D. Instead of being ISO 3200, it was actually ISO 4000. I've taken a screen grab of the Dpreview table below showing these results.
[attachment=10987:Dpreview...ating_5D.jpg]
DXOmark is clearly using a different standard to Dpreview. Their ISO results for the 5D indicate that Canon is overstating the true ISO, not understating. In fact, according to DXOmark, most camera manufacturers tend to overstate the ISO, which actually makes sense. It's a natural tendency of salesmanship.
[attachment=10988:DXOmark_..._D700_DR.jpg]
One can see from the above graph that the D700's very high-ISO DR is only marginally better than the 5D. The greatest improvement in DR is at ISO 200. What's remarkable is that the DR of the D700 at ISO 200 is actually at least one stop better than the DR of the 5D at ISO 100. That's a double benefit for those of us who walk around without a tripod. ISO 200 with the D700 gives not only 1 stop greater DR but twice the shutter speed.
In my books, that's a significant improvement beyond the pixel-peeping level. I've ordered a D700. Expect a few DR comparisons with the 5D in the near future.
-
(http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise1d3.gif)
From now on I'm going to take that chart with me on all my shoots, I reckon my clients will be really impressed.
Nick-T
-
Me too Nick. Hope that will impress the heck out of them. Maybe put it on my bag tag.
Just wondering if I can actually charge more for lower signal noise , now that would impress me. LOL
-
Me too Nick. Hope that will impress the heck out of them. Maybe put it on my bag tag.
Just wondering if I can actually charge more for lower signal noise , now that would impress me. LOL
I think I will make the homepage of my website with that chart.
-
i'd like DXO to test the current digital backs... but they make money with DSLR user, so they won't do it... Digital back and DSLR reminds me of the film era war : Leica M6 VS medium format (hassy, flex...) ...