Luminous Landscape Forum
Equipment & Techniques => Cameras, Lenses and Shooting gear => Topic started by: Ellis Vener on November 10, 2014, 02:44:13 pm
-
it's worth a read
http://www.dslrbodies.com/newsviews/do-you-believe-in-dxomark.html
At the end he makes some really interesting comments, ones I happen to agree with
"I guess to end this discussion, I should probably suggest what we Nikon users want to see on the DxOMark tests for the next Nikon. We’ve got it pretty good at the moment, after all. Incremental movement of the dynamic range chart upward would always be welcome, but I think we’re now very close to the point where we want to see something entirely different. We really want to see base dynamic range unbound (e.g. the two-shot type DR that the Apple iPhone 6 is doing, and which Sony’s latest patents all hint at). Beyond that, we want to see flattening of the dynamic range slope as ISO is boosted (less of a hit per doubling of ISO).
Simply put, we’re deep ending into small, incremental gains that aren’t pushing us very much further above the “good enough” bar. What we need is to push that bar far forward again, and that’s going to take another sensor generation/design, I think. Until then, you can all anguish over small differences in DxOMark measurements, but I’d be surprised that those show up anywhere near as clearly in your images. "
-
it's worth a read
http://www.dslrbodies.com/newsviews/do-you-believe-in-dxomark.html
At the end he makes some really interesting comments, ones I happen to agree with
"I guess to end this discussion, I should probably suggest what we Nikon users want to see on the DxOMark tests for the next Nikon. We’ve got it pretty good at the moment, after all. Incremental movement of the dynamic range chart upward would always be welcome, but I think we’re now very close to the point where we want to see something entirely different. We really want to see base dynamic range unbound (e.g. the two-shot type DR that the Apple iPhone 6 is doing, and which Sony’s latest patents all hint at). Beyond that, we want to see flattening of the dynamic range slope as ISO is boosted (less of a hit per doubling of ISO).
Simply put, we’re deep ending into small, incremental gains that aren’t pushing us very much further above the “good enough” bar. What we need is to push that bar far forward again, and that’s going to take another sensor generation/design, I think. Until then, you can all anguish over small differences in DxOMark measurements, but I’d be surprised that those show up anywhere near as clearly in your images. "
I think he hits it pretty good. The 7DmkII does pretty well at higher ISOs, but in the end, the $800 cheaper D7100 is a better bet from a image quality standpoint. He's a little more charitable concerning the progression from 7D to 7DmkII than I would have been. I would expect to see more than just a stop of high ISO performance out of a replacement for an older camera. And considering it is a year or more newer than the D7100 it should have been more competitive with the $850 cheaper D7100. Obviously, Canon is attempting to trade features and speed for their inability to deliver a better sensor. It's up to the purchaser to determine if that is a good trade for them.
I also agree with his final assessment of incremental gains. I hope that pushes all the makers to more usability and extensibility as a competitive area.
-
Obviously, Canon is attempting to trade features and speed for their inability to deliver a better sensor.
Perhaps what's also obvious is that there's more to a camera than the sensor. ;)
Particularly for an action shooter.
-
Perhaps what's also obvious is that there's more to a camera than the sensor. ;)
Particularly for an action shooter.
Considering that Thom is a very accomplished wildlife shooter, I don't think he has ignored that point.
-
Check to see what BIF guys say on Naturescapes.
-
By definition, DR has to go down by 1 stop for each stop of increased ISO. If it doesn't, it's either not a true 1-stop increase in ISO (stated and measured ISO can differ by a huge amount) or the sensor is being polluted by a large amount of read noise (like with Canon sensors at low ISO).
The way to increase DR at high ISO is to improve the capacity of the photon wells, thereby increasing DR at low ISO (which flows through to high ISO).
-
Beyond that, we want to see flattening of the dynamic range slope as ISO is boosted (less of a hit per doubling of ISO).
If your metric of dynamic range is the ratio of full-scale to the average signal level that produces a signal-to-noise ratio of around 10 or so, which happens to be mine, without a radical change to the design of the sensor, there's not much progress to be made here. If he's talking about engineering dynamic range, which is read-noise limited, I do think we can expect some improvement in the near future. The Sony Alpha 7S changes the conversion gain as you go from ISO 1600 to 2000, dramatically lowering the read noise. Designs that offer more granular adjustments to conversion gain might do what he wishes for.
Jim
-
Thom refers to "pragmatic dynamic range," which I guess takes into account his preferred software and workflow. He also mentions exposure blending and notes the 7Dii & D7100 can yield essentially identical results DR-wise via this method. IOW, in scenes exceeding the single-exposure DR of both cameras it doesn't matter which one you use.
-Dave-
-
Hi Jim,
I would suggest that there is a benefit of having low readout noise, and that would be that blacks and near blacks have little impulse type noise. Those areas, Zone 0 in Ansel Adams speak I guess, don't hold detail but would be ugly with colourful salt & pepper noise. It's a mix of red, green and (have you seen?) blue peppers.
Best regards
Erik
Ps. I need to check your blog again, your writing there is even better than your postings here! :-)
If your metric of dynamic range is the ratio of full-scale to the average signal level that produces a signal-to-noise ratio of around 10 or so, which happens to be mine, without a radical change to the design of the sensor, there's not much progress to be made here. If he's talking about engineering dynamic range, which is read-noise limited, I do think we can expect some improvement in the near future. The Sony Alpha 7S changes the conversion gain as you go from ISO 1600 to 2000, dramatically lowering the read noise. Designs that offer more granular adjustments to conversion gain might do what he wishes for.
Jim
-
I would suggest that there is a benefit of having low readout noise, and that would be that blacks and near blacks have little impulse type noise. Those areas, Zone 0 in Ansel Adams speak I guess, don't hold detail but would be ugly with colourful salt & pepper noise. It's a mix of red, green and (have you seen?) blue peppers.
Erik,
You raise a good point. However, if black is what you're looking for, black is pretty easy to get in post. If you want to automate it, you could apply a median filter through a luminance mask. I'm in the middle of a study of the spatial frequency characteristics of read noise. Read noise has in the past had reputation of being ugly, that ugliness stemming from relatively high low-frequency energy compared to the white noise that you get from the photon statistics. One of the things that I'm finding out, at least in the Nikon D810, which is the only camera for which I've fully crunched the data at this point, is that almost all of the low-frequency energy does not change from frame to frame. Computing an average read noise frame and subtracting that from a series of images results in almost all the read noise ugliness disappearing.
More on that as I get more results. Thanks for the compliments about my blog. I do put a little more effort in making the writing in my blog posts stylish. Having no (self-imposed) space limitations also helps.
Thanks,
Jim
-
You may want to try astronomy software that is capable of using bias frames.
http://en.wikipedia.org/wiki/Bias_frame
-
You may want to try astronomy software that is capable of using bias frames.
http://en.wikipedia.org/wiki/Bias_frame
I use Matlab.
-
Jim,
I may agree on that. Something I was thinking about is that raw developers should have a slider for near dar noise reduction.
One of my points is that I am very much in favour of a parametric workflow, where we are working from a raw image and just keep record of the necessary adjustments. Going out to Photoshop increases file size six times and throws away information.
That said, a parametric work flow needs a persistent image format, I don't feel today's raw formats fulfil that requirements, with DNG being a possible exception.
Best regards
Erik
Erik,
You raise a good point. However, if black is what you're looking for, black is pretty easy to get in post. If you want to automate it, you could apply a median filter through a luminance mask. I'm in the middle of a study of the spatial frequency characteristics of read noise. Read noise has in the past had reputation of being ugly, that ugliness stemming from relatively high low-frequency energy compared to the white noise that you get from the photon statistics. One of the things that I'm finding out, at least in the Nikon D810, which is the only camera for which I've fully crunched the data at this point, is that almost all of the low-frequency energy does not change from frame to frame. Computing an average read noise frame and subtracting that from a series of images results in almost all the read noise ugliness disappearing.
More on that as I get more results. Thanks for the compliments about my blog. I do put a little more effort in making the writing in my blog posts stylish. Having no (self-imposed) space limitations also helps.
Thanks,
Jim
-
Hi Jim,
I would say a good tool for engineers, who actually know what they are doing.
I am not really there…
Best regards
Erik
I use Matlab.
-
I would say a good tool for engineers, who actually know what they are doing.
I am not really there…
Erik, don't worry. If something comes of this, I'll come up with a way to tame it for ordinary, but tech-savvy photographers. DCRAW does dark-frame subtraction.
Jim
-
Check to see what BIF guys say on Naturescapes.
Link?
-
Erik, don't worry. If something comes of this, I'll come up with a way to tame it for ordinary, but tech-savvy photographers. DCRAW does dark-frame subtraction.
Jim
Cameras do dark frame subtraction with long exposure noise reduction. But you are talking about bias frames right? Astrophotography programs calibrate a stack of raws with bias frames, even the dark frames for a master dark.
-
Perhaps what's also obvious is that there's more to a camera than the sensor. ;)
Particularly for an action shooter.
Obviously. We are basically buying 2 things when purchasing a camera. The ability to get the images we want and the quality of the data that represents that image. I'm just happy I was forced to develop the technique to get the image with a 6fps camera with better quality at $850 less.
Tip: With a 7fps camera and a baseball batter, it is possible to take a sequence and never get the ball in the frame, let alone compressed on the bat. But with good technique, you can get it with 1 frame! Hint: Don't watch the bat, watch for the hands to come through the hitting zone.
-
Cameras do dark frame subtraction with long exposure noise reduction. But you are talking about bias frames right? Astrophotography programs calibrate a stack of raws with bias frames, even the dark frames for a master dark.
"...even the dark frames for a master dark." I'm not quite getting what your saying. I'm averaging 256 exposures of a dark field to get rid of the frame-to-frame randomness. Then I'm subtracting the averaged image from subsequent ones.
As I understand what LENR on cameras does, it only subtracts one dark frame from another, reducing the static read error, but actually increasing the amount of frame-random noise, since sqrt(firstFrameRandomNoise^2 + secondFrameRandomNoise^2) = 1.414 * firstFrameRandomNoise, if firstFrameRandomNoise = - secondFrameRandomNoise and the two are uncorrelated. I'm doing the average subtracting with short exposures. It doesn't work on all cameras.
Jim
-
Yes exactly. One dark frame actually increase noise but removes unwanted signal. A program like deep sky stacker will work with stacks of different frames. I'm certainly no expert but a typical scenario would be you have 20 (or more) light frames of your deep sky object, then during the night you also grab some 20 dark frames with the same duration, then you take (next day perhaps) some flat field images to remove vignetting. And you also put a lens cap on, fastest shutter speed and grab a series of bias frames. 20+ or 256 in your case. These are averaged together and are used to calibrate all of the other types of frames, basically to remove read pattern noise. Then you end up with master dark and flat field images that you can reuse with different light frames. More serious astrophotographers will keep a library of master darks for different sensor temperatures.
So are you averaging dark frames? I don't know theory and math but I guess the result would be the same more or less. For shorter exposures it might be practical. Dark + bias frames might be used only because it is impractical to grab so many long exposure darks.
-
So are you averaging dark frames? I don't know theory and math but I guess the result would be the same more or less. For shorter exposures it might be practical. Dark + bias frames might be used only because it is impractical to grab so many long exposure darks.
That's right. I haven't finished the testing, but I think I can use the same averaged image at shutter speeds from 1/60 to 1/8000 sec -- the handholding range -- on the D810.
http://blog.kasson.com/?p=7936
The technique doesn't work on the alpha 7S -- not at all at ISO 100, and very weakly at ISO 3200. It appears that there isn't much fixed pattern noise at those shutter speeds with that camera.
http://blog.kasson.com/?p=7971
Jim
-
And you also put a lens cap on, fastest shutter speed and grab a series of bias frames. 20+ or 256 in your case. These are averaged together and are used to calibrate all of the other types of frames, basically to remove read pattern noise.
Right. You never know how many frames you need until you do the averaging. When the standard deviation of the averaged image stops dropping, you've got enough.
Jim
-
Thom refers to "pragmatic dynamic range," which I guess takes into account his preferred software and workflow. He also mentions exposure blending and notes the 7Dii & D7100 can yield essentially identical results DR-wise via this method. IOW, in scenes exceeding the single-exposure DR of both cameras it doesn't matter which one you use.
-Dave-
But doesn't 'blending exposures' defeat the high fps and great action camera selling point for the 7DmkII? That is basically why you are paying $850 more for it in the 1st place. Obviously, the logic would have to be "you get a high speed action camera though with somewhat lower quality output, and when you are shooting other subjects you can use exposure blending to achieve dynamic range."
-
Hi,
My take is that dynamic range is a bit overrated. It depends a bit what you shoot. Personally, I am a low ISO shooter and I feel having good DR is an advantage. Would I shoot high ISO, DR may be less important to me.
But, high DR stuff also works well at high ISO, as high DR is essentially the same a clean signal processing.
If a photographer doesn't use base ISO, the DR advantage disappear and other properties play a greater role.
Best regards
Erik
But doesn't 'blending exposures' defeat the high fps and great action camera selling point for the 7DmkII? That is basically why you are paying $850 more for it in the 1st place. Obviously, the logic would have to be "you get a high speed action camera though with somewhat lower quality output, and when you are shooting other subjects you can use exposure blending to achieve dynamic range."
-
But doesn't 'blending exposures' defeat the high fps and great action camera selling point for the 7DmkII? That is basically why you are paying $850 more for it in the 1st place.
Are you talking real-world photography or what's theoretically achievable? From what I've seen specs-wise I doubt there'll be much real-world difference in image quality between the two. If you're gonna blend multiple exposures with one you'll probably do it with the other too. (Note that I have no interest in owning or using either camera. I'm merely curious about what Canikon is currently up to.)
-Dave-
-
Are you talking real-world photography or what's theoretically achievable? From what I've seen specs-wise I doubt there'll be much real-world difference in image quality between the two. If you're gonna blend multiple exposures with one you'll probably do it with the other too. (Note that I have no interest in owning or using either camera. I'm merely curious about what Canikon is currently up to.)
-Dave-
First, I'm talking real world as you can get cleaner images from having extra dynamic range. And yes, if you are blending multiple exposures, then you can overcome this handicap.
But let us make sure we understand the data. Below are the 3 Dynamic Range versus ISO curves Hogan cites. His original point was to debunk the notion that both the D300s and 7DmkII scored equivalently as the '70' overall rating indicated without regards to how the camera might be used. But it's a silly comparison between a discontinued 5 year old technology and a just release camera.
The other 2 graphs are much more relevant. The middle compares the 7DmkII to the current Nikon D7100 that is 1.5 years older and $850 cheaper ans shows the D7100 has 1.5 stops more DR at base ISO and never under preforms the 7DmkII at any ISO. The bottom graph, to me is much more ominous. It shows that 5 years of development has yielded only a 1 stop gain in DR from ISO 1000 and up!
-
First, I'm talking real world as you can get cleaner images from having extra dynamic range. And yes, if you are blending multiple exposures, then you can overcome this handicap.
But let us make sure we understand the data. Below are the 3 Dynamic Range versus ISO curves Hogan cites. His original point was to debunk the notion that both the D300s and 7DmkII scored equivalently as the '70' overall rating indicated without regards to how the camera might be used. But it's a silly comparison between a discontinued 5 year old technology and a just release camera.
The other 2 graphs are much more relevant. The middle compares the 7DmkII to the current Nikon D7100 that is 1.5 years older and $850 cheaper ans shows the D7100 has 1.5 stops more DR at base ISO and never under preforms the 7DmkII at any ISO. The bottom graph, to me is much more ominous. It shows that 5 years of development has yielded only a 1 stop gain in DR from ISO 1000 and up!
well you're correct of course. Sensorgen says the same thing. 7100 is a good two stops better in DR than the 7D II at base ISO bu they tighten up above 1600 and are very close at 6400
But a camera System is the sum of many parts. Now I don't know since I did own a 7100 but never a 7D II but the 7100 wasn't very video friendly. The 7D II appears to be a better companion for such matters. Both great crop sensor cameras. If we dissect out the sensors the Nikon appears to have some advantage. But in the real Real World it probably will come down to usage and which lenses are on your shelf.
Time will tell... maybe the 7D II will be "nailed to the shelf" and Canon will have to unveil the Rabbit in their hat earlier than they wanted to...
-
But a camera System is the sum of many parts. Now I don't know since I did own a 7100 but never a 7D II but the 7100 wasn't very video friendly. The 7D II appears to be a better companion for such matters. Both great crop sensor cameras. If we dissect out the sensors the Nikon appears to have some advantage. But in the real Real World it probably will come down to usage and which lenses are on your shelf.
Agreed. No camera ranking system is meant to have you go pick the highest rated that you can afford and call it a day. As an Amatuer and a generalist shooting everything from sports to studio portraits, I just find the 7DmkII very uncompelling against the D7100. 10fps is nothing to sneeze at, but I just don't need it to capture baseball, soccer, golf and other "action" areas I have experience in shooting. I've never had a super fast camera and have had to learn to execute without it. If I won a super duper wildlife photo trip, I would probably beg, borrow or steal a super fast camera because I lack the knowledge and experience in that area to shoot it well with a slow camera.
Time will tell... maybe the 7D II will be "nailed to the shelf" and Canon will have to unveil the Rabbit in their hat earlier than they wanted to...
The million dollar question on everyone's mind is "Does Canon have something in the hat worth pulling and when will they finally pull it?" To date, they haven't shown any indication they 1) have something to compete with the Sony sensors or 2) are willing to go outside to get it. I would have thought the 7DmkII would have been a good time, especially if the Rumored D9300 (24MP, DX, Expeed4b, 8fps (10fps w/ grip), D810 like body) materializes.
-
Yes exactly. One dark frame actually increase noise but removes unwanted signal. A program like deep sky stacker will work with stacks of different frames. I'm certainly no expert but a typical scenario would be you have 20 (or more) light frames of your deep sky object, then during the night you also grab some 20 dark frames with the same duration, then you take (next day perhaps) some flat field images to remove vignetting. And you also put a lens cap on, fastest shutter speed and grab a series of bias frames. 20+ or 256 in your case. These are averaged together and are used to calibrate all of the other types of frames, basically to remove read pattern noise. Then you end up with master dark and flat field images that you can reuse with different light frames. More serious astrophotographers will keep a library of master darks for different sensor temperatures.
So are you averaging dark frames? I don't know theory and math but I guess the result would be the same more or less. For shorter exposures it might be practical. Dark + bias frames might be used only because it is impractical to grab so many long exposure darks.
Raw Therapee lets you use dark frame and flat frame RAW on other RAWs. No bias frames yet.
The translation for others that may not be familiar with astro lingo, is you can store the noise and imperfections of your camera to be subtracted from all your future images. That is impressive in a raw converter.
-
The million dollar question on everyone's mind is "Does Canon have something in the hat worth pulling and when will they finally pull it?" To date, they haven't shown any indication they 1) have something to compete with the Sony sensors or 2) are willing to go outside to get it. I would have thought the 7DmkII would have been a good time, especially if the Rumored D9300 (24MP, DX, Expeed4b, 8fps (10fps w/ grip), D810 like body) materializes.
Thom has recently confirmed that Nikon had moved their Expeed range to ARM based processors. I wonder if that means that they have the possibility to parallelize processing?
That would limit the high speed challenge to sensor readout, shutter speed and battery life only. ;)
Cheers,
Bernard
-
If your metric of dynamic range is the ratio of full-scale to the average signal level that produces a signal-to-noise ratio of around 10 or so, which happens to be mine, without a radical change to the design of the sensor, there's not much progress to be made here. If he's talking about engineering dynamic range, which is read-noise limited, I do think we can expect some improvement in the near future. The Sony Alpha 7S changes the conversion gain as you go from ISO 1600 to 2000, dramatically lowering the read noise. Designs that offer more granular adjustments to conversion gain might do what he wishes for.
Jim
One can use the DXO Full SNR graphs to calculate the DR for any desired nose floor as described by Emil Martinec (http://www.luminous-landscape.com/forum/index.php?topic=42158.0). I downloaded the graphs for the Canon 7D MII and Nikon D7100 and calculated the DR for Jim's 10:1 SNR (20 dB) for base ISO of 100 and for ISO 1600. The DRs for SNR 1:1 (0 dB) as used by DXO are also calculated for verification of the methodology, and the values agree with the reported DXO DR at base ISO of 11.11 stops and 12.92 stops for the Canon and Nikon respectively.
At Jim's 10:1 floor, the Nikon has somewhat better DR at base ISO (8.04 stops vs 7.28 stops), but the DRs at ISO 1600 are very close, 4.35 stops and 4.51 stops respectively for the Canon and Nikon.
Bill
-
One can use the DXO Full SNR graphs to calculate the DR for any desired nose floor as described by Emil Martinec (http://www.luminous-landscape.com/forum/index.php?topic=42158.0). I downloaded the graphs for the Canon 7D MII and Nikon D7100 and calculated the DR for Jim's 10:1 SNR (20 dB) for base ISO of 100 and for ISO 1600. The DRs for SNR 1:1 (0 dB) as used by DXO are also calculated for verification of the methodology, and the values agree with the reported DXO DR at base ISO of 11.11 stops and 12.92 stops for the Canon and Nikon respectively.
At Jim's 10:1 floor, the Nikon has somewhat better DR at base ISO (8.04 stops vs 7.28 stops), but the DRs at ISO 1600 are very close, 4.35 stops and 4.51 stops respectively for the Canon and Nikon.
Bill, I don't look at the DxO numbers myself. Does DxO say how they get from a ratio to dB? 10*log10(ratio) vs 20*log10(ratio). I'd agree with you that electrons linearly convert to voltage and therefore it ought to be the latter, but it would be good to make sure.
Good work on the DR calcs.
Jim
-
My take is that dynamic range is a bit overrated. It depends a bit what you shoot. Erik
Hi Erik,
Dynamic range requirements depend entirely on what you shoot, not just a bit. ;)
The camera tries to mimic what the eye sees. Unfortunately, the camera is limited by a fixed aperture for every single shot, whereas the iris of the eye will automatically adjust the diameter of the eye's pupil (aperture) to accommodate changes in brightness.
Some of these adjustments in pupil size take place almost instantaneously, as for example when we peruse a landscape scene and shift our attention from the bright sky to the moss-covered roots of a fig tree which is in the shade, to the right of the composition we are about to photograph.
If the eye's aperture, after its adjustment to see the detail in the bright sky, could not instantaneously widen when we shift our gaze to the tree roots in the shade, we wouldn't be able to see any detail in the shade, just like a camera with a low dynamic range.
Sometime the adjustments in pupil size take a while, as for example when walking from bright light into a dark room such as a cinema theatre. If one were to combine the 2 situations of almost instantaneous adjustment of pupil size when perusing the average contrasty scene, and the slower adjustment required when moving to a darkened area, the total dynamic range of human vision would be in the order of 30 stops.
However, for general scenes with high contrast, such as the scene in one's living room combined with the view of the sky and landscape through the window, the eye can accommodate about 20 stops of dynamic range without any noticeable delay in pupil adjustment, which is much better than even a modern Nikon DSLR can manage with a single shot. ;)
-
Hi Erik,
Dynamic range requirements depend entirely on what you shoot, not just a bit. ;)
The camera tries to mimic what the eye sees. Unfortunately, the camera is limited by a fixed aperture for every single shot, whereas the iris of the eye will automatically adjust the diameter of the eye's pupil (aperture) to accommodate changes in brightness.
Some of these adjustments in pupil size take place almost instantaneously, as for example when we peruse a landscape scene and shift our attention from the bright sky to the moss-covered roots of a fig tree which is in the shade, to the right of the composition we are about to photograph.
If the eye's aperture, after its adjustment to see the detail in the bright sky, could not instantaneously widen when we shift our gaze to the tree roots in the shade, we wouldn't be able to see any detail in the shade, just like a camera with a low dynamic range.
Sometime the adjustments in pupil size take a while, as for example when walking from bright light into a dark room such as a cinema theatre. If one were to combine the 2 situations of almost instantaneous adjustment of pupil size when perusing the average contrasty scene, and the slower adjustment required when moving to a darkened area, the total dynamic range of human vision would be in the order of 30 stops.
However, for general scenes with high contrast, such as the scene in one's living room combined with the view of the sky and landscape through the window, the eye can accommodate about 20 stops of dynamic range without any noticeable delay in pupil adjustment, which is much better than even a modern Nikon DSLR can manage with a single shot. ;)
You are right that the human eye has an immense dynamic range: the inter-scene range is well over 40 f-stops. However, it only achieves a small portion of that range through pupil size variation. The range of the pupil is from about f/2 to about f/8, or about 4 stops.
Jim
-
You are right that the human eye has an immense dynamic range: the inter-scene range is well over 40 f-stops. However, it only achieves a small portion of that range through pupil size variation. The range of the pupil is from about f/2 to about f/8, or about 4 stops.
Jim
THIS
It's the 15 (or so)Teraflop computer between your ears and the intra-cellular biochem of the eye's system that does the heavy lifting under the broader DR of the real world...
-
Bill, I don't look at the DxO numbers myself. Does DxO say how they get from a ratio to dB? 10*log10(ratio) vs 20*log10(ratio). I'd agree with you that electrons linearly convert to voltage and therefore it ought to be the latter, but it would be good to make sure.
Good work on the DR calcs.
Jim
Jim,
They use 20*log10 as outlined here (http://www.dxomark.com/About/In-depth-measurements/Measurements/Noise).
I'm not an engineer, but understand that 20*log10 is for power and 10*log10 is for voltage. Perhaps you can explain the difference to us.
Bill
-
Hi,
I would say that human vision is complex.
Regarding my statement that DR is overrated is more based on experience
- I shoot mostly landscape
- Very few of my 75000 images have issues with DR
- On most occasions I shot HDR I could make an equivalent image from a single ETTR exposure
- Even the camera most limited in DR I am using, the P45+, is doing just fine in most situations
- When I got my Sony Alpha 99 SLT, giving almost 2EV gain in DR over my Sony Alpha 900, it took me three months to find a case where I could find an advantage in DR, and that was duping a extremely high contrast Velvia slide in a "totally" dark room. A Velvia slide can have a density range of about 13 steps, I think
But there are some other points.
- The DR that DxO reports is at SNR=1, and it is essentially readout noise. I think it is quite relevant for black areas.
- But areas affected by low light will be dominated by photon statistics.
- Read noise on modern sensors is close physical limits, I think. FWC is not growing fast (I would think). So I expect delopment to slow down
I would certainly suggest that Canon is lagging in the DR/shadow noise area, specially at low ISO-s, but low ISO work is probably not what the majority of their buyers does. Most buyers may not even shoot raw, expose ETTR or even use a tripod.
Best regards
Erik
-
Never noticed…
You may take it as a compliment.
Best regards
Erik
Jim,
They use 20*log10 as outlined here (http://www.dxomark.com/About/In-depth-measurements/Measurements/Noise).
I'm not an engineer, but understand that 20*log10 is for power and 10*log10 is for voltage. Perhaps you can explain the difference to us.
Bill
-
They use 20*log10 as outlined here (http://www.dxomark.com/About/In-depth-measurements/Measurements/Noise).
That's good.
I'm not an engineer, but understand that 20*log10 is for power and 10*log10 is for voltage. Perhaps you can explain the difference to us.
It's the other way 'round, actually. The idea is to make dB's related to power and dB's related to voltage match if a resistive load doesn't change. dB's started out as a power measurement; 10 dBs meant 10 times as much power. Then people wanted to measure voltage that way. If you increase the voltage into a resistive load by a factor of 10, you get 10 times the current and 100 times the power. Since the log10 of 100 is 2 and the log10 of 10 is 1, people figured that you could multiply the log10 of a voltage ratio by 20, and get the same power ratio as if you'd multiplied the log of the power ratio by 10.
Try it: let's start with one volt into one ohm. That means one amp, and thus one watt. Now raise the voltage by 20 dB. Since it's voltage, it takes 10 times the voltage to get the voltage up by 20 dB. (20*log10(10) = 20) Now we've got 10 volts, 10 amps, and 100 watts. Go back to the beginning, and raise the power by 20 dB. We have to go from 1 watt to 100 watts to do that. (10*log10(100) = 20)
Jim
-
That's good.
It's the other way 'round, actually. The idea is to make dB's related to power and dB's related to voltage match if a resistive load doesn't change. dB's started out as a power measurement; 10 dBs meant 10 times as much power. Then people wanted to measure voltage that way. If you increase the voltage into a resistive load by a factor of 10, you get 10 times the current and 100 times the power. Since the log10 of 100 is 2 and the log10 of 10 is 1, people figured that you could multiply the log10 of a voltage ratio by 20, and get the same power ratio as if you'd multiplied the log of the power ratio by 10.
Try it: let's start with one volt into one ohm. That means one amp, and thus one watt. Now raise the voltage by 20 dB. Since it's voltage, it takes 10 times the voltage to get the voltage up by 20 dB. (20*log10(10) = 20) Now we've got 10 volts, 10 amps, and 100 watts. Go back to the beginning, and raise the power by 20 dB. We have to go from 1 watt to 100 watts to do that. (10*log10(100) = 20)
Jim
Jim,
Thanks for the correction and physics lesson. Things make more sense now, and i can see the advantages of megavoltage transmission lines, but that takes us away from our photography discussion.
Bill
-
Hi Erik,
Dynamic range requirements depend entirely on what you shoot, not just a bit. ;)
The camera tries to mimic what the eye sees. Unfortunately, the camera is limited by a fixed aperture for every single shot, whereas the iris of the eye will automatically adjust the diameter of the eye's pupil (aperture) to accommodate changes in brightness.
Some of these adjustments in pupil size take place almost instantaneously, as for example when we peruse a landscape scene and shift our attention from the bright sky to the moss-covered roots of a fig tree which is in the shade, to the right of the composition we are about to photograph.
If the eye's aperture, after its adjustment to see the detail in the bright sky, could not instantaneously widen when we shift our gaze to the tree roots in the shade, we wouldn't be able to see any detail in the shade, just like a camera with a low dynamic range.
Sometime the adjustments in pupil size take a while, as for example when walking from bright light into a dark room such as a cinema theatre. If one were to combine the 2 situations of almost instantaneous adjustment of pupil size when perusing the average contrasty scene, and the slower adjustment required when moving to a darkened area, the total dynamic range of human vision would be in the order of 30 stops.
However, for general scenes with high contrast, such as the scene in one's living room combined with the view of the sky and landscape through the window, the eye can accommodate about 20 stops of dynamic range without any noticeable delay in pupil adjustment, which is much better than even a modern Nikon DSLR can manage with a single shot. ;)
I'm with you Ray, IMO the whole avoid harsh mid-day contrast that washes out your images is due to the lack of DR in the artificial imaging system. You go out in the day things look normal. You take a picture of it, it looks terrible. That is the camera to screen system failing to represent what our eyes see.
-
You are right that the human eye has an immense dynamic range: the inter-scene range is well over 40 f-stops. However, it only achieves a small portion of that range through pupil size variation. The range of the pupil is from about f/2 to about f/8, or about 4 stops.
Jim
Yes, my post is misleading. I gave the impression that the eye's adaption to changes in brightness and contrast is due only to changes in pupil diameter. The rods and cones also undergo a change in sensitivity to accommodate more extreme changes in brightness, and that process takes more time.
Nevertheless, if one views a contrasty scene that is clearly outside of the DR capability of a single shot from the best camera, such as an indoor scene which includes a view of a landscape and sky on a sunny day, as seen through the windows of a room, then the eye has no trouble making almost instantaneous adjustments when the gaze is redirected from the brightest cloud in the sky to the darkest corner in the room. There is no sense of impenetrable blackness or noise in those shadows.
Also, I imagine (but I'm no expert here) that in addition to the 'almost instantaneous' change in pupil diameter, in the situation described above, there will occur some additional adaption of the cones, over a period of just a few seconds, that will add to the effect of that 4 stop change from F8 to F2.
-
Regarding my statement that DR is overrated is more based on experience
Of course. I understand, Erik. We all have different experiences and different interests in different types of subjects. If you try taking shots in a Rainforest in sunny Australia, or in the ruins at Angkor Wat in Cambodia on a sunny day, you might be dissatisfied with the quality of detail in the shadows if you don't take multiple shots for merging to HDR, or use fill flash where practical, and/or if you don't have a high-DR camera.
I also rarely bother to use HDR now that I have a Nikon DSLR. My concern about the DR limitations of DSLRs was greatest when I was using Canon equipment. I used the 5D a lot, my first full-frame DSLR. It was my favourite camera at the time, although shadow noise could be an eyesore.
It's interesting to now compare the DR values of the Canon 5D and the Nikon D800E, on the DXOMark site. At base ISO the D800E has whopping 3.2 stops better DR, at equal print size. Even at the pixel level, the DR of the D800E is 2 & 1/2 stops better.
I recall when I first moved from the U.K to Australia I was struck by the increased brightness of the light. It was very noticeable. In England the sun rarely shines. Who needs a high-DR camera in the U.K. or Northern Europe! ;) Any Canon DSLR should be perfectly adequate. ;)
-
Retinal ganglion cells/Optic radiations and indeed the Occipital Cortical cells also join in the business of exposure. It' a System we can't even totally map. But one day it seems likely that a 30-40 stop DR may be achievable in a camera. Funny that some think the tehnolpg is near-plateau.
Just need more breakthroughs...
Yes, my post is misleading. I gave the impression that the eye's adaption to changes in brightness and contrast is due only to changes in pupil diameter. The rods and cones also undergo a change in sensitivity to accommodate more extreme changes in brightness, and that process takes more time.
Nevertheless, if one views a contrasty scene that is clearly outside of the DR capability of a single shot from the best camera, such as an indoor scene which includes a view of a landscape and sky on a sunny day, as seen through the windows of a room, then the eye has no trouble making almost instantaneous adjustments when the gaze is redirected from the brightest cloud in the sky to the darkest corner in the room. There is no sense of impenetrable blackness or noise in those shadows.
Also, I imagine (but I'm no expert here) that in addition to the 'almost instantaneous' change in pupil diameter, in the situation described above, there will occur some additional adaption of the cones, over a period of just a few seconds, that will add to the effect of that 4 stop change from F8 to F2.
-
I'm with you Ray, IMO the whole avoid harsh mid-day contrast that washes out your images is due to the lack of DR in the artificial imaging system. You go out in the day things look normal. You take a picture of it, it looks terrible. That is the camera to screen system failing to represent what our eyes see.
I don't agree with this. Harsh mid day sun is flat and has no depth. It's the same as using a flash on camera as opposed to off camera. 30 stops of DR isn't going to make an on camera flash look good.
-
Hi,
I don't think that the photons are there… Todays sensors are close natural limits.
To get much further I would say that ND filters are needed over half of the pixels or variable exposure time for groups of pixels.
Best regards
Erik
Retinal ganglion cells/Optic radiations and indeed the Occipital Cortical cells also join in the business of exposure. It' a System we can't even totally map. But one day it seems likely that a 30-40 stop DR may be achievable in a camera. Funny that some think the tehnolpg is near-plateau.
Just need more breakthroughs...
-
I don't agree with this. Harsh mid day sun is flat and has no depth. It's the same as using a flash on camera as opposed to off camera. 30 stops of DR isn't going to make an on camera flash look good.
Depends where you are and what time of the year it is.
At the right time of the year, mid-day sun is perfectly acceptable in polar and near-polar regions. At other times, so is midnight sun...
-
Hi,
I don't think that the photons are there… Todays sensors are close natural limits.
To get much further I would say that ND filters are needed over half of the pixels or variable exposure time for groups of pixels.
Best regards
Erik
The photons are there. You just need to expose for long enough so that even the darkest areas receive enough photons to show detail above the noise floor. Brighter areas, obviously, will receive more photons.
The challenge, then, is to have enough well capacity that the highlights don't blow out (i.e. The wells don't fill up) before the shadows receive enough light to show detail. In other words, the challenge is to engineer for brighter highlights, not darker shadows.
-
Hi,
Theere are two problems with your way of reasoning:
1) There is no noise floor, really. Readout noise is so low that shadow noise is dominated by photon flux variations.
2) The exposure is limited by FWC (Full Well Capacity). If full well capacity is exceeded clipping will result.
It is possible to make pixels larger, and that would increase the possible dynamic range of each pixel, but we would have less pixels so the effect on highlights would be nil, but large pixels give marginally lower shadow noise.
Adding ND filters on half of the pixels or using two different exposures on different pixel groups could expand dynamic range greatly. That would be HDR within a single exposure. Fuji has done it in different variations.
I guess that single exposure HDR is not what the market is asking for.
One real problem with HDR exposures is that they need to be downmapped (tone mapped) to something that can be seen in print or screen, doing that in a harmonious way is no easy feat. Think grungy HDRs. Obviously, it can be done in a nice way, but still it's a complex issue with pitfalls.
Best regards
Erik
The photons are there. You just need to expose for long enough so that even the darkest areas receive enough photons to show detail above the noise floor. Brighter areas, obviously, will receive more photons.
The challenge, then, is to have enough well capacity that the highlights don't blow out (i.e. The wells don't fill up) before the shadows receive enough light to show detail. In other words, the challenge is to engineer for brighter highlights, not darker shadows.
-
Hi,
Theere are two problems with your way of reasoning:
1) There is no noise floor. Readout noise is so low that shadow noise is dominated by photo flux variations.
That's exactly what I'm saying. You need to expose for long enough that there is detail in the shadows above the random noise generated by photon flux. I never said anything about read noise.
2) The exposure is limited by FWC (Full Well Capacity). If full well capacity is exceeded clipping will result.
Again, that's exactly what I'm saying, and is the main challenge in increasing DR.
You need to ensure that the wells don't reach capacity before sufficient shadow detail is recorded.
Therefore, the problem is in the highlights rather than the shadows. You can take care of shadow detail by exposing for long enough. But the wells need to have enough capacity that the highlights don't blow out in the time it takes to collect the shadow detail. And every 1-stop increase in dynamic range requires a doubling of the well capacity.
Perhaps the solution isn't in an expose-then-readout system as is used now, but in a continuous readout system, whereby the sensor is continuously read while the exposure is happening. Exposure could continue until sufficient detail is recorded in the shadows above the photon noise; the continuous readout would mean that the numbers from the highlights would simply keep adding up, rather than reaching the limit and being capped there. But that would require different sensor and readout architecture.
It is possible to make pixels larger, and that would increase the possible dynamic range of each pixel, but we would have less pixels so the effect on highlights would be nil, but large pixels give marginally lower shadow noise.
Adding ND filters on half of the pixels or using two different exposures on different pixel groups could expand dynamic range greatly. That would be HDR within a single exposure. Fuji has done it in different variations.
I guess that single exposure HDR is not what the market is asking for.
Or multiple exposures (either with continuous shooting or by shooting a video file) with the frames averaged. Photon shot noise would be averaged out, revealing shadow details that were previously hidden in the noise. And the highlights wouldn't be blown out, since each single exposure would be short enough for the wells not to reach capacity.
One real problem with HDR exposures is that they need to be downmapped (tone mapped) to something that can be seen in print or screen, doing that in a harmonious way is no easy feat. Think grungy HDRs. Obviously, it can be done in a nice way, but still it's a complex issue with pitfalls.
Every image needs to be tonemapped, HDR or not. Without anything to tell you what recorded luminance equals what brightness and colour in the output, it's just a meaningless list of numbers in a file. It's just that single-exposure images already have pre-made tone-mapping curves via RAW conversion software, while curves for HDR images need to be made manually - and, in the hands of the inexperienced, can be made very badly.
-
it's worth a read
http://www.dslrbodies.com/newsviews/do-you-believe-in-dxomark.html
At the end he makes some really interesting comments, ones I happen to agree with
"I guess to end this discussion, I should probably suggest what we Nikon users want to see on the DxOMark tests for the next Nikon. We’ve got it pretty good at the moment, after all. Incremental movement of the dynamic range chart upward would always be welcome, but I think we’re now very close to the point where we want to see something entirely different. We really want to see base dynamic range unbound (e.g. the two-shot type DR that the Apple iPhone 6 is doing, and which Sony’s latest patents all hint at). Beyond that, we want to see flattening of the dynamic range slope as ISO is boosted (less of a hit per doubling of ISO).
Simply put, we’re deep ending into small, incremental gains that aren’t pushing us very much further above the “good enough” bar. What we need is to push that bar far forward again, and that’s going to take another sensor generation/design, I think. Until then, you can all anguish over small differences in DxOMark measurements, but I’d be surprised that those show up anywhere near as clearly in your images. "
Frankly I don't see anything new on his comments. Some people are lazy, and most reporters are. I have been advocating to use the measurements tab here and on getdpi for both lenses and sensors for a long time.
He can say anything he wants about "Nikon is client" and "perhaps" Canon is not, but the true is that DXO numbers that show that the D4 vs the D3s for low light were a complicated story, or that the D750 super low light capabilities (against the D610) were software jpg stuff. The numbers also show that even the 6d has better dynamic range than the Nikons above iso 1600.
People should read and if someone takes a bad decision because they were too lazy to do a few clicks on a dynamic interface before making a massive investment they get what they deserve. They can also ask on the forums and someone will tell them how to use DxO data.
Overall I am glad that T.H post this, maybe now some people will stop the laziness (if this become something all the cool guys do).
In the other hand, it is important to underscore the importance of DxO testing. Before them we were listening to camera brand affiliated experts saying that the dynamic range of CCD base backs was 4 stops (8 vs 12, some times 8 vs 14) better than the best DSLR, even after the release of the D800.
T.H conclusions are very interesting too, the "what we should ask from Sony /Nikon is very valuable.
In a final note, I aside from the conspiracy, I the article is very good, and I know that as a columnist sometimes he has to stress the evident to help the community do better.
Best regards
-
I don't agree with this. Harsh mid day sun is flat and has no depth. It's the same as using a flash on camera as opposed to off camera. 30 stops of DR isn't going to make an on camera flash look good.
Lets consider it some more. With high sun you get the direct full black body spectrum plus off angle refracted light from the surrounding sky, which is mostly blue. So you are right that many textures get filled in with this blue light. The sun near the horizon has very direct yellow angle rays.
To my eye, strong overhead sun gives high contrast that the eye can manage. Overcast is washed out, mild looking. Sun near the horizon lets you work with the textures to make a certain look.
To the camera, it might interpolate the blue from overhead sun as fill, never thought about it before. What I see in many images is shadows go to black or the sky is blown. Overcast the camera loves. You crank the contrast in post bringing out lots of detail. Sun at the horizon again lets you define a look based on the direction of light.