Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: Abdulrahman Aljabri on January 26, 2010, 11:33:19 am
-
I see Phase and Hasselblad advertising higher dynamic range, up to 12.5 stops. For those who have both a digital 35mm and a digital MF cameras are you seeing a difference in captured dynamic range and is it significant? I am not sure what is the dynamic range on 35mm cameras but assuming it is 8-10, 12.5 stops should be significant improvement. Interior photography, particularly in which the light source is visible in the frame such as lamps, can benefit from this greatly. I often find myself in such situations taking several underexposed exposures to add back lost details in the original. This approach is far from ideal because blending is never straight forward with jpeg.
-
I see Phase and Hasselblad advertising higher dynamic range, up to 12.5 stops. For those who have both a digital 35mm and a digital MF cameras are you seeing a difference in captured dynamic range and is it significant? I am not sure what is the dynamic range on 35mm cameras but assuming it is 8-10, 12.5 stops should be significant improvement. Interior photography, particularly in which the light source is visible in the frame such as lamps, can benefit from this greatly. I often find myself in such situations taking several underexposed exposures to add back lost details in the original. This approach is far from ideal because blending is never straight forward with jpeg.
Without wanting to state the obvious- if dynamic range is a concern, why are you shooting JPEGs? JPEG compression is perceptual, so it is going to particularly clobber detail in the shadows and highlights because the algorithm knows that shadows are dark, highlights are bright, and human eyes are more sensitive to details in the mid-tones.
I'd have thought shooting RAW and subsequently processing in 16 bits per channel would do a lot of good to your details- the "recovery" slider in raw processors and Lightroom can help make the most of this and bring out the detail in the shots if it is there to start with. Multiple stacked exposures in JPEG sounds like a very labourious way to work- if you do need multiple combined exposures and some sort of HDR procedure, you'll definitely do better with RAW than with JPEG.
Cheers, Hywel Phillips
-
Without wanting to state the obvious
you just did, lol
Multiple stacked exposures in JPEG sounds like a very labourious way to work-
it is, and raw becomes better here.
if you do need multiple combined exposures and some sort of HDR procedure, you'll definitely do better with RAW than with JPEG
I am aware of this but my question is about MF vs 35mm, whether it is jpeg vs jpeg or raw vs raw, are MF cameras producing more dynamic range?
-
Take said cameras, point them at at backlit calibrated dynamic range chart - Stouffer wedges are commonly used for this, and count the number of wedges you can see for each camera. Post your answer and images on this forum for us all to enjoy.
Graeme
-
Take said cameras, point them at at backlit calibrated dynamic range chart - Stouffer wedges are commonly used for this, and count the number of wedges you can see for each camera. Post your answer and images on this forum for us all to enjoy.
Graeme
I wish I could but I do not have access to MF cameras in my area. The only way to get my hands on one is to buy it from the US and have it shipped over.
-
I wish I could but I do not have access to MF cameras in my area. The only way to get my hands on one is to buy it from the US and have it shipped over.
DXO has taken detailed measurements of many cameras, both DSLR and MFDB, and posted the results online. A few minutes of quality time with Google should get you all the info you need.
-
I have direct and long term experience with Leaf 75S, Leaf 22, P25+, Nikon D3, Nikon D300, Fuji S3 & S5; KING of DR is the Fuji S5, jpg vs jpg and raw vs raw! The DBs are very close in DR to the Fuji, nearly the same, but still fall very slightly short of it. In my experience neither the D3 nor the D300 are in the same league when it comes to DR, dxo figures or not.
-
you just did, lol
it is, and raw becomes better here.
I am aware of this but my question is about MF vs 35mm, whether it is jpeg vs jpeg or raw vs raw, are MF cameras producing more dynamic range?
If it is JPEG vs JPEG, there will be very little difference, because the limiting factor is the JPEG format and compression, not the cameras. JPEG is limited to 8 bits per channel anyway so only 8 stops (actually that's not strictly true because of the non-linear tone curve baked into the JPEG files, but still the principle is true because recovering those details from a lossy perceptual compression is tricky or impossible).
For lab measurements, as others have said DxO's website contains a mountain of careful measurements. In practice, shooting RAW on my Canon 5D vs my Hasselblad H3D-31, the Hasselblad does "feel" like it has more dynamic range (and much better image quality for a variety of other reasons).
But... as I suggested in my first post, if dynamic range is a primary concern for you, the first step is to turn the camera off JPEG always shoot RAW, whatever camera you are using. I'd certainly do that before considering the relative merits of an expensive MFDB system :-)
Cheers, Hywel.
-
Just a thought, but maybe we ought to have a FAQ on this forum for all people who want to know the differences between DSLR's and MF systems? Every couple months or so, someone wanders in and initiates the same discussion over and over.
To the OP - you should do a search - there's tons of discussion already here about this. Waste of time to repeat it -ESPECIALLY if you 'don't have access' and most likely won't ever use one.
-
I'm no tech-head and don't know much about the numbers. I'm just a photographer who's interested in doing great work. MFDB may have more dynamic range than small format, but it's hard to tell the difference in the finished shot. If you're hoping for MFDB to make a big difference that's going to make the skies part and the air sing with enlightenment, you're likely to be disappointed.
Dump the jpegs and shoot raw. There's a big difference there and it won't cost you anything. If you still want more and have cash to burn, then go with MFDB.
John
-
DXO has taken detailed measurements of many cameras, both DSLR and MFDB, and posted the results online. A few minutes of quality time with Google should get you all the info you need.
Some photographers are not swayed by scientific data, but take pictures of clouds and infer an expanded DR for their expensive MFDBs. I suspect that they see what they need to see to justify their cash expenditure. .
According to the DXO measurements, the Phase One P65+ has 11.51 EV of dynamic range at base ISO and the Nikon D3x has 12.8 EV of DR at base ISO. The Phase One does have better resolution and enables bigger prints, but I have yet to see photographic evidence that it has better DR than the D3x. Of course, I have neither camera and would like someone with both cameras to demonstrate the practical DRs of each instrument. However, I won't hold my breath for those results expect that subjective hype will continue to be offered.
The DXO measurements do have a wealth of data. Older dSLRs such as the Nikon D3 have their DR limited by the ADC and have relatively high read noise at base ISO. If read noise were constant across ISOs, each doubling of the ISO would halve the DR, since only half as many photoelectrons would be collected. If you look at the DR plot vs ISO for the D3 (which has only 11.92 EV of DR at base ISO), it doesn't change much for the first 2 doublings over base ISO. One collects fewer electrons, but the read noise decreases--for each stop of DR lost at the high end due to fewer collected photons, one gains nearly 1 stop of DR at the low end due to reduced read noise. As one reaches ISO 800, she does lose 1 stop of DR for each doubling of ISO. With this camera it makes little sense to set the ISO camera over 800 is you are shooting raw--it merely reduces highlight headroom. You do get a brighter preview on the LCD.
The DR vs ISO curve for the D3x is nearly linear, indicating low read noise at base ISO. The plot for the P65+ is linear, indicating that read noise does not change with ISO, but is relatively high due do the higher read noise with CCD designs as compared to CMOS. Other things being equal, one would expect that MFDBs would have better DR, since they collect many more photo electrons due to the large sensor size.
-
Some photographers are not swayed by scientific data, but take pictures of clouds and infer an expanded DR for their expensive MFDBs. I suspect that they see what they need to see to justify their cash expenditure. .
According to the DXO measurements, the Phase One P65+ has 11.51 EV of dynamic range at base ISO and the Nikon D3x has 12.8 EV of DR at base ISO. The Phase One does have better resolution and enables bigger prints, but I have yet to see photographic evidence that it has better DR than the D3x. Of course, I have neither camera and would like someone with both cameras to demonstrate the practical DRs of each instrument. However, I won't hold my breath for those results expect that subjective hype will continue to be offered.
The DXO measurements do have a wealth of data. Older dSLRs such as the Nikon D3 have their DR limited by the ADC and have relatively high read noise at base ISO. If read noise were constant across ISOs, each doubling of the ISO would halve the DR, since only half as many photoelectrons would be collected. If you look at the DR plot vs ISO for the D3 (which has only 11.92 EV of DR at base ISO), it doesn't change much for the first 2 doublings over base ISO. One collects fewer electrons, but the read noise decreases--for each stop of DR lost at the high end due to fewer collected photons, one gains nearly 1 stop of DR at the low end due to reduced read noise. As one reaches ISO 800, she does lose 1 stop of DR for each doubling of ISO. With this camera it makes little sense to set the ISO camera over 800 is you are shooting raw--it merely reduces highlight headroom. You do get a brighter preview on the LCD.
The DR vs ISO curve for the D3x is nearly linear, indicating low read noise at base ISO. The plot for the P65+ is linear, indicating that read noise does not change with ISO, but is relatively high due do the higher read noise with CCD designs as compared to CMOS. Other things being equal, one would expect that MFDBs would have better DR, since they collect many more photo electrons due to the large sensor size.
Here we go again! Sigh! So many arm chair psuedo scientific posts don't make it right. And its been pointed out numerous times why the DXO comparisons are not accurate for comparing MFDB to DSLR.
-
Folks, thanks for your feedback, but I am not looking for software analysis. I am looking for your practical findings. I personally can't think of too many places where I would need extra dynamic range in my professional photography work. That is why I posted a practical scenario, namely interior photography with light source in the frame. I was just carious if others found MF better in such setting for example.
Furthermore, I am not interested in buying a MF camera. Unless my work requires it I see no reason of making the investment. So far I have been asked about MF by only one advertising agency. That being said I was just interested to know if people saw any practical difference in a real shooting scenario.
-
Folks, thanks for your feedback, but I am not looking for software analysis. I am looking for your practical findings. I personally can't think of too many places where I would need extra dynamic range in my professional photography work. That is why I posted a practical scenario, namely interior photography with light source in the frame. I was just carious if others found MF better in such setting for example.
Furthermore, I am not interested in buying a MF camera. Unless my work requires it I see no reason of making the investment. So far I have been asked about MF by only one advertising agency. That being said I was just interested to know if people saw any practical difference in a real shooting scenario.
There are a few areas where I feel MF outperforms DSLRs in any situation, and since you so wisely asked for *practical* experience over bloated scientific analysis, I'll share a few of my thoughts based on my own experience.
First off, for what you do -- interiors with controlled lighting -- I would probably be hard pressed to recommend anything more than a high-end Canon body coupled to the 17 and 24 TSE II lenses. That combo has certainly proven itself in a very short time, and as a huge plus, it's very convenient and easy to use. It's also relatively inexpensive compared to a MF back, tech camera and digital lenses! Bottom line is if what you're currently using works for you and your customers are not complaining, then you are probably set.
That out of the way, the areas where I find MF will outperform the high-end Canon include: higher resolution, expanded DR, more accurate color, better tonality, better inter-pixel contrast. All of these factors allow you to push things around more aggressively in post to get to a desired result, including additional perspective corrections and selective pixel editing. So IMO the real question you need to ask yourself is if you need that ability, how often will you need it, and is it going to be worth the added investment (significant) for the MF solution?
As respects your uses, I think if you used one of the latest 60MP MF backs on a good tech camera with the best lenses, and processed both files optimally* and printed them out to 24x32, the MF print would show clear advantages; at 40x50 print sizes, the 60MP advantage will be patently clear and no contest. Of course you'll only be able to see the advantages if you compare the prints side-by-side -- and for at least in the 24x32, most folks would be entirely satisfied with the Canon print if they didn't see the MF print. Moreover, you aren't likely to see the color or tonal advantages in an online jpeg.
*The other issue is that not all MF shooters take the time to learn the back's software to process their files optimally, so we often see "comparable" results generated in online comparisons where the final files were not fully optimized -- usually these come from a DSLR shooter that borrowed or rented an MF back for a day. Sometimes we even see similar comparisons come from folks who have only read other analyses and then extrapolate their own hard scientific conclusion without ever picking up the cameras in question!
Hope this helps,
-
I have direct and long term experience with Leaf 75S, Leaf 22, P25+, Nikon D3, Nikon D300, Fuji S3 & S5; KING of DR is the Fuji S5, jpg vs jpg and raw vs raw! The DBs are very close in DR to the Fuji, nearly the same, but still fall very slightly short of it. In my experience neither the D3 nor the D300 are in the same league when it comes to DR, dxo figures or not.
I second that, though my experience is with Fuji S3 which has identical DR with Fuji S5.
-
Here we go again! Sigh! So many arm chair psuedo scientific posts don't make it right. And its been pointed out numerous times why the DXO comparisons are not accurate for comparing MFDB to DSLR.
I don't know what is pseudo in my post. Kindly elaborate on what is false.
Actually, DXO uses standard image analysis methods for dynamic range measurement. Their Noise Measurements (http://www.dxomark.com/index.php/eng/Technologies/Testing-protocols/Noise-protocol) are very scientific and are more precise than we could get at home by using a Stouffer wedge. Those who maintain that the DXO comparisons are not accurate for comparing MFDBs to dSLRs likely do not understand the normalization process. Acutally, DXO explains (http://Contrary%20to%20conventional%20wisdom,%20higher%20resolution%20actually%20compensates%20for%20noise) that contrary to conventional wisdom, higher resolution actually compensates for noise.
If you can supply a better evaluation for the DR of MFDBs, please supply a link or present data.
-
-
I don't know what is pseudo in my post. Kindly elaborate on what is false.
Actually, DXO uses standard image analysis methods for dynamic range measurement. Their Noise Measurements (http://www.dxomark.com/index.php/eng/Technologies/Testing-protocols/Noise-protocol) are very scientific and are more precise than we could get at home by using a Stouffer wedge. Those who maintain that the DXO comparisons are not accurate for comparing MFDBs to dSLRs likely do not understand the normalization process. Acutally, DXO explains (http://Contrary%20to%20conventional%20wisdom,%20higher%20resolution%20actually%20compensates%20for%20noise) that contrary to conventional wisdom, higher resolution actually compensates for noise.
If you can supply a better evaluation for the DR of MFDBs, please supply a link or present data.
bjanes, you can talk about DxO and scientific analysis all day, but in the end, if the MF print looks better than the DSLR print, then the conclusion is easy. The question is how much someone is willing to pay for how little a difference. One needs to look at the print and the wallet to decide, not the DxO website.
-
Hi,
The question is about MFDBs having more DR than 35 mm sensors. I'd suggest that the DxO tests give pretty good data on the issue. It may be that DxO and the person posting the question have a different interpretation of DR.
I would also say that the issue is a bit complex. Print cannot have DR larger than around 7. So the only way to put a DR of 10-12 into a print is using curves, that is, processing is involved. The question weather an MF print looks better than a DSLR print is probably not very much related to DR as DR used to be plenty. There are a lot of other factors.
The DxO data are probably accurate for some of the parameters using their technical definitions. The technical definitions may say little about how prints from two different system are perceived. If the MFDB image is better, is it better because DR is higher, or something else?
My feeling is that DR is like buzzword. Lot of talk but is it really DR we are talking about? Also, DR is about shadow detail, by definition. So if anyones sees differences in other areas than deep shadow it is not DR. The Fuji sensor may be an exception to this as it has actually an additional highlight sensing sensel and combines the two sensel in firmware, but MFDBs doesn't use similar technology.
Best regards
Erik
bjanes, you can talk about DxO and scientific analysis all day, but in the end, if the MF print looks better than the DSLR print, then the conclusion is easy. The question is how much someone is willing to pay for how little a difference. One needs to look at the print and the wallet to decide, not the DxO website.
-
The DxO data are probably accurate for some of the parameters using their technical definitions. The technical definitions may say little about how prints from two different system are perceived. If the MFDB image is better, is it better because DR is higher, or something else?
My feeling is that DR is like buzzword. Lot of talk but is it really DR we are talking about? Also, DR is about shadow detail, by definition. So if anyones sees differences in other areas than deep shadow it is not DR. The Fuji sensor may be an exception to this as it has actually an additional highlight sensing sensel and combines the two sensel in firmware, but MFDBs doesn't use similar technology.
Erik,
I think you hit the nail on the head with your reply. Michael and others whose opinions I trust do report an advantage for MFDB, so one has to come up with the reasons for that advantage. Having three or more times the resolution with no low pass filter is a big advantage. Also very high quality mostly prime lenses help. Top of the line electronics and a leisurely frame rate with 16 bit ADCs (analog to digital converters) are other advantages. It is difficult to make an ADC with a high frame rate and also a high dynamic range. The Hubble telescope reads out its images at a pretty low bit rate. DXO does not test for resolving power or some of these other factors.
For very large prints, the extra resolving power alone would be a big advantage, but some MDFB report a difference with 8 x 10 inch prints.
Bill
-
I rather suspect that most people, when they are talking about DR, are really referring to the film or sensor's ability to hold in highlight detail rather than anything else. Certainly that has always been my concern when shooting for B/W prints using digital sensors, or scanning B/W film for that matter. In colour, perhaps we are rather more concerned with shadow detail, because dead black shadows don't look very good in a colour print, whereas in B/W that's probably OK (at least with me). So control of highlights is very much the issue in B/W. Whether DR is 8 stops, 10 stops or 12 stops is rather a moot point - what really counts is the ability to make an exposure which will just give detail in the brightest parts of the scene that can be retrieved in the eventual print (specular highlights excepted). That's the tricky bit. A larger DR will help, of course, but no camera or film or sensor yet made can see things as the human eye does. So shooting tight in towards the sun on a bright day will always involve some compromise - do I keep the clouds and lose the shadows under those trees, or lose the clouds and open up the shadow areas? Since Fox Talbot, it has ever been thus.
John
-
Of course, I have neither camera and would like someone with both cameras to demonstrate the practical DRs of each instrument.
Exactly.
That was the only line worth writing on your post.
-
Exactly.
That was the only line worth writing on your post.
Perfect!
-
One thing to keep in mind in this discussion is that many photographers still think that DR is measured by the amount of blown highlights they manage to recover with their favorite raw converter...
Cheers,
Bernard
-
According to the DXO measurements, the Phase One P65+ has 11.51 EV of dynamic range at base ISO and the Nikon D3x has 12.8 EV of DR at base ISO. The Phase One does have better resolution and enables bigger prints, but I have yet to see photographic evidence that it has better DR than the D3x.
Great post, Bill. Spot on, as always. In addition to the resolution and other MFDB advantages you mentioned, another thing it has is improved color depth in the upper part of the dynamic range. The increase in tonal gradations is easy to measure, but I think the visual effect would be subtle.
-
You can look up the DR for most 35mm DSLRs, some compact, and selected digital backs, along with other quantitative info at: dxomark.com.
David Saffir
-
Since people are debating whether DR would show up in shadows or highlights let me provide an example
(http://farm5.static.flickr.com/4057/4315954868_202c775532_o.jpg)
My understanding is that extra DR would show up on both ends and would help where the red spots are placed. For each spot there is a light source shinning directly in the frame. The original shot was a jpeg file with some details blended back in with a -2 underexposed file. Obviously my mistake is using jpeg instead of RAW, but even with RAW this problem will still be there. Don't MF cameras show more details in the original file much like a 35mm file would after digital blending?
And since we are at it, how do people counter this problem? It is clear from the picture that I failed to restore details to the desirable level. In fact I resorted to a trick by painting detail with a color brush just to make those areas appear to hold detail.
Feel free to critique the picture if you like. Thanks in advance!
-
Since people are debating whether DR would show up in shadows or highlights
Really? I just thought people were yet again discussing how many angels can dance on the head of a pin.
That's what it sound like to me ...
-
DR is typically noise limited. Indeed, you can think of sensors only having more and more range in the shadows as the top end is fixed by it's clip point. However, that does not mean you should only look for extra DR in the shadows as it all depends where you place mid grey in the range the sensor offers, and hence how many stops above and stops below you can record.
Given that different cameras sensor sensitivities are different, you can't just put a similar lens and set the same shutter and f-stop on each camera and expect a fair comparison. You might see one didn't clip some highlights, but the other did, and conclude the first had more DR. However, the first could be less sensitive. The second camera might have much more shadow detail and hence could have been easily allowed to be under-exposed by a stop, recover the highlights and get a wider DR by bettering the first in the shadow department.
That is why real world scenes are not so good for measuring DR, and why calibrated backlit wedge tests are good, because it doesn't matter where on the scale you clip, as long as you clip near enough the top that you can count down into the noise floor at the bottom.
Graeme
-
And since we are at it, how do people counter this problem?
http://www.luminous-landscape.com/tutorial...-blending.shtml (http://www.luminous-landscape.com/tutorials/digital-blending.shtml)
http://luminous-landscape.com/forum/index....showtopic=40761 (http://luminous-landscape.com/forum/index.php?showtopic=40761)
http://www.luminous-landscape.com/tutorial...exposures.shtml (http://www.luminous-landscape.com/tutorials/blended_exposures.shtml)
http://www.luminous-landscape.com/forum/lo...php/t24466.html (http://www.luminous-landscape.com/forum/lofiversion/index.php/t24466.html)
http://www.luminous-landscape.com/forum/lo...php/t35171.html (http://www.luminous-landscape.com/forum/lofiversion/index.php/t35171.html)
-
http://www.luminous-landscape.com/tutorial...-blending.shtml (http://www.luminous-landscape.com/tutorials/digital-blending.shtml)
http://luminous-landscape.com/forum/index....showtopic=40761 (http://luminous-landscape.com/forum/index.php?showtopic=40761)
http://www.luminous-landscape.com/tutorial...exposures.shtml (http://www.luminous-landscape.com/tutorials/blended_exposures.shtml)
http://www.luminous-landscape.com/forum/lo...php/t24466.html (http://www.luminous-landscape.com/forum/lofiversion/index.php/t24466.html)
http://www.luminous-landscape.com/forum/lo...php/t35171.html (http://www.luminous-landscape.com/forum/lofiversion/index.php/t35171.html)
Thanks man, I am familiar with a couple of those, in fact the first link was the first tutorial I read about the subject when I was still new to photography several years ago. I have used those techniques with landscape images with very excellent results. For some reason, however, they don't seem to provide the same results with interior shots. Taking the highlight area in underexposed shots produces very dark tones which cannot blend with original picture tones. Also the transition in underexposed shots becomes very severe. Drop the exposure more and you get overly overexposed edges. Very weird, perhaps its the very high DR lights in frame require versus typical landscape scenes. I guess its like trying to paint back details in the sun.
-
To Graeme's point, revisit my post http://luminous-landscape.com/forum/index....8&hl=sensor (http://luminous-landscape.com/forum/index.php?showtopic=39318&hl=sensor)
In a test where the camera position, lens and lighting remained constant, my P65+ showed greater detail in hilights and shadows than my D3. It's not, however, enough detail to make up for the extreme contrast of many interiors scenes. You frequently have blown out light fixtures in the shot which may be the only area that isn't rendered with detail.
Sometimes I let these go, and sometimes I find them too distracting and will bracket for later compositing. Even if I have people in the view, I find that I can process a single P65+ file +1.5 and -1.5 with little to no loss in quality and then composite those images to achieve far greater tonal range than any recovery sliders can render.
In the past I've used the above linked technique, and I also like Photomatix, but I don't find that either yields a result that I can call final... I often blend the offensive areas (windows or fixtures) into a straight exposure that has nice contrast overall.
Will there ever be a sensor that has "enough" dynamic range? I truly have no idea. But I'm practically giddy to have twice the DR I used to have with chromes and don't mind a little compositing... for now.
Parting thoughts: RAW RAW RAW! Jpegs have less tonal range than an 8 bit Tif. Raws have more detail than a 16 bit Tif. Which do you think you should be shooting? Actually it often astounds me how much detail is there in the hilights of raw files if you play around with exposure and recovery. You just have to massage it out and into your Tifs.
To address your image specifically, I feel like the sconce and the ceiling niches are fine but the wall feature lights are very harsh and definitely need to be softened. It feels quite yellow overall and the shadow areas seem heavy. In the ceiling I would remove the speaker, the access panel, the can light on the top edge and the dead can light to start cleaning things up. Lastly, your shot is a little out of square. You needed to swing left a little more : )
-CB
-
Thanks man, I am going through the link as we speak
Parting thoughts: RAW RAW RAW! Jpegs have less tonal range than an 8 bit Tif. Raws have more detail than a 16 bit Tif. Which do you think you should be shooting? Actually it often astounds me how much detail is there in the hilights of raw files if you play around with exposure and recovery. You just have to massage it out and into your Tifs.
And, your shot is a little out of square. You needed to swing left a little more : )
-CB
yeah I got so used to using jpeg because they are mostly sufficient that I miss opportunities to shoot RAW when there is a real benefit to using it such as in this picture. Back when I used to live in Chicago I used to photograph the forest preserves in Palos Heights and I always used RAW for landscape. I would open the file in Photoshop RAW make zero adjustments and open the file as a 16bit jpeg. From there I would set the white and black point as I please and play with contrast.
-
One thing to keep in mind in this discussion is that many photographers still think that DR is measured by the amount of blown highlights they manage to recover with their favorite raw converter...
Bernard
Bernard,
Quite true, but a camera that has a lot of room for highlight recovery is not desirable. Highlight recovery is possible because the red and blue channels of a typical sensor are less sensitive than the green channel. This results in a white balance multiplier for the red and blue channels. For the Nikon D3 with Solux illumination, the red and blue multipliers are 1.55 and 1.61 respectively, as shown in the Rawnalize screen capture shown below. The green channel here is just short of clipping.
[attachment=19853:16_Rawnalize.png]
If we render this image in ACR at defaults, the image appears overexposed.
[attachment=19854:16_ACR_Default.png]
Using negative exposure, we appear to be recovering 1.05 stops of overexposure. However, there was actually no overexposure but merely a hot tone curve in ACR (which has a baseline exposure of +0.5 EV and a default brightness of +50).
[attachment=19855:16_ACR_MinusExp.png]
With more actual exposure of this image, the green channel would clip first but the blue and red channels would be intact and allow highlight recovery. However, there would likely be color shifts. The blue multiplier of 1.6 corresponds to approximately 0.69 stops, and this would be the maximal highlight recovery for this camera. It would be desirable if the blue and red channels had the same sensitivity as the green channel. Some photographers place a magenta filter over the lens to hold back some of the green light and balance the channels, allowing increased dynamic range.
Since the response of a CMOS or CCD sensor is linear, it makes no sense to talk about highlight and shadow dynamic range on either side of mid gray. If you expose for mid gray (18% saturation), the highlights will be 2.5 stops to the right. However, for digital exposure it does not make sense to meter from the midtones, but rather one should place highlights that must not be clipped just short of clipping. Shadow noise and effective DR will then be determined largely by the read noise of the sensor for a given full well. I'm sure you know all this, but others may be interested.
Finally a few words on DR. Dynamic range is defined as the full well capacity/read noise (both expressed in electrons). This is the engineering definition, and photographers may set the noise floor higher. For a recent MFDB sensor (the Kodak KAF 50100), the full well is 40,300 e- and the read noise is 12.5 e-, giving a DR of 11.7 stops, about the same as the Canon 1D MII (Roger Clark (http://www.clarkvision.com/imagedetail/evaluation-1d2/index.html)). The current champion of DR for dSLRs is the Nikon D3x due to its low read noise and decent full well capacity. I do not have the actual figures for that camera, but DXO reports the DR at 12.84 stops (screen), and they report 11.07 stops DR for the 1DMII (similar to Clark's data). This is per pixel DR, reported as "screen" on DXO. The screen DR for the Phase 1 p65+ is 11.51 stops. If one normalizes for a given print size, the DRs for the D3x, P65+ and 1D MII are 13.65, 12.97 and 11.11 stops respectively. I prefer these scientific data to subjective impressions DR derived by looking at pictures of clouds.
-
DR is typically noise limited. Indeed, you can think of sensors only having more and more range in the shadows as the top end is fixed by it's clip point. However, that does not mean you should only look for extra DR in the shadows as it all depends where you place mid grey in the range the sensor offers, and hence how many stops above and stops below you can record.
Given that different cameras sensor sensitivities are different, you can't just put a similar lens and set the same shutter and f-stop on each camera and expect a fair comparison. You might see one didn't clip some highlights, but the other did, and conclude the first had more DR. However, the first could be less sensitive. The second camera might have much more shadow detail and hence could have been easily allowed to be under-exposed by a stop, recover the highlights and get a wider DR by bettering the first in the shadow department.
You are refering to the fact that DR depends on sensor true sensitivity, and that sensor sensitivity being different with different bodies, it is difficult to compare DR.
True enough, but it doesn't change the fact that there is one and only one method to measure DR with digital sensor, and that method has absolutely nothing to do with mid-tones.
That correct method is to expose to the right (brighest significant highlight not blown in any channel) and measure shadow noise.
That method isn't just mathematical, it is closely related to the way digital cameras need to be exposed for optimal results.
Regards,
Bernard
-
Man, this all makes my head hurt. I just wanna make pictures.
-
Man, this all makes my head hurt. I just wanna make pictures.
+1
-
I hope it will be figured out or abandoned once and forever and the topic will be pinned to the top.
I'll go make some photographs in the meantime as well.
All those generic tests and mathematical calculations make no sense to me. One can only decide if IQ, DR, body design etc. suits him or her by examining his/her own real work produced with the equipment in question.
Best of luck to you anyway.
Alex
-
Practically though, it's hard to adjust a camera to be "just" under clipping, so we allow them to visibly clip in the chart and then we can measure from that point downwards. We do the same thing, just a bit easier to "prove" that we have captured the full range possible.
Graeme
You are refering to the fact that DR depends on sensor true sensitivity, and that sensor sensitivity being different with different bodies, it is difficult to compare DR.
True enough, but it doesn't change the fact that there is one and only one method to measure DR with digital sensor, and that method has absolutely nothing to do with mid-tones.
That correct method is to expose to the right (brighest significant highlight not blown in any channel) and measure shadow noise.
That method isn't just mathematical, it is closely related to the way digital cameras need to be exposed for optimal results.
Regards,
Bernard
-
Given the number of workshops that take place each year, between Michael's groups here and Jack and Guy's groups over on getdpi, I would think this matter could easily be put to rest if the same landscape were shot under identical lighting conditions with a 22 megapixel digital back and, say, a Nikon D3x, using the best glass of equivalent focal lengths on each camera. After all... it is often claimed that even the 22 mp digital backs will clearly beat the best of the 35mm offerings. Shoot the damned images at = iso, process in Capture One (or whatever the best software may be for the that given camera) and print them out at 24x36" on fine art paper. Dot gain will level the field somewhat... and the resulting images should clearly show whatever advantage one might have over the other. This subject has photographers so polarized that nothing short of a visual test on paper will ever settle the issue. Make the prints and label them on the back. Don't disclose the camera until the opinions are in. What better way than to let your eye's decide between two images shot under identical conditions?
-
... I would think this matter could easily be put to rest if the same landscape were shot under identical lighting conditions with a 22 megapixel digital back and, say, a Nikon D3x, using the best glass of equivalent focal lengths on each camera. After all... it is often claimed that even the 22 mp digital backs will clearly beat the best of the 35mm offerings. Shoot the damned images at = iso, process in Capture One (or whatever the best software may be for the that given camera) and print them out at 24x36" on fine art paper.
The main problem with your proposal is that the best glossy prints have a dynamic range of 8 stops at best, and the DR of fine art paper is considerably less. To show the entire DR of the capture would require tone mapping during the rendering, and good rendering is art as well as science (see Karl Lang (http://wwwimages.adobe.com/www.adobe.com/products/photoshop/family/prophotographer/pdfs/pscs3_renderprint.pdf): Rendering the Print: the Art of Photography). This brings up the raw developer and its settings as well as the skill of the photographer.
Shooting at the same ISO would not be advisable. At high ISO the D3x would blow away a P65+. I would suggest base ISO, since if DR is critical, one should optimize it, using a tripod if necessary. For landscapes most would use a tripod anyway.
You would likely want to normalize for print size as DXO does, but 24 by 36 inches is too much for a D3x, since the print resolution would only be 168 pixels/inch. Also, 3:2 is not the aspect ratio for most MFDBs, and you would have to crop the image.The difference in resolution with a P65+ would be obvious. 22 MP is not state of the art for MFDB. You would be comparing old vs new technology if you compared a 22 MP back with a D3x.
Then you would have to make certain that the photographed scenes are identical and that the exposure was optimum with the sensor at saturation so that the full DR of the camera is used. You can't rely on camera histograms or histograms on the raw converter.
As Michael showed with his megapixel shootout, controlling all the variables and reaching subjective conclusions was difficult even for those highly skilled and knowledgeable photographers. Finally, as Michael pointed out, regardless of the test procedures, there will be nit pickers.
Nonetheless, the test you proposed would be interesting. Other things being equal, a larger sensor would likely come out on top.
Bill
-
"The main problem with your proposal is that the best glossy prints have a dynamic range of 8 stops at best, and the DR of fine art paper is considerably less."
That's clear to me as well, Bill. It punishes both 35mm and mf digital equally, so there is no advantage to either. I print exclusively on fine art paper. Others prefer glossy or premium luster. Two prints could be made of each and this would make for an even more interesting comparison. I simply prefer to judge the quality of both files on a print rather than on screen.
"To show the entire DR of the capture would require tone mapping during the rendering, and good rendering is art as well as science (see Karl Lang (http://wwwimages.adobe.com/www.adobe.com/products/photoshop/family/prophotographer/pdfs/pscs3_renderprint.pdf): Rendering the Print: the Art of Photography). This brings up the raw developer and its settings as well as the skill of the photographer."
Personally I'd trust both Michael and Jack with this task... not that others on this and other forums wouldn't be competent as well. The point is not to make this such a difficult exercise that you hamstring the whole thing with a protracted exercise in nit picking. If MF digital is clearly superior it shouldn't require the efforts of Joseph Holmes to make this obvious. If the difference is barely visible on fine art paper, those who print with fine art paper can save themselves a ton of money and headaches by sticking with 35mm... if that's where they're coming from.
"Shooting at the same ISO would not be advisable. At high ISO the D3x would blow away a P65+. I would suggest base ISO, since if DR is critical, one should optimize it, using a tripod if necessary. For landscapes most would use a tripod anyway."
I never suggested a test where the D3x and mf digital are shot at 3200 iso. The point here is to shoot both at their optimum iso as that's the iso one would use when making a large fine art print.
"You would likely want to normalize for print size as DXO does, but 24 by 36 inches is too much for a D3x, since the print resolution would only be 168 pixels/inch. Also, 3:2 is not the aspect ratio for most MFDBs, and you would have to crop the image.The difference in resolution with a P65+ would be obvious. 22 MP is not state of the art for MFDB. You would be comparing old vs new technology if you compared a 22 MP back with a D3x."
Optimize the print dimensions however you like. One is 3:2 and the other is 4:3. Get on with it. No one has ever alleged that the best in 35mm digital is equal to a 65 mp back. Certainly not me. The numbers who have stated that a 22 mp digital back will beat the best 35mm has to offer are legion. This isn't about proving a P65+ is better than a D3X at optimum iso. I don't think anyone disputes that so why cook the books in favor of the P 65+? A 22 mp back is not old technology as many are still doing their commercial work with an Aptus with 22 mp. It was still being sold when the D3X was released. The point is about being able to see superior IQ at optimum iso and deciding whether the constraints of MF digital are worth the effort in switching platforms. If a 22 mp digital back will prove superior... a p30+, P39+, P40+ and the Dalsa equivalents will only make that more obvious. I'm sure many have this question.
"Then you would have to make certain that the photographed scenes are identical and that the exposure was optimum with the sensor at saturation so that the full DR of the camera is used. You can't rely on camera histograms or histograms on the raw converter."
Gee... lets just make this so impossible it's not worth the effort to do any test! Your make this sound like an exercise in pissing up a rope. Some of us would really like to see such a test so we can lay this dog to rest. You make it sound like you'd rather contribute to thread after thread on the same topic and throw obstacles in the path of anyone who is trying to get answers. Your prerogative and you're entitled to it. I have better things to do with my time.
I'm looking for definitive answers here.
"As Michael showed with his megapixel shootout, controlling all the variables and reaching subjective conclusions was difficult even for those highly skilled and knowledgeable photographers. Finally, as Michael pointed out, regardless of the test procedures, there will be nit pickers."
Especially when some of those "nit pickers" base their arguments on misquoting the actual resolution of the scans after repeated corrections by Michael!
"Nonetheless, the test you proposed would be interesting. Other things being equal, a larger sensor would likely come out on top."
Bill
I think you're right on that last point, Bill. My interest is in seeing just how noticeable the difference is on fine art paper. Others will prefer glossy. I picked an arbitrary size of 24x36 as it would tell us a lot about the quality of those mf digital pixels on a 22 mp digital back... just as it would about the D3X's output. I don't think there are many who have shot with a 22 mp mf digital back who would think they're at a disadvantage in such a test. Side by side prints are where the rubber meets the road.
Lawrence
-
I think you're right on that last point, Bill. My interest is in seeing just how noticeable the difference is on fine art paper. Others will prefer glossy. I picked an arbitrary size of 24x36 as it would tell us a lot about the quality of those mf digital pixels on a 22 mp digital back... just as it would about the D3X's output. I don't think there are many who have shot with a 22 mp mf digital back who would think they're at a disadvantage in such a test. Side by side prints are where the rubber meets the road.
Lawrence
If you're interested in an accurate comparison of DR, then a print is going to hinder you rather than help you. A print has such much less dynamic range than today's good cameras that even with a very carefully constructed development curve you're going to have to get into very careful secondary colour correction - carefully brightening dark areas (by hand or by whatever technique) - to map the large DR of the camera down onto the smaller DR of the screen or a print.
Graeme
-
The captured DR maps to the print, so if you saved highlights and shadows in a given capture, they have a chance to map to the print and while compressed, you can usually maintain visible tonal separation across the entire capture value even though it's compressed -- at least you can if you know what you're doing in post. In the case where the captured image fully clipped shadows and/or highlights and they cannot be processed out, compression at the print not only leaves them clipped, but in most cases the clipping edge is more obvious due to the compression, leaving visible edges where the blocked shadows and blown highlights transition in the print. (And if you've seen it, you know it's an ugly result.)
Capturing versus rendering. While the captured DR at the sensor may be measurably larger for camera A than it is for camera B, camera B's combination of file properties and processing software -- and the user's ability to better manipulate it -- can allow for a superior final rendering of a total DR for the purposes of that print. Thus when you compare the final prints, camera B's DR result appears superior to camera A's.
For *my* purposes, which is a print, I'll choose to use camera B because it renders superior DR, and I am able to get more of it visually into the print, usually represented by smoother transitions in the deep and high values. Granted, my absolute color values may not be as accurate in the deep shadows or bright highlights as the camera A result, but B's transition values can still look smoother/better for the aesthetic I'm after...
-
If you're interested in an accurate comparison of DR, then a print is going to hinder you rather than help you. A print has such much less dynamic range than today's good cameras that even with a very carefully constructed development curve you're going to have to get into very careful secondary colour correction - carefully brightening dark areas (by hand or by whatever technique) - to map the large DR of the camera down onto the smaller DR of the screen or a print.
Graeme
This is probably true to some extent when just looking at an image's light to dark component, but if you look at the image from a sideways path across the hues instead of luminance, you'll see one area of difference between camera platforms that goes largely without discussion. The richness of color or tonality that a camera is able to capture is really important to an image and IMHO contributes a significant palpable depth. So while most papers have a contrast of 100::1 or 200:1 and the DR (in luminance terms) is compressed, in some cases the subtle color transitions in things like flowers, plants, sky, and skin can still get through intact from camera to print. It seems that my CCD based cameras - Ixpress 528c, phase p20, and leica DMR all do a much better job at this than my canon 5d2.
-
Oh look, it's an MFDB vs. 35mm thread, how exciting !
They've become so rare, ....