Pages: 1 2 3 [4] 5   Go Down

Author Topic: Ten Years after….  (Read 16987 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Ten Years after….
« Reply #60 on: February 18, 2014, 01:38:06 pm »

Hi,

Some observations, contradicting yours a little bit. Obviously I have a limited experience shooting only Sony and P45+.

Washed out highlights come from overexposure. If you expose correctly the highlights will be OK. Digital backs underexpose 1-1.5 stops, protecting highlights.

If you use "filme curve" in C1 it will push exposure 1-1.5 stops. Try checking with linear curve to see what the raw image really looks like.

Regarding colour accuracy, I measured colour accuracy on my Sony Alpha 99 and my P45+ and the Sony was much more accurate than the P45, and I have made similar measurements with similar results from published tests on Hasselblad compared to Nikon D800. The colours may be better but more accurate they are not.






Best regards
Erik




This isn't exactly what you are asking for, but I did shoot the same image (outdoor architecture, all artificial lighting, night shot) with a Canon 5DII and an Aptus 7 (33 MP) on a tech cam with a 35 mm lens. The Aptus image was clearly superior, especially in highlight detail. Where the Canon had clipped channels and made the wood go yellow, the Aptus kept both accurate color and impressive detail. It was good enough that if you needed a bigger image, you could just uprez it in photoshop as all the useable detail in the scene was captured. I was impressed.

The client didn't see any appreciable difference and went with the Canon shot because it was a little wider and they liked the 2x3 perspective. I vastly preferred shooting with the Canon because the camera was mounted on a very tall tripod and where the Canon could be completely controlled with Capture One, the tech cam required making adjustments to the lens/shutter blind by reaching high overhead. The tripod was on a hill in a snow bank.

Now, I have not done a similar shoot with a D800 or A7/r, and that could be interesting. Especially an A7/r with a Canon T/S lens.

I got to play in a studio with a model and a D800 and IQ180. All the picture from the D800 were technically good, but I had a hard time getting a great picture from it. It just seemed a bit flat. It took all my C1 mojo to make them look good. The IQ180/DF (might have been DF+) was a different story. Getting an acceptably focused image was very difficult because of the DF and full size chip, but the ones that were in focus were spectacular and presented a very different challenge in C1 -- the images could be made to look any way that you wanted. While the DSLR had IMO one correct way to process them, the IQ180 files were so flexible that your own creativity was the limiting factor.

What's the point of all this? Well, if you are looking at legacy backs, I think a bigger sensor is a better sensor as long as you don't need the advantages of a smaller sensor (higher ISO, capture speed). The more recent digital backs (P30+, P40+, Aptus 7, Aptus 8) are more color accurate than any DSLR. There are limitations of the DF body, especially with focusing. This is more of a problem with a large chip because of DOF. The digital backs (any of them) tether better and more reliably than a DSLR.

So, from an overall technical view, I would say there are advantages, but at a cost. To me, the real reason to invest in a 22 mp back is because you want to upgrade to a newer technology back at some point, and want to start building a camera system. I would be surprised if there is any advantage to a 22mp back over a current DSLR. When you move up to a 33 - 39 mp back, I think that changes.
Logged
Erik Kaffehr
 

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Ten Years after….
« Reply #61 on: February 18, 2014, 01:52:34 pm »

Digital backs underexpose 1-1.5 stops, protecting highlights.

Best regards
Erik


Actually I find the opposite… If one trusts his light meter, MF backs (usually) overexpose. But again no image area "exposes", the photographer does.
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
less analog gain before ADC is NOT less exposure
« Reply #62 on: February 18, 2014, 03:12:30 pm »

Digital backs underexpose 1-1.5 stops, protecting highlights.
Erik,
    Can we stop confusing

exposure, meaning how much light the sensor receives (determined by illumination, aperture ratio and shutter speed)

with

how much analog amplification is applied to the sensor's signal before ADC?

The difference is clearly important, as less exposure reduces photon counts and so leads to a worse ratio of signal to photon shot noise, whereas less amplification causes no change in the effect of photon shot noise, but only in the noise introduced after that amplification, which with either a CCD or CMOS sensor with column-parallel ADC is just quantization noise from the ADC. (Canon sensors are about the only remaining exceptions to this.)

What many MF backs are doing is applying less amplification than some other cameras, less analog gain before ADC, and this leaves more headroom for highlight recovery in the raw file.  With a gap of two or the stops between the capabilities of ADCs delivering 16-bit and sensors delivering only 12-13 stops of DR, this make perfect sense: even base ISO amplification level probably puts the noise levels in the signal entering the ADC comfortably above the noise floor of the ADC, so that further analog gain does nothing to improve overall noise levels; it just increases the number of the least significant bits that are pure noise, and increase the risk of losing highlights that are within the sensor's range but get amplified into clipping.

In particular, if I am reading DXO's documents correctly, its measurement of what it mis-describes as "true ISO" are done with equalized exposure level at equal camera ISO setting. DXO measures all sensors with no lens (and so no aperture setting) and the same combination of exposure time and subject illumination for each given ISO setting. So there is no variation in in exposure, just a variation in the raw level placement for the output of that exposure.

By the way, it should be obvious that when manual settings of aperture and shutter speed are used, the level of exposure is dictated by those choices, and the camera cannot mess with the exposure level. Nor is there any evidence that MF backs are fudging the exposure in auto-exposure modes, by choosing a lower shutter speed and/or a lower aperture ratio than over cameras do at equal ISO speed setting.
« Last Edit: February 18, 2014, 03:14:17 pm by BJL »
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: less analog gain before ADC is NOT less exposure
« Reply #63 on: February 18, 2014, 08:23:28 pm »

In particular, if I am reading DXO's documents correctly, its measurement of what it mis-describes as "true ISO" are done with equalized exposure level at equal camera ISO setting. DXO measures all sensors with no lens (and so no aperture setting) and the same combination of exposure time and subject illumination for each given ISO setting. So there is no variation in in exposure, just a variation in the raw level placement for the output of that exposure.

By the way, it should be obvious that when manual settings of aperture and shutter speed are used, the level of exposure is dictated by those choices, and the camera cannot mess with the exposure level. Nor is there any evidence that MF backs are fudging the exposure in auto-exposure modes, by choosing a lower shutter speed and/or a lower aperture ratio than over cameras do at equal ISO speed setting.

My understanding was that for a given shutter speed/aperture/ISO the level of raw exposure reached by a back at ISO100 corresponds to the level of raw exposure reached by DSLRs at ISO30.

The raw processing software compensate for this when opening the raw file, which does not give the appearance of underexposure, but actually protects highlights from being blown. This resulting in the impression that backs have more "highlight" DR.

Am I mistaken?

Cheers,
Bernard

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: less analog gain before ADC is NOT less exposure
« Reply #64 on: February 18, 2014, 08:24:42 pm »

Hi,

I agree on exposure is something what we choose, but our exposure is either based on exposure index or on histograms.Normally I only use histograms, but it is also a question how exact they are.

See enclosed images:

Note:

1) IQ 180 has an ISO rating well above measured ISO
2) This doesn't seem to apply to P45+
3) The standard film curve is much brighter than the linear film curve, tricking you into exposing less (this is on P45+)

I also enclose the raw histogram from the same image.

As a side question, with the P45+ and the IQ 180 having exposure ratings so much apart, who is confusing, me or Phase One?

Best regards
Erik

Erik,
    Can we stop confusing

exposure, meaning how much light the sensor receives (determined by illumination, aperture ratio and shutter speed)

with

how much analog amplification is applied to the sensor's signal before ADC?

The difference is clearly important, as less exposure reduces photon counts and so leads to a worse ratio of signal to photon shot noise, whereas less amplification causes no change in the effect of photon shot noise, but only in the noise introduced after that amplification, which with either a CCD or CMOS sensor with column-parallel ADC is just quantization noise from the ADC. (Canon sensors are about the only remaining exceptions to this.)

What many MF backs are doing is applying less amplification than some other cameras, less analog gain before ADC, and this leaves more headroom for highlight recovery in the raw file.  With a gap of two or the stops between the capabilities of ADCs delivering 16-bit and sensors delivering only 12-13 stops of DR, this make perfect sense: even base ISO amplification level probably puts the noise levels in the signal entering the ADC comfortably above the noise floor of the ADC, so that further analog gain does nothing to improve overall noise levels; it just increases the number of the least significant bits that are pure noise, and increase the risk of losing highlights that are within the sensor's range but get amplified into clipping.

In particular, if I am reading DXO's documents correctly, its measurement of what it mis-describes as "true ISO" are done with equalized exposure level at equal camera ISO setting. DXO measures all sensors with no lens (and so no aperture setting) and the same combination of exposure time and subject illumination for each given ISO setting. So there is no variation in in exposure, just a variation in the raw level placement for the output of that exposure.

By the way, it should be obvious that when manual settings of aperture and shutter speed are used, the level of exposure is dictated by those choices, and the camera cannot mess with the exposure level. Nor is there any evidence that MF backs are fudging the exposure in auto-exposure modes, by choosing a lower shutter speed and/or a lower aperture ratio than over cameras do at equal ISO speed setting.
Logged
Erik Kaffehr
 

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: less analog gain before ADC is NOT less exposure
« Reply #65 on: February 18, 2014, 09:09:59 pm »

My understanding was that for a given shutter speed/aperture/ISO the level of raw exposure reached by a back at ISO100 corresponds to the level of raw exposure reached by DSLRs at ISO30.
Bernard (and Erik),

Firstly some of that is the lower base ISO sensitivity of many digital,backs, due to tue absense of microlenses and such. Beyond that, the short answer seems to be that you (and DXO) are talking about measurements of numerical levels in raw files (relative to maximum level), which are basically the product of
- how much _exposure_ the sensor gets
by
- the conversion factor in "levels per electron",
The latter in turn depends on the degree of _amplification_ applied before ADC, measured by the gain in "volts per electron". The differences you are talking about are differences in the choice of _amplification_, not a different level of _exposure_.

Also, please note the very important difference between
- losing highlights due to overfilling electron wells (too much exposure), which I call "blown highlights"
and
- losing higlights from photosites that did not get overfilled, but whose signal was then amplified beyond the maximum voltage that the ADC can handle (over-amplification), which I call "clipped highlights".

Amplifying a bit less, but still enough that the noise introduced by the ADC is insignificant compared to the noise in the analog signal coming from the sensor, is an unequivocally good thing for image qualify (less risk of highlight clipping with no significant decrease in SNR), so it puzzles me why so many people wish to characterize as a defect or deception or falsificTionof the ISO speed with inaccurate talk about "underexposure".  And let us be frank: the prefix "under-” is clearly pejorative, implying an error, which is not the case.

I agree with your characterization that this gives the appearance of more "highlight DR", and your inference that this is rather mythical entity, but if the practical consequence is that standard light metering is less likely to lead to highlight problems and with no downside, I do not underatand your cynical attitude to this approach. Can you point to any disadvantage that I am overlooking?
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
ISO exposure index vs highlight-based base-ISO speed as measured by DXO
« Reply #66 on: February 18, 2014, 09:29:25 pm »

Erik,
I am surprised that you have not read my previous explanations of this, or have forgotten them.
The name "ISO" is ambiguous with multiple related meanings defined in ISO standard 12232, including:
- the exposure index used by a camera's internal light metering to adjust (or in manual mode) how much exposure is given to the sensor.
-  the _minimum_ exposure index consistent with placing mid-tones at least three stops below maximum level (related to the "base ISO speed".)
The former is what the "ISO" setting on a camera refers to; in fact this is mandated for Japanese cameras by their industry association.
The latter is roughly what DXO measures, and clearly there is no reason why this ISO-standard recommended _minimum_ should the the same as the former. In fact the better the ADC, the more it makes sense to increase the highlight headroom, by _amplifying_ by less than the amount that would give that minimum recommended space between metered mid-tone and maximum level, which DXO confusingly describes as a lower "ISO".

What your graphs show is two variations to this approach of amplifying a bit less.

The IQ180 with a Dalsa sensor keeps the amplification the same over the first three ISO setting levels, and then (with abundant highlight protection in place) amplifying in proportion to ISO setting beyond that. One suspicion is that the subsequent increase in "DXO measured level placement" is done by bit shifting, and there is no actual variation in analog gain, as indeed there does not beed to be of the ADC is good enough.

In the P45+ with a Kodak sensor, the first step up in ISO exposure index goes with doubled gain, presumably getting to a level where ADC noise is irrelevant, and after that the analog gain is fixed, with further level adjustments on the outout left to be done in raw conversion.

The difference is probably related to a fundamental difference between Dalsa and Kodak CCDs: the Dalsa ones do charge-to-voltage conversion on the sensor chip, at the corners, and output a voltage, whereas Kodak sensors do not, and output the actual charges from each photosite.

And of course, these choices are noted in the raw file and taken account of in conversion to JPEG, probably with simple bit-shifting (totally noise-free amplification!). The fact that increased analog amplification can introduce noise and nonlinearity(and require more complicated designs with more different levels of analog gain required) whereas digital amplification is perfectly linear and noise free, is another argument in favor of this approach of less analog gain, more digital gain.
« Last Edit: February 18, 2014, 09:54:54 pm by BJL »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: less analog gain before ADC is NOT less exposure
« Reply #67 on: February 18, 2014, 09:46:40 pm »

As a side question, with the P45+ and the IQ 180 having exposure ratings so much apart, who is confusing, me or Phase One?
Why do you insist on calling these difference in exposure when in fact they are essentially just differences in how the information from the sensor is encoded in the intermediate stage of raw file levels, not evidence of any changes in how much exposure the sensor was given?

The differences are of course noted in the raw file and taken account of by all common raw conversion software, so I do not see why you should get confused: these are mainly just different ways of encoding the same information into the numerical raw levels. People who look at photographs rather than raw histograms should not be affected or confused in any way by these variations; the only people who need to care are the authors of raw conversion software.
Logged

synn

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1235
    • My fine art portfolio
Re: less analog gain before ADC is NOT less exposure
« Reply #68 on: February 18, 2014, 10:01:19 pm »

People who look at photographs rather than raw histograms should not be affected or confused in any way by these variations; the only people who need to care are the authors of raw conversion software.

Can't +1 this enough.
Logged
my portfolio: www.sandeepmurali.com

Kumar

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 754
    • http://www.bskumarphotography.com
Re: Ten Years after….
« Reply #69 on: February 18, 2014, 10:25:59 pm »

I think your explanation referred to single-shot vs. multi-shot backs. With scanning backs, each pixel receives the full RGB exposure . So, what exactly is the difference between them and multi-shot backs?

Kumar
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: less analog gain before ADC is NOT less exposure
« Reply #70 on: February 19, 2014, 12:56:50 am »

Hi,

You don't blow out highlights if you don't clip.

But, noise levels depend on how well the sensor is utilized. If you consider that an exposure E is just below saturation of the sensor you would have a certain SNR (Signal Noise Ratio). Reducing exposure one step would increase SNR by 41 percent.

So reducing exposure 1 stop from optimal reduces the SNR by 41 percent and reduces DR by 1EV (because you sensels are not fully utilized). It could also be said that a sensor like mine P45+ has twice the size of the sensor of my Alpha 99, so it can collect about twice the number of photons. So if I reduce exposure one stop on the P45+ it will collect about the same number of photons as the Alpha 99. So most of the advantage of the larger sensor would be gone.

The larger sensor of course can conserve more of lens MTF (due to lower sampling frequency for equal print size) and that advantage remains, and that is frankly the only advantage I have found with the P45+.

The images below were shot same evening. The clouds obviously moved a lot, so they are not the same. Alpha on the left P45+ on the right.

Best regards
Erik


Bernard (and Erik),

Firstly some of that is the lower base ISO sensitivity of many digital,backs, due to tue absense of microlenses and such. Beyond that, the short answer seems to be that you (and DXO) are talking about measurements of numerical levels in raw files (relative to maximum level), which are basically the product of
- how much _exposure_ the sensor gets
by
- the conversion factor in "levels per electron",
The latter in turn depends on the degree of _amplification_ applied before ADC, measured by the gain in "volts per electron". The differences you are talking about are differences in the choice of _amplification_, not a different level of _exposure_.

Also, please note the very important difference between
- losing highlights due to overfilling electron wells (too much exposure), which I call "blown highlights"
and
- losing higlights from photosites that did not get overfilled, but whose signal was then amplified beyond the maximum voltage that the ADC can handle (over-amplification), which I call "clipped highlights".

Amplifying a bit less, but still enough that the noise introduced by the ADC is insignificant compared to the noise in the analog signal coming from the sensor, is an unequivocally good thing for image qualify (less risk of highlight clipping with no significant decrease in SNR), so it puzzles me why so many people wish to characterize as a defect or deception or falsificTionof the ISO speed with inaccurate talk about "underexposure".  And let us be frank: the prefix "under-” is clearly pejorative, implying an error, which is not the case.

I agree with your characterization that this gives the appearance of more "highlight DR", and your inference that this is rather mythical entity, but if the practical consequence is that standard light metering is less likely to lead to highlight problems and with no downside, I do not underatand your cynical attitude to this approach. Can you point to any disadvantage that I am overlooking?

« Last Edit: February 19, 2014, 12:59:30 am by ErikKaffehr »
Logged
Erik Kaffehr
 

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: less analog gain before ADC is NOT less exposure
« Reply #71 on: February 19, 2014, 02:48:04 am »

Logged

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Ten Years after….
« Reply #72 on: February 19, 2014, 04:05:03 am »

I think your explanation referred to single-shot vs. multi-shot backs. With scanning backs, each pixel receives the full RGB exposure . So, what exactly is the difference between them and multi-shot backs?

Kumar
Scanning backs can only be used with LF cameras of 4x5 image area or more, while MS backs can be used with both MF cameras and LF cameras, the time required to capture the image is significantly less with MS backs which is extremely beneficial as far as constant lighting is concerned and success of the process. MF lenses are (usually) superior than LF lenses (especially if 4x5 image area capable lenses are concerned), testing the scene for accuracy (by using a single shot tethered pre-shot) is important to pre visualise the outcome (vital), with MS one can "stitch" a part of the image (like a curtain for example in an interior scene) using a single shot for the part of the scene he may have "movement detected" without altering any of the parameters, with MS one can stitch easily two or more images due to the time difference and thus shoot a much larger subject (e.g. a huge wall painting), or use an LF camera and do the stitching on the whole image area that the scan back would use, a big advantage for MS is that the MF cameras are more resistant to vibrations. Usually (if one masters the process and uses adequate equipment), a 16x MS shot is as easy as a longish (about a minute) exposure. Of course one can use the same MF back for single shot too.
 Of course if both are correctly done, the result is superb with either… only that scanning is more difficult to get it right with and to control the obstructive parameters.
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: less analog gain before ADC is NOT less exposure
« Reply #73 on: February 19, 2014, 08:33:36 am »

Amplifying a bit less, but still enough that the noise introduced by the ADC is insignificant compared to the noise in the analog signal coming from the sensor, is an unequivocally good thing for image qualify (less risk of highlight clipping with no significant decrease in SNR), so it puzzles me why so many people wish to characterize as a defect or deception or falsificTionof the ISO speed with inaccurate talk about "underexposure".  And let us be frank: the prefix "under-” is clearly pejorative, implying an error, which is not the case.

I agree with your characterization that this gives the appearance of more "highlight DR", and your inference that this is rather mythical entity, but if the practical consequence is that standard light metering is less likely to lead to highlight problems and with no downside, I do not underatand your cynical attitude to this approach. Can you point to any disadvantage that I am overlooking?

As far as I am concerned I think that what Phaseone does makes a lot of sense because most photographers are a lot more uncomfortable with blown highlights than they are with slightly noisier shadows.

As a result, I don't see "under exposure" at capture or through amplification as a negative thing, but I still think that it is reasonnable to call the approach under exposure relative to a supposed ideal ETTR exposure.

Cheers,
Bernard
« Last Edit: February 19, 2014, 09:01:31 am by BernardLanguillier »
Logged

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Ten Years after….
« Reply #74 on: February 19, 2014, 09:50:15 am »

OK… Let's go back to the (now old) 22Mp backs then… Their DR is superb because one can "dig" deep into the shadows with less noise, their mid tones linear part is more contrasty right out of the box, they are easier with older lenses, they present no pixelation if blown up to 200%… which means that the image can be up-sampled and can be printed really large, they have enough detail to compete with a 36mp DSLR, if one has an MS version he can shoot the best still image by far, Their 100 Iso is best exposed with EV value as if it was (more or less)  200 Iso of DSLRs, they keep the WA character of WA-MF lenses,  and they are great (if not the optimum) to use with LF cameras as well as to bring life to older LF equipment… The Drawbacks now…, is mainly some (rare-but they are more prone to it than the next resolution step up) moire, which can be treated to much extend with modern software and (for pixel peepers only since these things are very sharp) a bit less resolution…
Did I miss something….? Oh yes!  :othe price that someone can get one and have a perfectly capable back to experience MF and LF!  ;)
« Last Edit: February 19, 2014, 10:15:42 am by T.Dascalos »
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
one more time: these are not differences in _sensor_exposure_ at all
« Reply #75 on: February 19, 2014, 11:10:54 am »

As far as I am concerned I think that what Phaseone does makes a lot of sense because most photographers are a lot more uncomfortable with blown highlights than they are with slightly noisier shadows.

As a result, I don't see "under exposure" at capture or through amplification as a negative thing, but I still think that it is reasonnable to call the approach under exposure relative to a supposed ideal ETTR exposure.

Cheers,
Bernard
I agree that there is noting wrong with the approach but let me try to get this across one more time:
there is no difference in the exposure received by the sensor, so describing it as under-exposure is flatly wrong, and misleading.
Thus there is no deviation from "supposed ideal ETTR exposure."
The differences are just in the subsequent positioning of the numeric raw levels, so are essentially just different ways of using discrete numerical levels to encode the signal from the sensor.

Just for fun, let me add to the complication by noting that the effect of the difference between 16-bit and 14-bit ADCs is far more important to the anachronistic(*) part of the ETTR ideal about maximizing numerical raw levels.
That difference in bit-depth means that MF backs's lower raw level placements relative to maximum level (which is what DXO measures) still often correspond to the same or higher actual numerical levels. For example:
- if a camera with 14-bit raw output places mid-tones three stops below maximum level (the bare minimum tolerated by the ISO 122323 standard, misinterpreted by some as the one and only correct placement) the raw level is 2^14/2^3 = 2^11, about 2000.
- if a camera with 16-bit raw output places mid-tones four stops below maximum level (halving its "DXO" rating) the raw level is 2^16/2^4 = 2^12, about 4000.
Thus the latter has twice the density of raw levels: twice as many raw levels corresponding to any given range of sensor signal levels, and so is "twice as good" according to the part of the original ETTR idea about maximizing the fineness of raw levels.

So which camera is falling further short of that ETTR ideal?!


(*) Anachronistic because the part of the original ETTR idea about striving to have as many raw levels as possible was relevant when 10-bit and 12-bit ADC themselves imposed a potential limit on the accuracy at which sensor output is recorded, but is largely irrelevant when a modern 14-bit or 16-bit ADC has a wider "dynamic range" (counted for example in bit depth) than the sensor signal (counted for example in f-stops from the brightest photosite's signal down to the noise floor.)
« Last Edit: February 19, 2014, 11:14:43 am by BJL »
Logged

Gandalf

  • Full Member
  • ***
  • Offline Offline
  • Posts: 112
Re: Ten Years after….
« Reply #76 on: February 19, 2014, 03:13:32 pm »

Hi,

Some observations, contradicting yours a little bit. Obviously I have a limited experience shooting only Sony and P45+.

Washed out highlights come from overexposure. If you expose correctly the highlights will be OK. Digital backs underexpose 1-1.5 stops, protecting highlights.

If you use "filme curve" in C1 it will push exposure 1-1.5 stops. Try checking with linear curve to see what the raw image really looks like.

Regarding colour accuracy, I measured colour accuracy on my Sony Alpha 99 and my P45+ and the Sony was much more accurate than the P45, and I have made similar measurements with similar results from published tests on Hasselblad compared to Nikon D800. The colours may be better but more accurate they are not.

I suppose everyone's experience is different, and as you pointed out, it is very important to be specific about what cameras and backs we are comparing. Perhaps my architecture test says more about the Canon than the Leaf. I will say that when I shot my Sony A900 with adapted Leica lenses against an Aptus 8/DF/80mm, while I enjoyed shooting with the Aptus/DF more, the files were extremely similar to my eye. That said, it was sunset and the light was fading quickly, and exposure times were bumping up against the capability of the Aptus back.

I would expect the colors from a Sony A99 to be more accurate than a P45+, particularly in the green/yellow spectrum. Not sure about reds. That said, I stand by my post.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Ten Years after….
« Reply #77 on: February 19, 2014, 03:21:55 pm »

Hi,

The green/yellow part was most problematic for me, I created DCP profiles that I am quite happy with.

White balance is a major factor. I may think it is a bit warmer with C1 / P45+ than with LR5 and SLT99.

Best regards
Erik

I would expect the colors from a Sony A99 to be more accurate than a P45+, particularly in the green/yellow spectrum. Not sure about reds. That said, I stand by my post.
Logged
Erik Kaffehr
 

Theodoros

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2454
Re: Ten Years after….
« Reply #78 on: February 19, 2014, 03:42:50 pm »

I thought this is a conversation for the position (and usefulness) of older 22mp "fat pixel" (9μm) backs… "ten years after" they where introduced (as being the "high end" of the days) and their position in todays market. 
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Re: Ten Years after….
« Reply #79 on: February 19, 2014, 05:14:04 pm »

My apologies for the tangent in the conversation; back you your original topic:
OK… Let's go back to the (now old) 22Mp backs then… Their DR is superb because one can "dig" deep into the shadows with less noise ...
I have seen no visual evidence of this claim that those older (and noisier) 22MP CCD sensors give better shadows or less noise than modern high resolution CCDs (let alone the coming wave of medium format CMOS sensors) in any relevant comparison of "end products", meaning viewing prints or on-screen images from various cameras at the same size. Because printing (or otherwise displaying) an image with more pixels at the same size and thus at higher PPI increases the SNR of the signal received by the viewers' eyes and improves the fineness of perceived tonal range, cleanness of shadows and such; this is due to "dithering", roughly speaking. This benefit of "more pixel per image" has to be offset against the engineer's measure of the DR of the individual pixels.

A familiar illustration is seen in comparisons between prints of the same size made using the same film emulsion in different formats: the emulsion in each case has the same "DR", but the larger prints with a lower degree of enlargement show improved tonality and cleaner shadows.

P. S. Traditional silver halide negatives have a "per pixel DR" of about 1: if you look close enough, the pixels are all either pure black (silver from an exposed silver halide crystal) or white. All tonal gradations seen are due to "dithering" of that information, due to the size scale of the "pixels" being far below what the eye can resolve.
« Last Edit: February 19, 2014, 05:45:30 pm by BJL »
Logged
Pages: 1 2 3 [4] 5   Go Up