Ad
Ad
Ad
Pages: « 1 [2]   Bottom of Page
Print
Author Topic: Canon 5DMkII ISO usage and LR/ACR  (Read 31558 times)
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 578


WWW
« Reply #20 on: January 21, 2009, 12:47:02 PM »
ReplyReply

[quote name='tlooknbill' date='Jan 21 2009, 04:24 AM' post='253509']

Thank you for this very useful tip!
Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 578


WWW
« Reply #21 on: January 24, 2009, 09:55:50 AM »
ReplyReply

A new discovery concerning light metering on the 5D2:
At EV 4, I found a gross inconsistency with the Asahi, whose output was  confirmed by the Nikon D200 and the Metered Light Pocket Spot meter. So I set up a little experiment: From the starting point at EV 4 at available daylight, I illuminated the gray card with 2 Solux lamps, whose voltage I regulated. Metering method on the 5D2 was "partial", = 8% in the center of the image.

[attachment=11076:Asahi_vs...ering__1.jpg]

When I looked at the results, I found them so incredible that I took a second run:

[attachment=11077:Asahi_vs...ering__2.jpg]

This shows at least a more transparent pattern.
The camera had been left in the "studio", but switched off,  while I looked at the results of the 1st take, and maybe it had acclimatized. The temperature in the "studio" is about +15 Celsius, vs. ca. 22 in my living room. (In the meantime, daylight had gone down so much that I needed the Solux lamps all the way.)

Conclusion: Apart from the 2/3 EV offsett that results from the ISO cheating, this gap increases to 1/1 EV below EV 6 (5 and 4).



Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 578


WWW
« Reply #22 on: January 25, 2009, 08:17:27 AM »
ReplyReply

The  conversion to DNG adds 1/3 EV to Canon raws, according to the indikation in Rawnalyze. However, if I expose a ColorChecker according to the Asahi spot meter and open the DNG in ACR, the row 4/column 4 patch shows RGB readings of 98/98/98, which is close to the correct value for Pro Photo (101/102). How come?? (Exposure 0, Brightness +50, Contrast +25, medium Tone Curve).

Logged

Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1324



WWW
« Reply #23 on: January 25, 2009, 09:02:13 PM »
ReplyReply

Quote from: eronald
The guys here are photographers, not scientists. So, exposure to them is A PRIORI exposure, meaning a decision which is taken on the ground prior to making a shot, not A POSTERIORI terms which you as a scientist use to analyze  the resultant pixel map.

In this context your remarks are misleading the poor pragmatic men and women in this forum who are trying to earn a living with their cameras. Exposure for them is something to be done with respect to an expected reflectance in a given lighting, with precautions for headroom for speculars (forehead, nose), tableware or in-image lights (lamps in architecture), and acceptable shadows.

Exposure is a term of the art that is very much alive. As a rule of thumb, I'd recommend one stop headroom over the white of a Colorchecker.
I wonder why a photographer could not be interested in knowing what's going on with the different ISO settings allowed by his camera. If you are a RAW shooter, Gabor's findings are very valuable and help maximise the quality of results of any photographer (i.e. a person who takes photographs).

One stop over the white of a Colourchecker as a general rule can easily be non-optimum, leading to loose DR in the shadows or in the highlights. I, as arquitecture and interiors photographer, would check my histogram instead, and make sure I got it exposed to the right. And would do additional overexposed shots to improve DR in the shadows if needed in medium or high DR situations.

BR
« Last Edit: January 25, 2009, 09:03:59 PM by GLuijk » Logged

Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #24 on: January 25, 2009, 09:56:57 PM »
ReplyReply

Quote from: Hening
The  conversion to DNG adds 1/3 EV to Canon raws, according to the indikation in Rawnalyze. However, if I expose a ColorChecker according to the Asahi spot meter and open the DNG in ACR, the row 4/column 4 patch shows RGB readings of 98/98/98, which is close to the correct value for Pro Photo (101/102). How come??
ACR applies the adjustment as "exposure correction", but it does not appear on the slider. This is the worse aspect of it: the vast majority of users don't know about the automatic adjustment and are misled. For example if the adjustment is positive, the users will tend to expose lower than ideal.

ADDED:

if you want to see this demoed, pls upload a raw file from your brand new 5D2.
« Last Edit: January 25, 2009, 10:34:09 PM by Panopeeper » Logged

Gabor
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #25 on: January 25, 2009, 10:46:54 PM »
ReplyReply

Quote from: eronald
The guys here are photographers, not scientists. So, exposure to them is A PRIORI exposure, meaning a decision which is taken on the ground prior to making a shot, not A POSTERIORI terms which you as a scientist use to analyze  the resultant pixel map

Edmund, the subject of this thread is a camera of the price $2700. You are talking about cameras, which cost ten times more.

1. The 5D2 shows an RGB histogram, which can be turned into a more or less raw channel histogram.

2. The 5D2 can make three exposure bracketed shots within a second, even without flapping the mirror in between. One can review the clipping indications and the histograms immediately and fire another triplet (it is the biggest idiocy from Canon to limit the bracketing so primitively. Why on earth are Japs allowed in function design, when they have no idea of photograpy?)

3. You can afford to be less vigilant with a P45+ because of the large DR than with a 5DMkII (or a lesser one). If I am sloppy with my 40D, then I burn the clouds or push the dark side of the mountain in the night.
Logged

Gabor
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 578


WWW
« Reply #26 on: January 26, 2009, 11:03:05 AM »
ReplyReply

Hi Gabor

thank you for your answer.

How can I upload a raw? The max. single upload size is 2 MB.

>ACR applies the adjustment as "exposure correction", but it does not appear on the slider.

Yes that's how I understood it. But what I still do not understand is that this does not show in the density of the CC middle gray patch.


Logged

Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #27 on: January 26, 2009, 02:21:46 PM »
ReplyReply

Quote from: Hening
How can I upload a raw? The max. single upload size is 2 MB

Use yousendit.com, it is still the best and less circumstancial of these services:

- you don't need any registration up to 25MB per file. You don't even have to use your own email address or the recipient's address, you can use fictional ones. After uploading you receive a URL and if you post that, everyone can download it 7 days long. Of course you can send it to my address directly.

- you need a free registration up to 100 MB per file

Anyway, if you want to measure the intensity ("density"), you must not apply brightness, nor contrast, nor saturation, and use "linear" instead of the curve.
Logged

Gabor
eronald
Sr. Member
****
Offline Offline

Posts: 4207



« Reply #28 on: January 26, 2009, 07:20:33 PM »
ReplyReply

Gabor, I have no quarrel with any of your experimental findings, as you know very well. However, as always we have differences over the way such findings should be internalized to make *pratical* photographics decisions *when imaging real scenes*. I am warning people who read your charts tht they had better get their heads around their meaning before they start doing dumb things like ETTR in real scenes. Burning the cloud detail  or losing nose skin  is a typical price you'll pay for listening to the ETTR sirens. As to the significance of in-camera histograms, when you relaize they are computed on sRGB previews you also understand their lack of worth!

Edmund

Quote from: Panopeeper
Edmund, the subject of this thread is a camera of the price $2700. You are talking about cameras, which cost ten times more.

1. The 5D2 shows an RGB histogram, which can be turned into a more or less raw channel histogram.

2. The 5D2 can make three exposure bracketed shots within a second, even without flapping the mirror in between. One can review the clipping indications and the histograms immediately and fire another triplet (it is the biggest idiocy from Canon to limit the bracketing so primitively. Why on earth are Japs allowed in function design, when they have no idea of photograpy?)

3. You can afford to be less vigilant with a P45+ because of the large DR than with a 5DMkII (or a lesser one). If I am sloppy with my 40D, then I burn the clouds or push the dark side of the mountain in the night.

Logged
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #29 on: January 26, 2009, 07:47:50 PM »
ReplyReply

Quote from: eronald
Gabor, I have no quarrel with any of your experimental findings, as you know very well
Edmund, be assured, that I was not reading any imaginary statements from your post :-)

Most of the differences come from the different usage of different cameras. I am shooting exclusively with a total neutral setup, which renders the embedded JPEGs practically worthless except for the recognition of the scenery, for that is helpful with my targets (landscapes and city scapes) and with my camera, which is lacking a few stops of DR (compared to my preference). This may not be the prevalent position on LL, but I am among the vast majority of DSLR owners with such problems.

Re the usefulness of in-camera histograms: the cameras need a good portion of prodding to create a histogram, which resembles the raw histogram despite demosaicing, color space conversion, white balancing and non-linear mapping, but it can be done with high reliability, at least with those cameras we have tried.
Logged

Gabor
Murray Fredericks
Sr. Member
****
Offline Offline

Posts: 290



WWW
« Reply #30 on: January 27, 2009, 12:53:20 AM »
ReplyReply


 
Gabor,

have you looked at the noise characteristics of the different ISO setting and noise?

The tradeoff at long exposures would also be interesting - higher iso vs longer exposure etc?

Cheers

Murray
Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 578


WWW
« Reply #31 on: January 27, 2009, 12:29:29 PM »
ReplyReply

Hi Gabor,

a raw file of my ColorChecker is uploaded to Yousendit.

When I take the CC image that was exposed according to the Asahi reading, open the DNG in ACR with everything zero and linear, chose the same camera profile that was used when shooting (Camera neutral), the CC 4/4 patch shows RGB values of 66. When I move the Exposure slider until the densitometer reads 100, the exposure slider has been moved to +1,10.

The Asahi exposure is 2/3 EV lower than the camera meter. So if the DNG conversion adds 1/3 EV, + 0,33 in Exposure should do the trick.

In the same image, exposed according to the camera meter, i.e. 2/3 EVs higher, the middle gray patch has RGB values of 86. Achieving 101 requires Exposure + 0,45. If the DNG converter adds +1/3, it should require -1/3.

So it looks like middle gray is middle gray AFTER application of a particular tone curve. I thought the D 0.7 was the pivoting point for the rest.

Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 578


WWW
« Reply #32 on: January 27, 2009, 04:02:55 PM »
ReplyReply

In the meantime, I have found this thread:
"Exposing to the right"
http://luminous-landscape.com/forum/index....24354&st=40

post #45 by Bill Janes
>
With the saturation method, an exposure of an 18% gray card based on a standard light meter reading will result in an image with a grey level of 18%/√2 = 12.7% of saturation, which corresponds to a pixel value of 100 in a gamma 2.2 space. The square root of 2 is to allow 0.5 EV of headroom for specular highlights. Since the standard light meter is calibrated on the basis of the equivalence of 12% reflectance, you have to add 0.5 EV of exposure if you want the gray card image to have a pixel value of 118 in the gamma 2.2 file.
>

This seems to fit quite good with some of my findings of post #19. For the ACR Camera Profile Adobe Standard, I found that the exposure according to the Asahi meter gave RGB values of 98; with +1/3 EV (if not 1/2): 118.

This is after the DNG conversion has applied its +1/3 EV and with 50/25 brightness/contrast and medium tone curve; but it is something like this I will use in the end. And Bill Janes' post refers to a gamma 2.2 color space, not linear data.

This exposure is 1/3 EV below camera metering. The latter however is the one that places the camera histogram in the middle of its window. So this should be the exposure to choose according to  Thom in "Meters don't see 18% gray", http://www.bythom.com/graycards.htm.

Oh wait: with the camera exposure and without the DNG bonus of 1/3 EV, i.e. with the Exposure slider in ACR set to -0.33, everything would come out right? - Indeed! quite close, 116!

Getting closer, bye and bye...

Logged

Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #33 on: January 27, 2009, 08:09:36 PM »
ReplyReply

Quote from: Hening
a raw file of my ColorChecker is uploaded to Yousendit
In turn, you can download http://www.panopeeper.com/Download/Hening_LL_0153+3EV.dng and http://www.panopeeper.com/Download/Hening_LL_0153-3EV.dng, only as a demo that ACR does apply the adjustment w/o telling you a word about it.

Quote
Bill Janes' post refers to a gamma 2.2 color space, not linear data

There are two "curves" involved here.

1. the mapping in sRGB or Adobe RGB or ProPhoto RGB. This is the "gamma 2.2" if you select Adobe 98 RGB.

2. the "tone curve", often mentioned as "S curve". This is, what you can set to "linear".

Anyway, IMO all this is unnecessary if you can afford pursuing true ETTR. Expose as high as you can afford in the given setting without raw saturation and adjust it in raw processing. Keep in eye: reducing the intensity in raw processing does not create noise; just the opposite, it reduces the noise.
Logged

Gabor
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #34 on: January 27, 2009, 08:18:55 PM »
ReplyReply

Quote from: Murray Fredericks
have you looked at the noise characteristics of the different ISO setting and noise?

The tradeoff at long exposures would also be interesting - higher iso vs longer exposure etc?
I'm sorry, but I don't have any long exposure shots; the longest is I think 1.6sec, and that's only one shot
Logged

Gabor
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 578


WWW
« Reply #35 on: January 28, 2009, 10:42:06 AM »
ReplyReply

Hi Gabor,

thank you for your files. If I understand it right, what they demo is that somebody who knows how to do it can manipulate the DNG file by just any amount of exposure without this being reflected in the Exposure indication in ACR.

Yes I can see that all this is irrelevant for ETTR. But as I wrote earlier, alongside with the ETTR-exposures, I want to make one exposure for the mid-tones as my reference for how dark/light the scene was.

Also, when making the camera profile in the DNG Profile Editor, I need a reference for middle gray in the visual space.

Happy I got this far with my workflow, at last.
Logged

ejmartin
Sr. Member
****
Offline Offline

Posts: 575


« Reply #36 on: January 28, 2009, 11:05:39 AM »
ReplyReply

According to John Sheehy, the in-between ISO's on the 5D2 are implemented as they are on the 40D/50D, via software multiplication after the sensor signal is quantized (this is different from the 5D, which implemented intermediate ISO's via analog amplification, as in the 1 series).  This observation means

1) That there is no difference between using ISO 125 vs underexposing ISO 100 by 1/3 stop and compensating with +1/3 EC during RAW conversion, except that by using ISO 100 one retains an extra 1/3 stop of highlight data.  Similarly for 250 (use 200 underexposed 1/3 EV), 500 (use 400 instead), 1000 (use 800 instead), etc.

2. That there is no difference between using ISO 160 vs overexposing ISO 200 by 1/3 stop and compensating with -1/3 EC during RAW conversion.  However, by using ISO 200 and overexposing, one is explicitly aware that one is doing ETTR by 1/3 stop at the time of the exposure, rather than having the camera lie to you about it.

These two points also apply to the 40D/50D, which implement the intermediate ISO's in the same way.  Note that there is a custom function on all these cameras which allows one to disable the intermediate ISO's.  

Furthermore, as long as we are on the subject of what ISO's it makes sense to use for RAW, any ISO above 1600 on Canons (or any other current DSLR, for that matter) has zero noise advantage compared to underexposing at 1600 and applying exposure compensation during RAW conversion.  As in point (1) above, underexposing at 1600 has the advantage of retaining additional stops of highlight headroom that are thrown away by using a higher ISO.
« Last Edit: January 28, 2009, 11:07:00 AM by ejmartin » Logged

emil
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1324



WWW
« Reply #37 on: January 29, 2009, 10:46:42 AM »
ReplyReply

Quote from: ejmartin
Furthermore, as long as we are on the subject of what ISO's it makes sense to use for RAW, any ISO above 1600 on Canons (or any other current DSLR, for that matter) has zero noise advantage compared to underexposing at 1600 and applying exposure compensation during RAW conversion.
This is one of the most interesting conclusions I made from your SNR curve plots. But Emil, is this an empirical rule you concluded from those noise measurements, or there is some physical limit that makes this statement be true on any present or future camera?


(the yellow part of the SNR improvement curve when rising ISO becomes flat, i.e. no improvement for rising ISO)


And another question: I never understood very well what Roger N. Clark used to call 'unity gain' ISO, i.e. the ISO value for which every extra converted photon would account for one extra ADU in the encoded RAW file. He claimed going beyond that ISO value is nonsense in terms of SNR. But I never understood how he extrapolated this no matter which bitdepth had the sensor when there should be a big difference in that ISO value for 12, 14 or 16 bit encodings. Has the 'unity gain' ISO any relation with what you said?

BR
« Last Edit: January 29, 2009, 10:51:20 AM by GLuijk » Logged

ejmartin
Sr. Member
****
Offline Offline

Posts: 575


« Reply #38 on: January 29, 2009, 10:19:37 PM »
ReplyReply

Quote from: GLuijk
This is one of the most interesting conclusions I made from your SNR curve plots. But Emil, is this an empirical rule you concluded from those noise measurements, or there is some physical limit that makes this statement be true on any present or future camera?

It has entirely to do with the noise of the ISO amplifier, which is the limiting factor in low ISO DR.  If the signal processing circuitry downstream of the photosite array had as much DR as the photosite array itself (what Roger Clark calls sensor DR, as opposed to camera DR), then there would be little distinction between underexposing at a fixed ISO, vs applying EC during RAW conversion.  The only reason (apart from jpeg generation) for the camera to have variable gain is the benefit in SNR that it confers at lower EV.  The rest of the electronics has less DR than the sensor itself, therefore that lower limit forces one to choose the ISO amplification to decide what part of the sensor DR to access.  If the rest of the electronics had a DR that met or exceeded the sensor DR, then the all that the sensor captured would be available starting at the lowest ISO, and so all that raising the ISO would do is remove parts of highlight headroom without adding any more range at the lower end.

Current Canon and Nikon offerings leave anywhere from 1-2.5 EV of DR unrealized due to shortcomings in the DR of the electronics that processes the photosite signal.  It is that which limits the utility of changing analog ISO amplification in the camera; once the read noise is dominated by the photosite noise rather than the amplifier/ADC noise, there is no further advantage to raising the ISO.  Empirically this happens to be at about ISO 1600 on current Canons, and the D3/D700 from Nikon.  I haven't crunched the numbers, but the lower amplifier noise due to the parallel ADC architecture in the D300/D3x probably means that those cameras benefit little or none from raising the ISO past 800.

Quote
And another question: I never understood very well what Roger N. Clark used to call 'unity gain' ISO, i.e. the ISO value for which every extra converted photon would account for one extra ADU in the encoded RAW file. He claimed going beyond that ISO value is nonsense in terms of SNR. But I never understood how he extrapolated this no matter which bitdepth had the sensor when there should be a big difference in that ISO value for 12, 14 or 16 bit encodings. Has the 'unity gain' ISO any relation with what you said?

Unity gain is explained by Clark as follows: "Unity Gain ISO is the ISO of the camera where the A/D converter digitizes 1 electron to 1 data number (DN) in the digital image. Further, to scale all cameras to equivalent Unity Gain ISO, a 12-bit converter is assumed. Since 1 electron (1 converted photon) is the smallest quantum that makes sense to digitize, there is little point in increasing ISO above the Unity Gain ISO".

This would be fine if there were no noise in the image processing chain.  Noise means that unity gain is merely the point where on average one electron captured raises the raw level by one unit.  That does not mean that the sensor is counting electrons, which is what would be required for there to be no advantage to raising the gain above the unity gain ISO.  But the noise is much more than one raw level at the unity gain ISO, washing out the accuracy of the raw levels relative to electron counts.  

Indeed, because the electronics downstream of the photosite array still has a substantial contribution to read noise at unity gain ISO in many cameras, raising the ISO beyond unity gain decreases their contribution relative to signal (since their contribution to noise stays fixed while the signal is amplified), and SNR improves.  Unity gain for the 1D3 is about ISO 500 if one takes the literal definition of one electron per raw level (Clark instead fudges and rescales 14 bit data to 12 bits to avoid the bit depth issue you raised), but my SNR plot at fixed exposure shows a definite improvement from ISO 400 to 800 to 1600:

http://theory.uchicago.edu/~ejm/pix/20d/te...DRwindow1d3.png
« Last Edit: January 29, 2009, 10:28:24 PM by ejmartin » Logged

emil
Pages: « 1 [2]   Top of Page
Print
Jump to:  

Ad
Ad
Ad