The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Clearing up confusion with Brumbaer/Sinar backs

B

brumbaer

Guest
Back to the original question. Why is the exposure, color balance and tint so far off with the Brumbaer DNGs?
The exposure is not off it's different :) I will not go into details, because it has to do with the way CaptureShop and I assume XPosure as well treats pixels, and it's not up to me to talk about that.
The important thing is that (ignoring hilight recovery) the number of distinct colors in the file is the same. Just imagine both files contain values form 0 to 6789. One spec defines 16378 as white and the other 6789. Both files contain the same amount of information, but one looks underexposed.

The colors are not off on my system. Have you calibrated the colors with the help of a gretag card ?

Differences between editing applications are created by different ways of treating DNG data. ACR and Lightroom give quite close results Raw Developer gives different results. Aperture gives also different results and is a special case, because it ignores the Profile data in the DNG and uses it's own profile. For that reason the gretag calibration does not influence the display in aperture.

Regards
SH
 

robsteve

Subscriber
Is there a way to embedd in the DNG a close starting point so that when opened in ACR, we don't have to move the exposure slider to lighten the image? In other words, have it open with the exposure slider alread in a good starting point.

Robert
 
T

thsinar

Guest
I believe that there is certainly a way, although probably different from one DNG application to another.

I base my belief on the fact that e.g. the Brumbaer DNG does open in ACR with the "colour noise reduction" in ACR set to a value of 25 automatically, when eXposure leaves it to 0.

But don't ask me how to do this.

Best regards,
Thierry

Is there a way to embedd in the DNG a close starting point so that when opened in ACR, we don't have to move the exposure slider to lighten the image? In other words, have it open with the exposure slider alread in a good starting point.

Robert
 
P

Panopeeper

Guest
Is there a way to embedd in the DNG a close starting point so that when opened in ACR, we don't have to move the exposure slider to lighten the image? In other words, have it open with the exposure slider alread in a good starting point.

Robert
Actually, there are several ways to achieve that via different DNG metadata. The raw-to-DNG converters/creators decide if, which way and how much. Generally, this depends on the camera, but different raw converters may decide for different ways and values.

For example the Sinar e22 at ISO 400: as there is no ISO gain, the exposure has to be adjusted in raw conversion. When converting the very same image (a sample provided by Graham), Brumbauer's instructs the raw processor to add 2.8 EV, while eXposure decides for +3.31 EV.

I guess you need to convince the makers of your favourite DNG converter, that the current setting is not good.

Gabor
 

David K

Workshop Member
Stephan,

Good to see you on this forum. As a new Sinar back user let me add my voice to the many that have thanked you for your efforts. The rest of this discussion is way beyond my comprehension. I know a good file when I see it, the rest is academic to me.
 
P

Panopeeper

Guest
I'm afraid this particular segment of the thread has been thrown out of joint.

First, let me make this clear: I do not need to be convinced about how good MFDBs are. I have not stated anywhere, anytime that 36mms or smaller were equals to MFDBs.

I took exception to two claims:

1. that a certain posted (excellent) shot was without noise removal,

2. higher ISOs with the Sinar back.

The first issue has no relevance whatsoever; it is only about the understanding of what is happening to the image. I come back to that in a separate post, for some members appear intimidated by technical details, and that issue is really technicality.

However, the ISO issue is relevant IMO. Some were asking, what this has to do with photographing. Well, it has a lot.

ISO with sensors does not mean sensitivity but gain; I guess this is common knowledge. In technical terms - sorry, so much has to be - adjusting the exposure in raw processing (a misnomer) means a multiplication of the original pixel values without adding any information. On the other hand, increasing the gain not only increases the pixel values but adds low order values to it. An example: after multiplying by two, the values 4, 5, 6 become 8, 10, 12. Nothing new, no finer shades. However, increasing the gain by one stop means, that this 4 may become 8 or 9, the 5 may become 10 or 11, etc., i.e. intermediate values, more shades are created. (If those new values are reliable or noise depends on the ISO capability of the sensor.)

The great dynamic range of MFDBs is a substitute for higher ISOs; the clean pixels of the very low range are useful even there, where other sensors offer only noise. So, there is no problem with this. For example some of the images I have analyzed in the past two days have been underexposed by three full stops (measured from the very highest possible exposure, so that is not a truly 3 EV underexposure) and still have very clean data.

Where I see a problem is the lack of consciousness on the photographer's part.

What does it mean for the photographer, when the pixel data received from the sensor does not really depend on the selected ISO?

1. Possibly avoidable underexposure. If my camera has good ISO gain at least in the lower range, let's say 200 or perhaps 400, then increasing the ISO in this range is quasy equivalent to increasing the shutter or the aperture. If the ISO setting without gain is used this way, then the underexposure is given.

2. One-two, or with an MFDB three or more stops underexposure (from the right edge) is usually not a big issue. However, if this is coupled with a fictional ISO 400, then the 3 EV underexposure becomes 5 EV or 6 EV, and that may be too much.

3. Nominal overexposure using no-gain ISO is usually no problem, for it does not necessarily induce factual clipping. However, due to the automatic adjustment by the raw converting/processing software, the factually immaculate exposure will appear overexposed (apparent clipping) in the raw processing. This can be countered by negative exposure compensation, but again, the user has to be aware of the situation.

Facit: the photographer has to be very aware of this situation, understanding, that selecting higher ISO is practically the same as a negative exposure bias.

That was my point.


Addendum

The question is now justified: if higher ISO is equivalent to negative exposure bias, then why higher ISO?

The software (at least the DNG converters, but I guess Capture One too) supports this by the automatic compensation; exposure bias would not be compensated for.
 
Last edited:
I'm a newbie in the digital arena, so feel free to correct me, but let's see if I've got this straight:

ISO with sensors does not mean sensitivity but gain; I guess this is common knowledge. In technical terms - sorry, so much has to be - adjusting the exposure in raw processing (a misnomer) means a multiplication of the original pixel values without adding any information. On the other hand, increasing the gain not only increases the pixel values but adds low order values to it. An example: after multiplying by two, the values 4, 5, 6 become 8, 10, 12. Nothing new, no finer shades. However, increasing the gain by one stop means, that this 4 may become 8 or 9, the 5 may become 10 or 11, etc., i.e. intermediate values, more shades are created. (If those new values are reliable or noise depends on the ISO capability of the sensor.)
So increasing the ISO (gain) means you get a more accurate range of tonal values (noise aside).

2. One-two, or with an MFDB three or more stops underexposure (from the right edge) is usually not a big issue. However, if this is coupled with a fictional ISO 400, then the 3 EV underexposure becomes 5 EV or 6 EV, and that may be too much.
We expose to the right edge to get a more gradated set of tonal values. Underexposure reduces the accuracy of these values.

Upping the fictional ISO will result in underexposure, which would REDUCE the quality of the tonal range, but is compensated for by the increase in levels provided by upping the gain?
 
T

thsinar

Guest
Do the eMotion backs achieve higher ISO with hardware gain or software multiplication?

Robert
It's more like a multiplication of the gain!

;)

Seriously, I know little, but what I know is that I look at the (an) image and compare it with an(other) image, at 100%, not analyzing tables, graphics and values on paper.

1. that a certain posted (excellent) shot was without noise removal, ...
Panopeeper:

- your analysis is of great value, thanks for it. I shall forward it to our engineers. May be I will get some comments/information back to share with you.

- I still wish to repeat and emphasize the following once again, since in one of your sentences is implied again that there might be a noise reduction somewhere during the DNG conversion, by saying "I took exception to two claims: a certain posted (excellent) shot was without noise removal, ....
and further then "The first issue has no relevance whatsoever ..."

It has relevance to me: there is absolutely no noise reduction being applied, neither in the Brumbaer eMotion DNG Converter, nor in the Sinar eXposure. I shall repeat here my explanation to you in response to your PM to me this morning for the understanding of all, since I feel there is still a misunderstanding and that 2 issues have possibly lead to a confusion:

There is a "denoise" OPTION in Brumbaer's DNG converter, which has NOTHING to do with a noise reduction in the image data.
The word "denoising" is Stefan's own term. There is no intention to mislead or whatsoever with the word noise reduction. And it is very clear what "denoising" is doing, namely it takes away noise from a file (called "white shading") which is applied to the image file, by subtracting its information from the image. Why does one need and use a "white shading"? When there are vignetting issues due to the lens fall-off, or due to the shifts, tilts or swings done with the camera, or when the sensor produces those shifts in terms of (colour) uniformity. The solution is to create another file, under the exactly same shooting conditions (same f-stop, same camera setting, same focus), with an opal glass in front of the lens: this will "capture" all this light (and colour) fall-off and ONLY this, which can then be subtracted from the image data. Due to the opal glass in front, it needs to be shot with about 2 stops more light than the image data on which it shall be applied. Obviously this will create sometimes (most of the time) some additional and unwanted noise which will be added to the image (by subtraction), simply because those 2 stops more light needed for the right exposure are obtained by increasing the exposure time (f-stop cannot be changed). The result is then a noisy image, but noise created by the white shading file, not by the sensor and not in the image data shot. So it has to be taken away and therefore one has the possibility (again, free choice for the user here as well) to "denoise" this white shading file before applying it to the image data. The whole has therefore absolutely nothing to do with a NR done in the image data. And that is also the very reason (or I suppose so) why Stefan calls it "denoising", not noise reduction.

Best regards,
Thierry
 
Last edited:
P

Panopeeper

Guest
Upping the fictional ISO will result in underexposure, which would REDUCE the quality of the tonal range, but is compensated for by the increase in levels provided by upping the gain?
Increasing the pixel values afterwards compensates for the lightness of the resulting image, but not for the intermediate tonal levels.

Do the eMotion backs achieve higher ISO with hardware gain or software multiplication?
By software multiplication. However, I think this needs to be explained in more detail.

Many cameras substitute certain ISO steps with numerical adjustment. My camera, the Canon 40D offers 14 ISO settings: 100 to 1600 in 1/3 stop steps and 3200. However, nine out of these are numerical derivatives of the true ISO settings; they are plainly fakes.

What is worse, this is happening in-camera (that too is software). Accordingly, ISO 3200 is roughy the doubling of the ISO 1600 values. The consequence is, that the resulting values occupy the numerical range of the sensor, effectively reducing the dynamic range by one stop without giving anything in return. One would think that this does not matter, for one uses so high ISO only when the light is low - but even in such situations there may be some bright spots, like street lights, etc. Using 3200 can cause highlight clipping. I suggest raw shooters to avoid ISO 3200, but when recording JPEG in-camera, it may be useful.

I find the solution of Sinar better, for the compensation occurs in raw processing, i.e. the higher ISOs do not reduce the dynamic range. This compensation is the equivalent of moving the "Explosure" slider of ACR.

Note: I wrote above, that there are different ways to instruct the raw processor to adjust the lightness of the resulting image. Though the effect is the same, it can be confusing. After loading the Brumbaer's converted ISO 400 image of a Sinar e22, ACR starts with "Exposure +2.80"; load the eXposure converted image, and the Exposure slider is at 0, even though ACR displays a much brighter image (0.4 EV difference). The reason is, that the two converters chose different ways to tell the raw processor to increate the lightness.

This difference has almost no relevance, but one needs to be aware of it.
 
T

thsinar

Guest
nothing adding to the debate, but I wish to correct it:

Graham has used a Sinar eMotion 54 LV for his images:
the Sinar e22 (eMotion 22) does not exist anymore since a long time. The eMotion 54 LV has a completely different electronic sensor board (among others).

Best regards,
Thierry

... image of a Sinar e22
 
P

Panopeeper

Guest
Graham has used a Sinar eMotion 54 LV for his images:
the Sinar e22 (eMotion 22) does not exist anymore since a long time
Graham posted a serie of raw files at the beginning of this thread, with ISO 50, 100, 200 and 400. The metadata of these DNG files, converted by Brumbaer's, contains

Make: SINAR
Model: eMotion 22
Unique camera model: e54 on Rollei 600x

Make and Model usually come from the camera; the unique camera model comes from the raw converter. However, I do not have the original raw files to confirm this. I though e54 is a sub-classification of eMotion 22.

The full image size is 4008x5344pix, perhaps this helps in cearing, what was happening.

Anyway, my statements regarding the ISO were not based on the a priory knowledge of the model but on the actual data in these raw files.
 
Upping the fictional ISO will result in underexposure, which would REDUCE the quality of the tonal range, but is compensated for by the increase in levels provided by upping the gain?
Maybe I didn't explain myself properly.

If the scene is underexposed, we reduce the amount of tonal levels that are being recorded around a certain luminance value, therefore we are losing detail. But you are saying that if the gain is hardware gain, those intermediate values are not actually lost, but recorded correctly. Software gain would result in those values being lost.
 

Graham Mitchell

New member
Graham posted a serie of raw files at the beginning of this thread, with ISO 50, 100, 200 and 400. The metadata of these DNG files, converted by Brumbaer's, contains

Make: SINAR
Model: eMotion 22
Unique camera model: e54 on Rollei 600x
I can confirm they were taken with an eMotion 54LV
 
P

Panopeeper

Guest
If the scene is underexposed, we reduce the amount of tonal levels that are being recorded around a certain luminance value, therefore we are losing detail. But you are saying that if the gain is hardware gain, those intermediate values are not actually lost, but recorded correctly. Software gain would result in those values being lost.
I have been thinking for a while about a fitting analogy. Here is one; I am not fully happy with it, but I don't have any better. Suggestions are welcome, we will share the royalty fees.

Let's imagine we have water in a container and we want to measure its volume.

First we measure it in gallons. If the amount is large, like in a swimming pool, we get a large number and a relatively good measurement. However, if there is much less water there, the resulting number is lower, and the difference between the measurement of two containes with almost identical amount of water is very small to zero. In order to get a more accurate measurement, we measure the leftover (less than a gallon) as well, in quarts. If the amount is even less, we have to measure the leftover after the quart, so we turn to a pint.

This way we can measure and compare small amounts as well. However, there is a downside: the smaller the unit, the less the accuracy of measurement, some measurements will be off.

A pixel well captures photons; these turn into measurable electric charge. The higher the exposure, the higher the difference in the charges of two pixels, which were not equally illuminated.

The closer we "look", the more fine measurement we can take, thus we can distinguish between like charges. At the same time, more and more measurements will be off.

This off-measurement is one source of noise.

Now, to the substitution of ISO gain by multiplication: if we measure everything in gallons and multiply the result by eight, then we get that very measurement in pints - but that's not the same, as if we had measured it in pints.
 
Last edited:
I have been thinking for a while about a fitting analogy. Here is one; I am not fully happy with it, but I don't have any better. Suggestions are welcome, we will share the royalty fees.

Let's imagine we have water in a container and we want to measure its volume.

First we measure it in gallons. If the amount is large, like in a swimming pool, we get a large number and a relatively good measurement. However, if there is much less water there, the resulting number is lower, and the difference between the measurement of two containes with almost identical amount of water is very small to zero. In order to get a more accurate measurement, we measure the leftover (less than a gallon) as well, in quarts. If the amount is even less, we have to measure the leftover after the quart, so we turn to a pint.

This way we can measure and compare small amounts as well. However, there is a downside: the smaller the unit, the less the accuracy of measurement, some measurements will be off.

A pixel well captures photons; these turn into measurable electric charge. The higher the exposure, the higher the difference in the charges of two pixels, which were not equally illuminated.

The closer we "look", the more fine measurement we can take, thus we can distinguish between like charges. At the same time, more and more measurements will be off.

This off-measurement is one source of noise.

Now, to the substitution of ISO gain by multiplication: if we measure everything in gallons and multiply the result by eight, then we get that very measurement in pints - but that's not the same, as if we had measured it in pints.
The more you explain this, the more little pieces fall into place!

However, (and please don't think the student is trying to tell the master here), I thought (unless things have changed in the last 6 or 7 years), the measurements for one 'exposure' of an image were the same size across that image's DR, which is why measurement accuracy falls off with underexposure. I can see that using software gain (and measuring in pints) doesn't change anything, but I didn't think gallons and pints were mixed during one exposure. Or are you saying that that is what effectively happens when using hardware gain?

And how are we, as non-engineers, supposed to know which ISO settings are using hardware or software-based solutions? You said in the case of the Canon it was part one and part the other.
 
P

Panopeeper

Guest
and please don't think the student is trying to tell the master here
pleeeze :)

I thought (unless things have changed in the last 6 or 7 years), the measurements for one 'exposure' of an image were the same size across that image's DR, which is why measurement accuracy falls off with underexposure
You are right. Analogies always carry the risk of misleading.

If you are judging, that the exposure won't be enough (based on metering or your feeling), you decide for the "finer units" and that applies to all pixels.

how are we, as non-engineers, supposed to know which ISO settings are using hardware or software-based solutions?
We, non-engineers (and even those of us, who are engineers) are not supposed to know this. You don't find any specification from major camera makers stating this. The maximum is, that the ISO settings past the last real one are characterized as "expansion" or "high" and often they can be blocked from accidental selection; furthermore they are excluded from auto ISO selection. Likewise, the 1/3 stop ISOs can be blocked, which is comfortable for raw shooters.

Often there is an ultra-low ISO as well, for example 50; that too causes losing one stop from the DR, and that too is treated as "extra".

However, manufacturers generally are very tight-lipped about anything, which could lead to better understanding of the cameras. For example Nikon never explained in cleartext, what the lossy compression means; they stated only, that in most cases it does not cause any image degradation. Canon do not confirm the method applied in the compression, even though it is known widely (all programs, which are processing Canon raw files on their own have to know the compression).

Perhaps this is self-generating: all feel, that they would get into a disadvantaged position by revealing some of the restrictions, because others are not revealing that and customers would make a wrong comparison.

This is happening already. Example: the Canon 1DsMkIII's highest ISO setting is 3200 (I do not know if that is real or fake, I have not seen any raw file yet). Nikon's D3 goes up to 25600. Most Nikon owners herald it as three stop advantage over the Canon, but the D3's 12800 and 25600 are fake. Sometimes I have to laugh reading the owners boosting about the greatness of ISO 25600.

Anyway, analysis of the raw files, often the histogram alone reveals these tricks. Followings are the fine histograms from the Canon 40D. Each red, green and blue column of pixels in the histogram represents a distinctive pixel level; gaps indicate, that there is no pixel with that level:

Canon 40D ISO 100

Canon 40D ISO 125

Canon 40D ISO 160

Canon 40D ISO 200

ISO 3200 (fake) vs 1600 is particularly interesting. The ISO 1600 pixel values get doubled, therefor half of the original pixel values occupy the entire numerical range of the pixel depth. Consequently, pixel values within the top EV of the dynamic range get clipped - without any gain in the shadows.

The following layered TIFF demostrates the effect:
Canon40D ISO 1600vs 3200. The file contans the exlanation of the layers.

The principle is the same for all cameras doing this trick. On the other hand, the "true gain" in details depends on the camera.

Now, compare the above with the way the Sinar back is operating: higher ISO does not affect the raw pixel data at all. See the histograms of the ISO 100 and 200 shots with the e54: the pixel values reflect the lower exposure with ISO 200. This way the "higher ISO" (which is none) does not induce clipping:

Sinar e54 ISO 100

Sinar e54 ISO 200

(The displayed shutter time is incorrect; Graham has explained, that the camera does not pass this info on the back.)
 

David K

Workshop Member
Took this shot this evening while scoping out a location to do a stitched pano. Processed the files in both Exposure and with the Brumbaer software with the latter yielding far better results due to the way it handled the setting sun...
 
Top