Ettl vs ettr

Breakfastographer · 12-26-2020, 04:56 AM

Originally posted by Class A

That does not make any sense.

ETTR is the best approach to exploit the full dynamic range of the sensor. Any lower exposure wastes some of the range.

This statement may (I haven't checked it) make sense for JPEGs. It certainly does not make sense for RAW data. The latter has a linear correspondence to the light levels in the scene and there are no "shoulders" as we know them from film, neither in the highlights, nor in the shadows.

Not sure what you mean by shoulder - the raws are clipped at whatever is the maximum value for the bit depth, e.g. a value of 16384 for each channel in 14 bits, or 4096 for 12 bits. It is my understanding based on observation that some raw converters recover more highlight detail than is actually in the recorded image, which is really trivially easy to do if you don't take it too far (i.e., it's wrong, but it "works").

**Cerebum** · 12-26-2020, 08:42 AM

Originally posted by Class A

That does not make any sense.

ETTR is the best approach to exploit the full dynamic range of the sensor. Any lower exposure wastes some of the range.

.

Yeah, people in YouTube land aren't always right, which muddies the water somewhat

texandrews · 12-26-2020, 10:10 AM

Originally posted by Class A

This statement may (I haven't checked it) make sense for JPEGs. It certainly does not make sense for RAW data. The latter has a linear correspondence to the light levels in the scene and there are no "shoulders" as we know them from film, neither in the highlights, nor in the shadows.

I can only speak from my field experience with 4 camera brands, Oly, Sony FF and APSC, Canon FF, and Pentax medium format and FF. I nearly always have to underexpose my Pentax shots versus my ones with other cameras (which is a PITA now at work, where I sometimes have to shoot Canon and my own Pentax cameras side by side, and I sometimes forget about not underexposing with the Canon's---in fact I can slightly overexpose with them. I can almost never do that with my Pentax cameras). With the exception of time-lapses and the very, very occasional grab shots, I shoot raw. I find that the 645Z is more prone to overexposure than my K1mkII, but that's maybe a bit more feeling then true difference---yet it's a somewhat strong feeling.

Quote:

It is true that the Sony sensors used in Pentax cameras have very little digital noise, so shadow recovery is well-supported but that is true for practically every modern digital camera nowadays.

I don't find that to be true of the Canon's I'm using, 5DmkIV's.

Quote:

The JPEG engine within the camera may or may not have a harsher transition to clipping than other cameras, I really don't know, but I'd say anyone wondering about the level of improvement that ETTR provides over regular exposure, should definitely shoot RAW, instead of JPEG, and then forget about what the in-camera tone curves for JPEG development look like.

I don't think one should forget about them entirely, because they are all we have, but it continues to be aggravating that we can't have a raw histo readout in 2020. Should be a standard feature---if shooting in raw the readout is for the raw, and if in jpg for the jpg. Maybe this is a stupid statement for a technical reason I don't yet know about.

**GUB** · 12-26-2020, 01:57 PM

In Pentax raw ETTR is the way to go to get max dynamic range with minimum Noise.
But if you are utilising iso to achieve this ETTR then you are kidding yourself. The result will be no different in post to the mid range histo of the same shutter and aperture at base iso.
Better to think of ETTR at base iso.

**Cerebum** · 12-26-2020, 02:58 PM

Originally posted by GUB

In Pentax raw ETTR is the way to go to get max dynamic range with minimum Noise.
But if you are utilising iso to achieve this ETTR then you are kidding yourself. The result will be no different in post to the mid range histo of the same shutter and aperture at base iso.
Better to think of ETTR at base iso.

Do you mean high iso or iso in general? I am currently staying below iso640

**GUB** · 12-26-2020, 03:17 PM

Originally posted by Cerebum

Do you mean high iso or iso in general? I am currently staying below iso640

Iso in general.
Once you take the iso out of the equation then it becomes obvious - the more light you can put on the sensor via your aperture and shutter the better signal to noise ratio you have. But increasing your iso to achieve this won't improve your signal / noise ratio.

In case it is not understood, our modern sensors are iso invariant which means there is no difference in quality between increasing iso when taking the shot and increasing exposure in post processing.

Class A · 12-26-2020, 03:49 PM

Originally posted by Breakfastographer

Not sure what you mean by shoulder...

I mentioned that term when talking about JPEG processing (not RAW data).

The term refers to the roll-off at the highlights (the corresponding roll-off at the shadows is sometimes referred to as "toe"). The presence of "toe" and "shoulder" results in an "S-curve", i.e., a non-linear translation of scene values into image data.

**ehrwien** · 12-28-2020, 01:08 PM

Originally posted by GUB

But if you are utilising iso to achieve this ETTR then you are kidding yourself. The result will be no different in post to the mid range histo of the same shutter and aperture at base iso.

Reminds me of the time I changed my drive mode to Bracketed Exposures thinking I would maximise the dynamic range captured - it took a few minutes for me to recognise that I had been shooting in TAv mode prior to that

**Cerebum** · 12-29-2020, 01:01 AM

OK, I am seeing a definite improvement in the final image. It is much closer to what I saw on the day. Everything seems to really pop. This isn't a fascinating image but it was a good test. For once I had decent light. When I was underexposing, the colours weren't as rich. I have done very little to this image.

There's also quite a difference between the jpeg histogram on the camera and the raw histogram in lightroom. I am going to keep comparing so that I get it fixed in my mind where the raw exposure is in relation to the on camera histo. I feel like I am moving forward again. Thanks guys for all the input

**GUB** · 12-29-2020, 01:22 AM - **1 Like**

Originally posted by Cerebum

OK, I am seeing a definite improvement in the final image. It is much closer to what I saw on the day. Everything seems to really pop. This isn't a fascinating image but it was a good test. For once I had decent light. When I was underexposing, the colours weren't as rich. I have done very little to this image.

There's also quite a difference between the jpeg histogram on the camera and the raw histogram in lightroom. I am going to keep comparing so that I get it fixed in my mind where the raw exposure is in relation to the on camera histo. I feel like I am moving forward again. Thanks guys for all the input

My rather basic understanding of the camera histogram is it reflects the current settings of the jpeg even though you are in raw and not concerned with those settings. So if you you want a consistent relationship between the in camera histo and LR it probably is a good idea tohave things like highlight correction off "auto" and probably off all together. Any "auto" setting could cause variation.
I find the opening Darktable histogram has a small margin of error (less than a stop) of spare highlight compared to on camera.

Class A · 12-29-2020, 01:45 AM

Originally posted by GUB

My rather basic understanding of the camera histogram is it reflects the current settings of the jpeg even though you are in raw and not concerned with those settings.

That's correct.

It is worth playing with JPEG settings, like contrast, fir instance, to achieve better predictability regarding the highlights headroom.

However, most will strive to maintain a balance between predictability and the playback image still being useful for visual feedback.

stevebrot · 12-29-2020, 11:26 AM - **1 Like**

Originally posted by Cerebum

There's also quite a difference between the jpeg histogram on the camera and the raw histogram in lightroom.

Neither of those is the same as the RAW histogram from a tool such as RawDigger. The lack of a graph of the actual response data makes it difficult to guide exposure from the histogram.

Steve

**GUB** · 12-29-2020, 01:50 PM

Originally posted by stevebrot

The lack of a graph of the actual response data makes it difficult to guide exposure from the histogram.

It would have to be a very precise exposure requirement for this to hold true.
As discussed above there is a need for personal calibration and consistency of jpg settings to work the histogram but that is not difficult.
For me the main requirement is no clipping so I am happy to live with that margin of error built in. In the unlikely event of being a complete stop unintentionally underexposed it is not a biggy in post processing. Iso invariance means you have just taken the shot at one stop more Iso.

stevebrot · 12-29-2020, 02:52 PM

Originally posted by GUB

It would have to be a very precise exposure requirement for this to hold true.

I guess it would be more helpful if I were to provide actual comparison histograms. The difference is pretty striking for RawDigger vs. Lightroom.

I agree that conservative in-camera JPEG settings help provide a more dependable histogram and that using that as a guide to avoid clipping should be quite adequate. At least, that is what I do.

Steve

**GUB** · 12-29-2020, 02:59 PM

Originally posted by stevebrot

The difference is pretty striking for RawDigger vs. Lightroom.

I ended up with an argument elsewhere about raw digger. At the time I was convinced that it had highlight reconstruction on by default. Turning it on in DT had an identical response. Highlight reconstruction messes with the highlight end of the histo so if you are trying to expose via the histo it complicates things.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
K-5 vs MZ-S vs LX vs PZ-1p vs ist*D vs K10D vs K20D vs K-7 vs.......	Steelski	Pentax K-5 & K-5 II	2	06-28-2017 04:59 PM
ISO Invariance ETTL	indy	Photographic Technique	14	03-30-2016 06:42 AM
Should Pentaxians use ETTR?	y0chang	Pentax DSLR Discussion	13	12-18-2014 01:55 AM
K10D and ETTR & High ISO	Crosshair	Pentax DSLR Discussion	2	02-22-2014 11:43 AM
Wireless ettl	P12	Pentax Camera and Field Accessories	8	10-01-2011 04:10 AM

12-26-2020, 04:56 AM	#16
Breakfastographer Pentaxian Join Date: Nov 2018 Posts: 673	Originally posted by Class A That does not make any sense. ETTR is the best approach to exploit the full dynamic range of the sensor. Any lower exposure wastes some of the range. This statement may (I haven't checked it) make sense for JPEGs. It certainly does not make sense for RAW data. The latter has a linear correspondence to the light levels in the scene and there are no "shoulders" as we know them from film, neither in the highlights, nor in the shadows. Not sure what you mean by shoulder - the raws are clipped at whatever is the maximum value for the bit depth, e.g. a value of 16384 for each channel in 14 bits, or 4096 for 12 bits. It is my understanding based on observation that some raw converters recover more highlight detail than is actually in the recorded image, which is really trivially easy to do if you don't take it too far (i.e., it's wrong, but it "works").

12-26-2020, 08:42 AM	#17
Cerebum Site Supporter Join Date: Oct 2017 Location: Lancaster Photos: Gallery Posts: 3,828 Original Poster	Originally posted by Class A That does not make any sense. ETTR is the best approach to exploit the full dynamic range of the sensor. Any lower exposure wastes some of the range. . Yeah, people in YouTube land aren't always right, which muddies the water somewhat

12-26-2020, 10:10 AM	#18
texandrews Moderator Join Date: Apr 2014 Location: Baltimore Photos: Albums Posts: 3,390	Originally posted by Class A This statement may (I haven't checked it) make sense for JPEGs. It certainly does not make sense for RAW data. The latter has a linear correspondence to the light levels in the scene and there are no "shoulders" as we know them from film, neither in the highlights, nor in the shadows. I can only speak from my field experience with 4 camera brands, Oly, Sony FF and APSC, Canon FF, and Pentax medium format and FF. I nearly always have to underexpose my Pentax shots versus my ones with other cameras (which is a PITA now at work, where I sometimes have to shoot Canon and my own Pentax cameras side by side, and I sometimes forget about not underexposing with the Canon's---in fact I can slightly overexpose with them. I can almost never do that with my Pentax cameras). With the exception of time-lapses and the very, very occasional grab shots, I shoot raw. I find that the 645Z is more prone to overexposure than my K1mkII, but that's maybe a bit more feeling then true difference---yet it's a somewhat strong feeling. Quote: It is true that the Sony sensors used in Pentax cameras have very little digital noise, so shadow recovery is well-supported but that is true for practically every modern digital camera nowadays. I don't find that to be true of the Canon's I'm using, 5DmkIV's. Quote: The JPEG engine within the camera may or may not have a harsher transition to clipping than other cameras, I really don't know, but I'd say anyone wondering about the level of improvement that ETTR provides over regular exposure, should definitely shoot RAW, instead of JPEG, and then forget about what the in-camera tone curves for JPEG development look like. I don't think one should forget about them entirely, because they are all we have, but it continues to be aggravating that we can't have a raw histo readout in 2020. Should be a standard feature---if shooting in raw the readout is for the raw, and if in jpg for the jpg. Maybe this is a stupid statement for a technical reason I don't yet know about.

12-26-2020, 01:57 PM	#19
GUB Loyal Site Supporter Join Date: Aug 2012 Location: Wanganui Photos: Gallery \| Albums Posts: 5,760	In Pentax raw ETTR is the way to go to get max dynamic range with minimum Noise. But if you are utilising iso to achieve this ETTR then you are kidding yourself. The result will be no different in post to the mid range histo of the same shutter and aperture at base iso. Better to think of ETTR at base iso.

12-26-2020, 02:58 PM	#20
Cerebum Site Supporter Join Date: Oct 2017 Location: Lancaster Photos: Gallery Posts: 3,828 Original Poster	Originally posted by GUB In Pentax raw ETTR is the way to go to get max dynamic range with minimum Noise. But if you are utilising iso to achieve this ETTR then you are kidding yourself. The result will be no different in post to the mid range histo of the same shutter and aperture at base iso. Better to think of ETTR at base iso. Do you mean high iso or iso in general? I am currently staying below iso640

12-26-2020, 03:17 PM	#21
GUB Loyal Site Supporter Join Date: Aug 2012 Location: Wanganui Photos: Gallery \| Albums Posts: 5,760	Originally posted by Cerebum Do you mean high iso or iso in general? I am currently staying below iso640 Iso in general. Once you take the iso out of the equation then it becomes obvious - the more light you can put on the sensor via your aperture and shutter the better signal to noise ratio you have. But increasing your iso to achieve this won't improve your signal / noise ratio. In case it is not understood, our modern sensors are iso invariant which means there is no difference in quality between increasing iso when taking the shot and increasing exposure in post processing.

12-26-2020, 03:49 PM	#22
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by Breakfastographer Not sure what you mean by shoulder... I mentioned that term when talking about JPEG processing (not RAW data). The term refers to the roll-off at the highlights (the corresponding roll-off at the shadows is sometimes referred to as "toe"). The presence of "toe" and "shoulder" results in an "S-curve", i.e., a non-linear translation of scene values into image data.

12-28-2020, 01:08 PM	#23
ehrwien Site Supporter Join Date: May 2016 Posts: 2,782	Originally posted by GUB But if you are utilising iso to achieve this ETTR then you are kidding yourself. The result will be no different in post to the mid range histo of the same shutter and aperture at base iso. Reminds me of the time I changed my drive mode to Bracketed Exposures thinking I would maximise the dynamic range captured - it took a few minutes for me to recognise that I had been shooting in TAv mode prior to that

12-29-2020, 01:01 AM	#24
Cerebum Site Supporter Join Date: Oct 2017 Location: Lancaster Photos: Gallery Posts: 3,828 Original Poster	OK, I am seeing a definite improvement in the final image. It is much closer to what I saw on the day. Everything seems to really pop. This isn't a fascinating image but it was a good test. For once I had decent light. When I was underexposing, the colours weren't as rich. I have done very little to this image. There's also quite a difference between the jpeg histogram on the camera and the raw histogram in lightroom. I am going to keep comparing so that I get it fixed in my mind where the raw exposure is in relation to the on camera histo. I feel like I am moving forward again. Thanks guys for all the input

12-29-2020, 01:22 AM - 1 Like	#25
GUB Loyal Site Supporter Join Date: Aug 2012 Location: Wanganui Photos: Gallery \| Albums Posts: 5,760	Originally posted by Cerebum OK, I am seeing a definite improvement in the final image. It is much closer to what I saw on the day. Everything seems to really pop. This isn't a fascinating image but it was a good test. For once I had decent light. When I was underexposing, the colours weren't as rich. I have done very little to this image. There's also quite a difference between the jpeg histogram on the camera and the raw histogram in lightroom. I am going to keep comparing so that I get it fixed in my mind where the raw exposure is in relation to the on camera histo. I feel like I am moving forward again. Thanks guys for all the input My rather basic understanding of the camera histogram is it reflects the current settings of the jpeg even though you are in raw and not concerned with those settings. So if you you want a consistent relationship between the in camera histo and LR it probably is a good idea tohave things like highlight correction off "auto" and probably off all together. Any "auto" setting could cause variation. I find the opening Darktable histogram has a small margin of error (less than a stop) of spare highlight compared to on camera.

12-29-2020, 01:45 AM	#26
Class A Pentaxian Join Date: Aug 2008 Location: Wellington, New Zealand Posts: 11,251	Originally posted by GUB My rather basic understanding of the camera histogram is it reflects the current settings of the jpeg even though you are in raw and not concerned with those settings. That's correct. It is worth playing with JPEG settings, like contrast, fir instance, to achieve better predictability regarding the highlights headroom. However, most will strive to maintain a balance between predictability and the playback image still being useful for visual feedback.

12-29-2020, 11:26 AM - 1 Like	#27
stevebrot Otis Memorial Pentaxian Join Date: Mar 2007 Location: Vancouver (USA) Photos: Gallery \| Albums Posts: 42,007	Originally posted by Cerebum There's also quite a difference between the jpeg histogram on the camera and the raw histogram in lightroom. Neither of those is the same as the RAW histogram from a tool such as RawDigger. The lack of a graph of the actual response data makes it difficult to guide exposure from the histogram. Steve

12-29-2020, 01:50 PM	#28
GUB Loyal Site Supporter Join Date: Aug 2012 Location: Wanganui Photos: Gallery \| Albums Posts: 5,760	Originally posted by stevebrot The lack of a graph of the actual response data makes it difficult to guide exposure from the histogram. It would have to be a very precise exposure requirement for this to hold true. As discussed above there is a need for personal calibration and consistency of jpg settings to work the histogram but that is not difficult. For me the main requirement is no clipping so I am happy to live with that margin of error built in. In the unlikely event of being a complete stop unintentionally underexposed it is not a biggy in post processing. Iso invariance means you have just taken the shot at one stop more Iso.

12-29-2020, 02:52 PM	#29
stevebrot Otis Memorial Pentaxian Join Date: Mar 2007 Location: Vancouver (USA) Photos: Gallery \| Albums Posts: 42,007	Originally posted by GUB It would have to be a very precise exposure requirement for this to hold true. I guess it would be more helpful if I were to provide actual comparison histograms. The difference is pretty striking for RawDigger vs. Lightroom. I agree that conservative in-camera JPEG settings help provide a more dependable histogram and that using that as a guide to avoid clipping should be quite adequate. At least, that is what I do. Steve

12-29-2020, 02:59 PM	#30
GUB Loyal Site Supporter Join Date: Aug 2012 Location: Wanganui Photos: Gallery \| Albums Posts: 5,760	Originally posted by stevebrot The difference is pretty striking for RawDigger vs. Lightroom. I ended up with an argument elsewhere about raw digger. At the time I was convinced that it had highlight reconstruction on by default. Turning it on in DT had an identical response. Highlight reconstruction messes with the highlight end of the histo so if you are trying to expose via the histo it complicates things.