Ugly Hedgehog - Photography Forum
Home Active Topics Newest Pictures Search Login Register
Main Photography Discussion
ETTR TO THE FAR RIGHT
Page 1 of 5 next> last>>
May 26, 2014 09:31:58   #
abc1234 Loc: Elk Grove Village, Illinois
 
CAUTION: What follows will be very technical and is not for everyone. This is a long, detailed post.

The idea for this post comes from a reference in another forum. Since I would like to understand this approach so I can decide to incorporate it into my exposure method, I will keep my comments as neutral as possible and invite all objective, clear and rational responses. Here is the link to the reference.

http://www.johnshawphoto.com/category/exposure/

I shot the appropriate photos as raw's according to this link and looked at them in LR. I cannot post the pictures because they are raw's and the jpg's will not reflect accurately what we need to know. Anyone interested in getting the raw's should let me know and I will send them by dropbox. Here is specifically what I did.

1. I set the camera on a tripod and took a basic picture at ISO 800 (to introduce noise), aperture priority (f/7.1) and 1/30. Canon 60D, 18-200 mm lens.
2. I switched to manual exposure and increased exposure until "blinkies" started to show. This occurred at f/7.1 and 1/20.
3. I continued increasing the shutter speed in increments equivalent to 1/3 stop. This obviously made the pictures lighter and shifted the histogram to the right.

The closest I could get the histogram to the right was f/7.1 and 1/10. This is the end of the technical, factual part of the experiment.

This last shot was way too bright for my taste and I presume most people would say it is severely overexposed. As expected, when I zoom into a shadow, I find good detail and no noise. When I decrease the brightness to a pleasing level, the histogram obviously shifts to the left. Here is the rub. Comparing this adjusted photo with the one in which blinkies first occurred, both pictures have the same shadow detail, noise and comparable mid-tones and highlights. Bear in mind that I am not a pixel-peeker. While I looked at these pictures and 1:1 and 3:1, I base my final conclusion upon "normal" viewing distance. The closer you look, the more you will find wrong.

Here are my conclusions and I invite people to comment critically, clearly and objectively.

1. Exposing beyond the right and adjusting the brightness to a desirable level do not improve noise or shadow detail.
2. The overall appearance of the standard exposed-to-the-right and adjusted exposed-beyond-the-right pictures are comparable.

Since I am looking to improve my picture taking, I am hoping someone will find something wrong with I have done and offer a compelling reason supported by examples to merit exposing beyond the right. Again, I am willing to share my raw's.

Reply
May 26, 2014 09:59:08   #
WAL
 
Can we assume that this is all done in raw? If you are doing this in JPG it has no meaning.
The idea was originally based on comments from Thomas Knowl who with his brother developed Photoshop. It is based on the distribution of information in the digital file. There is more information in the upper end. Shifting the exposure to the highlights as high as possible w/o clipping will improve shadow detail. The idea started in 2003. This link goes to the source on Luminous Landscape.

http://www.luminous-landscape.com/tutorials/expose-right.shtml

Reply
May 26, 2014 11:11:13   #
minniev Loc: MIssissippi
 
abc1234 wrote:
CAUTION: What follows will be very technical and is not for everyone. This is a long, detailed post.

The idea for this post comes from a reference in another forum. Since I would like to understand this approach so I can decide to incorporate it into my exposure method, I will keep my comments as neutral as possible and invite all objective, clear and rational responses. Here is the link to the reference.

http://www.johnshawphoto.com/category/exposure/

I shot the appropriate photos as raw's according to this link and looked at them in LR. I cannot post the pictures because they are raw's and the jpg's will not reflect accurately what we need to know. Anyone interested in getting the raw's should let me know and I will send them by dropbox. Here is specifically what I did.

1. I set the camera on a tripod and took a basic picture at ISO 800 (to introduce noise), aperture priority (f/7.1) and 1/30. Canon 60D, 18-200 mm lens.
2. I switched to manual exposure and increased exposure until "blinkies" started to show. This occurred at f/7.1 and 1/20.
3. I continued increasing the shutter speed in increments equivalent to 1/3 stop. This obviously made the pictures lighter and shifted the histogram to the right.

The closest I could get the histogram to the right was f/7.1 and 1/10. This is the end of the technical, factual part of the experiment.

This last shot was way too bright for my taste and I presume most people would say it is severely overexposed. As expected, when I zoom into a shadow, I find good detail and no noise. When I decrease the brightness to a pleasing level, the histogram obviously shifts to the left. Here is the rub. Comparing this adjusted photo with the one in which blinkies first occurred, both pictures have the same shadow detail, noise and comparable mid-tones and highlights. Bear in mind that I am not a pixel-peeker. While I looked at these pictures and 1:1 and 3:1, I base my final conclusion upon "normal" viewing distance. The closer you look, the more you will find wrong.

Here are my conclusions and I invite people to comment critically, clearly and objectively.

1. Exposing beyond the right and adjusting the brightness to a desirable level do not improve noise or shadow detail.
2. The overall appearance of the standard exposed-to-the-right and adjusted exposed-beyond-the-right pictures are comparable.

Since I am looking to improve my picture taking, I am hoping someone will find something wrong with I have done and offer a compelling reason supported by examples to merit exposing beyond the right. Again, I am willing to share my raw's.
CAUTION: What follows will be very technical and i... (show quote)



I hope this very interesting topic will survive a while but it is already sinking in the page list so I'll revive it.

It appears that you did your experiment at an ISO of 800. It is my impression that these experiments are best conducted at base ISO.

I know ETTR helps me get more out of the relatively small sensor in my m43 camera. I first read of it on Luminous Landscape, with the article referenced in this thread by WAL above. Most of my references, for obvious reasons, are related to the use of ETTR for my type of sensor, but some have good applicability for all types of cameras/sensors. I found this article to be very readable and detailed. http://www.dpreview.com/articles/6641165460/ettr-exposed This is the second of Gollywop's articles, the first is linked to this one.

Reply
 
 
May 26, 2014 11:14:45   #
abc1234 Loc: Elk Grove Village, Illinois
 
Wal, you are indeed correct. Everything was shot raw. The Shaw article makes that point. Thanks for the history. Shaw had it too but I did not look at it. Do you have an opinion on all this?

Reply
May 26, 2014 12:09:51   #
abc1234 Loc: Elk Grove Village, Illinois
 
minniev wrote:
I hope this very interesting topic will survive a while but it is already sinking in the page list so I'll revive it.

It appears that you did your experiment at an ISO of 800. It is my impression that these experiments are best conducted at base ISO.

I know ETTR helps me get more out of the relatively small sensor in my m43 camera. I first read of it on Luminous Landscape, with the article referenced in this thread by WAL above. Most of my references, for obvious reasons, are related to the use of ETTR for my type of sensor, but some have good applicability for all types of cameras/sensors. I found this article to be very readable and detailed. http://www.dpreview.com/articles/6641165460/ettr-exposed This is the second of Gollywop's articles, the first is linked to this one.
I hope this very interesting topic will survive a ... (show quote)


Hi Minnie. I do not care if it sinks. I know this is long and complicated but I am sure I will get some good answers, agreeing or disagreeing.

I wanted noise because someone claimed that the method also reduces it. I used to stay at 100 as much as possible but with my Canon 60D, I see no appreciable noise in well-lit scenes.

I am glad you use ETTR. I think it is the way to go and think the people who shoot manually are merely guessing at exposure. I did enough of that in my film days and find ETTR a pretty good way of going for myself. I will read that article later.

Thanks for your interest and I will see you in other discussions.

Reply
May 26, 2014 12:15:04   #
selmslie Loc: Fernandina Beach, FL, USA
 
abc1234 wrote:
CAUTION: What follows will be very technical and is not for everyone. ...

Try to cut through some of the fog in all of these discussions. Proponents of ETTR point out that the technique usually improves the rendition of tonalities (virtually impossible to prove) and reduces noise but they overlook the simple fact that that is seldom necessary.

At low ISO (200 or less) noise is almost never a problem with one of the better cameras, even with a crop sensor. You can go to the trouble of calculating the point beyond the blinkies and the level at which you actually start to blow out your highlights in your raw image at base ISO, but you may find that this margin is not the same at higher ISO values or with a different subject.

If you do not have both a lot of dark as well as bright areas in a contrasty scene, you will seldom notice noise in the dark areas at low ISOs. If any shows up it is easy to post-process it out. You will always have noise at high ISOs to some degree depending on the camera and ISO setting and the noise will progressively spread from the darker into the lighter areas as the ISO increases.

If you are not sure of the proper exposure, simply bracketing your shots will provide you the insurance you need without agonizing over the technicalities.

Reply
May 26, 2014 12:22:08   #
abc1234 Loc: Elk Grove Village, Illinois
 
Thanks Scotty. We are pretty much on the same page, especially about cutting through the fog.

Not only have I used ETTR, I have advocated it for years while being open to better methods. Despite the hype, I cannot say EBTR is such an improvement.

My philosophy is to shoot the best possible picture presuming ETTR allows you to capture the most information about the scene. Then in post-processing, I can throw away any information I want. I cannot throw away information that is not there anymore than I can make up information. So far, I am delighted with this approach.

Reply
 
 
May 26, 2014 16:12:44   #
selmslie Loc: Fernandina Beach, FL, USA
 
abc1234 wrote:
... We are pretty much on the same page, especially about cutting through the fog. ...

To take another slice at the fog, consider this.

Suppose that you meter a scene for ISO 400 and decide that a "normal" exposure would be f/11 at 1/500. You decide that a 1-stop exposure to the right would reduce your noise level without blowing your highlights, so you open up to f/8 or cut the shutter speed to 1/250. Either of these adjustments doubles the amount of light throughout the image and doubles the signal-to-noise (S/N) ratio for all zones. So far, so good.

Now suppose that instead, you simply lower the ISO to 200 and don't make any adjustment for ETTR. You still might expose at f/11 and 1/250 or f/8 and 1/500 and you would get the same amount of light and the same S/N ratios throughout as in the ETTR version metered at ISO 400.

There is no difference between either of these methods when it comes to noise avoidance. But ETTR advocates have another, more complicated argument.

Bit-wise, the crudest graduation between values within the raw file happens in the darkest zones. For example, if Zone I has 8 value steps, Zone II will have 16, etc. and Zone X will have 4,096 steps. Expose one step to the right and the number of steps in each zone doubles. From a purely mathematical viewpoint it would seem logical that this would matter, but in reality it does not. Can we really see tonality differences or details within Zone I or Zone II? Probably not. And do we really need 4,000 or 8,000 tonality steps in the highest values to see the shapes of clouds or snow? Not likely.

In reality the human eye cannot really distinguish more than about 256 graduations from darkest black to paper white. Line them up across your screen or print and they will just look like a smooth and gradual transition.

A 12- or 14-bit raw file gets converted to a 256-bit JPEG or a 65,536-bit TIFF file in a non-linear fashion. After conversion, each zone ends up with the same number of graduations. So what you are working with in your editor has already spread out the graduations and leveled the playing field for all zones. The extra bits in a TIFF file are there to prevent banding while adjusting contrast and colors.

Reply
May 26, 2014 16:35:08   #
abc1234 Loc: Elk Grove Village, Illinois
 
selmslie wrote:
To take another slice at the fog, consider this.

Suppose that you meter a scene for ISO 400 and decide that a "normal" exposure would be f/11 at 1/500. You decide that a 1-stop exposure to the right would reduce your noise level without blowing your highlights, so you open up to f/8 or cut the shutter speed to 1/250. Either of these adjustments doubles the amount of light throughout the image and doubles the signal-to-noise (S/N) ratio for all zones. So far, so good.

Now suppose that instead, you simply lower the ISO to 200 and don't make any adjustment for ETTR. You still might expose at f/11 and 1/250 or f/8 and 1/500 and you would get the same amount of light and the same S/N ratios throughout as in the ETTR version metered at ISO 400.

There is no difference between either of these methods when it comes to noise avoidance. But ETTR advocates have another, more complicated argument.

Bit-wise, the crudest graduation between values within the raw file happens in the darkest zones. For example, if Zone I has 8 value steps, Zone II will have 16, etc. and Zone X will have 4,096 steps. Expose one step to the right and the number of steps in each zone doubles. From a purely mathematical viewpoint it would seem logical that this would matter, but in reality it does not. Can we really see tonality differences or details within Zone I or Zone II? Probably not. And do we really need 4,000 or 8,000 tonality steps in the highest values to see the shapes of clouds or snow? Not likely.

In reality the human eye cannot really distinguish more than about 256 graduations from darkest black to paper white. Line them up across your screen or print and they will just look like a smooth and gradual transition.

A 12- or 14-bit raw file gets converted to a 256-bit JPEG or a 65,536-bit TIFF file in a non-linear fashion. After conversion, each zone ends up with the same number of graduations. So what you are working with in your editor has already spread out the graduations and leveled the playing field for all zones. The extra bits in a TIFF file are there to prevent banding while adjusting contrast and colors.
To take another slice at the fog, consider this. b... (show quote)


At a light level of ISO 400, f/11, 1/500, I would not expect any noise except, perhaps, in a very contrasty situation. As much as I like the science and technology of photography, ultimately I am a practitioner and all I know is that the results improve every year, you get more function for the money, and the process gets easier and faster. And you and I still agree.

Reply
May 26, 2014 17:54:38   #
selmslie Loc: Fernandina Beach, FL, USA
 
abc1234 wrote:
... all I know is that the results improve every year, you get more function for the money, and the process gets easier and faster. And you and I still agree.

When the original Luminous Landscape tutorial appeared in 2003 the ETTR argument was based on a 12-but raw image. Even when it was updated in 2011, the 12-bit raw file was still being presented as typical despite the fact that manufacturers had been offering a 14-bit raw file option for some time. So 14-bit raw files pretty much obsoleted that part of the ETTR argument.

During the same time, noise present in the sensor and capture process has been dropping significantly - to the extent that newer cameras are performing with much better S/N ratios well beyond 400 or 800 ISO. This is making the noise argument a diminishing reason for applying ETTR methods.

I am certainly not averse to following old methods in order to be on the safe side. After all, I still do more film than digital simply because I have a preference for medium and large formats. But I do so with my eyes open and I try to keep up with the benefits and trade-offs of doing the extra work.

Reply
May 26, 2014 18:25:03   #
abc1234 Loc: Elk Grove Village, Illinois
 
selmslie wrote:
When the original Luminous Landscape tutorial appeared in 2003 the ETTR argument was based on a 12-but raw image. Even when it was updated in 2011, the 12-bit raw file was still being presented as typical despite the fact that manufacturers had been offering a 14-bit raw file option for some time. So 14-bit raw files pretty much obsoleted that part of the ETTR argument.

During the same time, noise present in the sensor and capture process has been dropping significantly - to the extent that newer cameras are performing with much better S/N ratios well beyond 400 or 800 ISO. This is making the noise argument a diminishing reason for applying ETTR methods.

I am certainly not averse to following old methods in order to be on the safe side. After all, I still do more film than digital simply because I have a preference for medium and large formats. But I do so with my eyes open and I try to keep up with the benefits and trade-offs of doing the extra work.
When the original Luminous Landscape tutorial appe... (show quote)


Thank you for the history and putting this whole thing into perspective. And happy developing. I always thought large format gave the best prints by far.

Reply
 
 
May 26, 2014 19:11:16   #
pete-m Loc: Casper, WY
 
That looks like the digital explanation of the "expose for the highlights" from the film days. The idea was to prevent any blockage of detail of the whites in the prints and hopefully you could bring out the shadow detail during printing. Of course you ignore the spectral highlights. Used for high contrast subjects.

Pete :D

Reply
May 26, 2014 19:38:50   #
abc1234 Loc: Elk Grove Village, Illinois
 
pete-m wrote:
That looks like the digital explanation of the "expose for the highlights" from the film days. The idea was to prevent any blockage of detail of the whites in the prints and hopefully you could bring out the shadow detail during printing. Of course you ignore the spectral highlights. Used for high contrast subjects.

Pete :D


I guess not much has changed other than to give it a new name and for a real-time histogram. Now, you do not have to wait to take the film out of the hypo to have a sense of how it will print. Just look at the histogram. It is not the final word but it does help find a reasonable starting point.

Reply
May 26, 2014 20:14:23   #
rebride
 
Expose for the highlights applied to slides/positve film.
For BW negative it is expose for the shadows, develop for the highlights.
My Fuji dynamic range modes for contrast control expose to the left. It stops down 1 or 2 stops (for the raw file) from the set exposure. It than holds the highlights in place and tone maps the lower end up for the jpg image.
Expose for the highlights, develop for shadows.
ETTL ???

Reply
May 26, 2014 21:07:53   #
MW
 
abc1234 wrote:
Hi Minnie. I do not care if it sinks. I know this is long and complicated but I am sure I will get some good answers, agreeing or disagreeing.

I wanted noise because someone claimed that the method also reduces it. .


"Avoid" may be a better term than "reduce". Noise appears more in the darker parts of a digital image than in the lighter. I think this is because the noise is constant across the sensor but the brighter parts of the image the more information hence the signal to noise ratio is lowest in the darker parts and higher in the brighter regions. So if ETTR is working as claimed when you adjust the exposure in PP the noise in the darker regions is driven toward the black point.

I know from personal experience that the opposite is true, i.e., if you underexpose and try to compensate in PP previously unperceived noise can appear in the shadows.

Lots of detail or texture can obscure noise and this needs to be considered when making your experiments. If there is a lot of detail in the shadows you won't see the noise so much although it is really there.

Disclaimer: the above is my understanding of the subject and there is no guarantee my understanding is perfect.

Reply
Page 1 of 5 next> last>>
If you want to reply, then register here. Registration is free and your account is created instantly, so you can post right away.
Main Photography Discussion
UglyHedgehog.com - Forum
Copyright 2011-2024 Ugly Hedgehog, Inc.