Fred Picker's "Maximum Printable Negative' Implications

[Tom Montgomery]

But at what point do you quit talking about "process" and start making images?

Exactly.  Fred Picker often responded to technical questions with a big rubber stamp that said "TRY IT!". 

[RFPhotography]

Just a comment. In Lightroom you don´t have a RAW Histogram. It is a histogram in ProPhotoRGB color space. You need a tool like Rawnalize to see the RAW histogram

Since the developer of Rawnalyze has died and no new updates have been made in a few years, is there another tool that would be similar?

[Alan Klein]

Can anyone point me to two final pictures that show the difference between taking the shot with the heads on exposure vs. ETTR where both were post processed to try to get the best result.  Preferably some landscape shot, not a test chart.  I'm new to this stuff and am curious to see what everyone's talking about in the result.

[Schewe]

Try this one:
ETTR

[Alan Klein]

Jeff:  That was an excellent presentation of ETTR.   Your explanation made it easy for me to understand.  Move as much of the data to the light side without clipping so the sensors can gather more of the info in the shadow areas so the final presentation is cleaner.  I have four questions, the first you answered to a degree, but I'd like to flesh it out:

1. What size end product is required to see the noise difference? Certainly you can see it at 300%.  But when does ETTR become worth the trouble?  Have you done prints at different sizes and compared at "normal" viewing distance?  

2.  Why are we able to live with shadow areas in pictures that don't have the room on the right to move too?  If nobody cares or notices in some of the pictures, why get excited about only some of them?

3. Since you had to brighten the "normal" exposed picture in post processing, doesn't that mean that the exposure wasn't set at the correct point when you exposed the middle picture?  If you set it higher to begin with, then the shadow areas would not have had to be processed so bright in the first place so the noise would have been less afterwards.

4.  Did you make a 300% comparison of the light areas to see if there was any difference there caused by the move to the right?  Your presentation doesn't go into that.  The right most histogram shows more clipping before and after the adjustments than the others, I believe.

I ask these questions, because I 'm not a pro and doubt (or hope) I don't need to do this.  I'm having enough trouble just with Elements that I would prefer not having to feel I'm doing something wrong just because others are doing it.  Thanks for the education.  Alan.

[Schewe]

1 Depends...the better the image, the bigger you can go. I chose 300% because you could clearly see the differences.

2 If the image already has a full range such as with a full lit scene, the contrast range will disguise a lot of noise in the shadows.

3. Don't know...I set the camera to expose a normal and plus/minus 1.33. After looking at the image in Camera Raw, I decided to increase the brightness. I matched the overall range with the other exposures. Could the "normal" have been better exposed? Sure. But that's life...you want further info, do it yourself and get back to us...

4. The lighter side of each exposure was just fine after normalizing. Nothing was wrong with the +1.33 exposure that tone mapping didn't resolve. The biggest difference in the results with in the midtones and shadows.

[NikoJorj]

Not trying to be picky here, but in what sense is a  'genuine' raw histogram useful?

Well, I'd say it's because you

'capture the data' rather than 'get the shot'.

ETTR only means two constraints :
- not clipping the data at one end (even if Jeff's very didactic page shows there definitely may be some leeway here, there are cases where there ain't any due to hue shifts caused by recovery tools),
- and doing the best not to bury the data in noise at the other end of the histogram (and underquantization, though experiments like this one let me think we got some fair amount of leeway in this respect).

Worrying about a defined space and context means rather dealing with a rendering of the data, and :

Worry about the tones when you prepare a file for printing.

[Alan Klein]

I'll accept that ETTR works.  But I can't seem to get a practical response as where it will help in the real world because doing it just adds another layer of tweaking and I want to know I'm not wasting my time.  If anyone can provide a practical guidepost, that would be helpful.  Something like, if you're shooting with a 12meg camera and printing on any paper over 8x10", then you'll be able to see the difference from two feet  It doesn't have to be exact, just some guideline.  Otherwise I might as well continue to bracket my landscape shots as I have been doing for years not for ETTR but for my mistakes in correctly calculating the exposure in the first place.  And if I get to the point when one of those shots would get me better end results had I used ETTR to begin with, I'll have the +1 shot to work with.  Anything wrong with that approach.

[ErikKaffehr]

Sorry,

I don't think it can be said that way. The idea with ETTR is to keep noise down. Todays DSLRs are pretty clean. Noise is mostly related to photon statistics. Low exposure -> "bad statistics" ->much noise. The noise is mostly present in the shadows. On the other hand noise in shadows may be not very visible. I would say:

- If you have clean shadows you don't need to work on your exposure
- For a high contrast scene ETTR will get very similar to exposing for highlights

ETTR is only applicable to "raw capture", it doesn't really add complexity.

On the other hand, I find DR on todays DSLRs quite impressive. I tried HDR several times, but essentially found that exposing for the highlight (which is essentially is ETTR) still gives very good shadow detail.

Best regards
Erik

I'll accept that ETTR works.  But I can't seem to get a practical response as where it will help in the real world because doing it just adds another layer of tweaking and I want to know I'm not wasting my time.  If anyone can provide a practical guidepost, that would be helpful.  Something like, if you're shooting with a 12meg camera and printing on any paper over 8x10", then you'll be able to see the difference from two feet  It doesn't have to be exact, just some guideline.  Otherwise I might as well continue to bracket my landscape shots as I have been doing for years not for ETTR but for my mistakes in correctly calculating the exposure in the first place.  And if I get to the point when one of those shots would get me better end results had I used ETTR to begin with, I'll have the +1 shot to work with.  Anything wrong with that approach.

[ErikKaffehr]

Hi,

Do an experiment, if you bracket for say three exposures, select all and apply "auto". Can you see any difference between the exposures? Now which is the best one?

The reason I never bracket is that it gives me three times the work when postprocessing the pictures. What I normally try to do is to take a picture and check the histogram to see if it's all right. If the histogram is less than optimal I adjust exposure and reshoot.

Not saying that there is one and only optimal workflow.

Best regards
Erik

Otherwise I might as well continue to bracket my landscape shots as I have been doing for years not for ETTR but for my mistakes in correctly calculating the exposure in the first place.  And if I get to the point when one of those shots would get me better end results had I used ETTR to begin with, I'll have the +1 shot to work with.  Anything wrong with that approach.

[Schewe]

I'll accept that ETTR works.  But I can't seem to get a practical response as where it will help in the real world because doing it just adds another layer of tweaking and I want to know I'm not wasting my time.

Well, ya know, only YOU can tell if you are wasting YOUR time.

I spent the time to show proof that ETTR can be helpful. That was all I was trying to do. Now it's up to you to decide whether or not it's relevant to your work...I mean really, how much spoon feeding do you need? Go out and try it for yourself.

[Alan Klein]

Erik:  The reason I'm thinking about braketing still is because I'm going to shoot film as well and also to play around with HDR programs to see what they can do in lieu of grad HD filters.

Jeff:  Thanks for your analysis of ETTR. 

[ErikKaffehr]

Hi,

Lightroom and ACR have a very decent "gradient filter", I would play with that before resorting to HDR.

I have also found that luminosity masking in Photoshop is quite useful, from time to time. See this article:

http://echophoto.dnsalias.net/ekr/index.php/photoarticles/46-fixing-sky-with-luminosity-mask

I have done some experiments with HDR, but have not yet found some real benefits. That said I'd say that Photoshop CS Merge To HDR with Local Adaption is quite OK.

Non HDR (from -2 exposure)
http://echophoto.smugmug.com/Special-methods/HDR/HDR/13306153_DcZHj#1002864735_dkeci-A-LB

HDR (using -2, 0 and +2) exposures
http://echophoto.smugmug.com/Special-methods/HDR/HDR/13306153_DcZHj#966794997_wt4h6-A-LB

This folder contains my published HDR samples: http://echophoto.smugmug.com/Special-methods/HDR/HDR

Best regards
Erik

[Alan Klein]

Erik.  That seems relatively normal.  The shadows are darker relative to the sky  but heck they are shadows.  The overall picture looks "normal".  That's the kind of thing I'm after not the graphic, cartoony look you usually see in HDR.  Just something you might see in real life some time.  Tks  Alan.

[fdisilvestro]

Since the developer of Rawnalyze has died and no new updates have been made in a few years, is there another tool that would be similar?

There is another possibility to display a RAW histogram (If you know others please share) that involves two tools: dcraw + Histogrammar

dcraw lets you to convert an unprocessed RAW file into a tiff file, with the -D option

The command I use is: dcraw -v -D -4 -T filename.ext,  (where the options mean: verbose, document mode, linear 16 bit, tiff)

Histogrammar is a tool developed by Guillermo Luijk, a contributor on this forum (Gracias Guillermo por tus aportes en este foro y en tu página web),

There is an option to show the RAW histogram provided the tiff file has been processed with the -D option in dcraw

dcraw has been updated recently to support new cameras (as the Nikon D7000). You can find recent compiled versions here

The attached file shows the raw histogram for the same image in Histogrammar and Rawnalize. The scales of the histogram are different, but you can play with the parameters to get a very detailed graph.

[JimAscher]

There is another possibility to display a RAW histogram (If you know others please share)...

My portable Sanho HyperDrive Album will show me a RAW histogram -- and in the field too.  However, Sanho has recently discontinued its manufacture and sales, as the LCD displays for it are no longer available from the Sanho supplier.  Bummer.