Best capture technhique and processing tools ?

[doug]

In his posting, Marv (Marpel) did not merely say that combining ETTR with Raw capture was the BEST way to shoot. He said (and I quote) "ETTR, as discussed by others, is really ONLY applicable to Raw." (my bold-face emphasis). The idea expressed here (and very frequently by many in the DPReview forums) seems to be that once the camera software makes choices in rendering a JPEG and throws away data not conforming to that "baked in" result, you are stuck with that result and further adjustments will be futile. Instead of leaving the reader with the impression that somewhat more extensive adjustments MAY be possible from the Raw data and achieved more easily, the impression is left that if you don't shoot Raw, you are sort of stuck with the out-of-camera result.

A couple years ago, Den helped tutor me through PWP adjustments to a JPEG that had white balance and other issues. I thought that the changes were rather dramatic. But perhaps that was just a figment of our imagination (lol).

Heck, the BIRD photo used in the PWP tutorial is a JPEG. So even Jonathan and Kiril seem to think that JPEGs are sufficiently adjustable to showcase their software. Am I wrong to believe that ETTR has some benefit toward capturing the maximum possible data (without blowing highlights, of course), even with a JPEG?

I just re-read this posting, and it might sound like I'm trying to start a flame war with Marpel. I'm really not. So I apologize if anyone takes it that way -- especially Marv.

Doug

[Marpel]

Doug,

No offence taken with your comments and I don't take them as any attempt to initiate a flame war. One of the benefits of a forum such as this is the opportunity to not only exchange ideas but to learn other's methods and perspectives.

Perhaps I misspoke a bit and didn't ensure my comments were meant as my opinion rather than a fact above challenge and reproach. For that I apologize if I inadvertently mislead.

Having said that, I don't believe I stated that one could not edit a jpeg, rather my comments about jpegs were related to ETTR. In fact, I agree that a jpeg is not exempt from manipulation. Editing is possible and tone/colour etc can be changed, albeit somewhat less than a TIFF or Raw before getting into trouble. Although I now only shoot Raw, I have done post processing on jpegs in the past and I know others do the same. You just don't have the elbow room that a Raw allows.

Regarding ETTR and jpeg - it does not seem logical (to me) if the main purpose of ETTR is to get the best from an image file, to use a format that automatically retricts that quality right off the bat and greatly restricts the value of ETTR. As you probably know, a jpeg (8 bit, sRGB) has 256 levels of tone, while shooting Raw one can get 4096 levels (at 12 bits). The brightest f-stop holds a potential 2048, the next 1024 and on it goes by half for each f-stop. It's this large quantity of levels in the bright end of the file that make ETTR of value. Since the introduction of ETTR, all conversations I have witnessed have been in relation to Raw and I would have expected if it was of much value to jpegs, that it would have had some play time as well. I am certainly no expert (not even close) so perhaps Den or others could comment on the real value of ETTR for jpeg. I just don't see its applicability to jpeg (I may be proven wrong however!!).

Marv

[Dieter Mayr]

ETTR is usefull for scenes with a quiet low range of tones, like a fogy landscape as a extreme example.
If the scene itself has the full rage of tones already, lets say a sunny landscape with bright lights and deep shadows, the histogram bumps already to the left and right so there is no room for shifting it.
The only difference between 8 and 16 bit images is the numbers of steps between white and black, the sensor pixel can be without electrons for totally black or almost overflow with electrons for white, so the dynamic range of the sensor is a constant for that particular sensor and has no relation with the bit depth of the image. So if there is a image with a histogram with space to the left and right, one can move it to the left or right without loosing quality even when it's a 8 bit image.
The advantage of 16bit comes out when it comes to spreading the histogram, the result gets smoother because of the more steps in between 2 given values.
JPEG has a lossy compression, so it looses a bit of data / image quality, 100% agree on that.
But, if one prints out mostly images of a size 13x18cm (5x7 inches) or full DIN A4 pages, like I guess most of the non professionalists do here, I doubt one can tell from just looking at the picture if it was from JPEG or RAW, even at 50x70cm printouts I did not see a difference.
I take advantage of RAW in cases where it can point with it's feaures, like in delicate light situations, when mayor color or contrast editing has to be made and such things. For a "normal" picture that does not need any special afforts in exposure I use JPEG with the highest quality.
And if a little color or exposure correction is needed it can be done with a JPEG based image too.

The above does not take any claim to be the totally truth, its my understanding of how things work, mixed with a few years of expierience, so if i am wrong in some cases, my apologies and please correct me.

[tomczak]

Could people post some examples of the ETTR-exposed and then software-corrected vs. somewhat more 'conservatively'-exposed images, showing the actual benefit of the technique in controlling the noise? (e.g. a black cat on a pile of coal may do...).

[rfuerman]

1) @Den: thank you for your offer. Disclaimer - I am not seriously into digital photography as most of you folks on this msg board; that makes me one of Dieter's "non professionalists," I suppose. But I try to follow & learn about PWP from your (& the other's) posts, hints & comments. I know I can reach you all with questions via this board.

2) @bbodine9: how do *you* capture images? RAW, jpeg, scanner ??

3) @Dieter: Re-reading the "Caveats" section of the LL article; I would say that the author agrees with you.

4) [OT] @everyone: click on the 'User Control Panel' link at the top of the page, please; click the 'Profile' tab & enter your location in the box #7. Thank you. I just think it is interesting to see where all you PWP users are located.

boB

[couman]

There is also the question of what you are trying to communicate with the image. Using Dieter’s example of a foggy landscape, you may completely destroy the mood of the scene by either expanding to full dynamic range or shifting to the right. On the other hand, if you’re trying document a particular feature of the scene, then it’s probably best to attempt to capture as much information as possible that is relevant to that feature.

I think Den has said it many times, but it’s worth repeating – treat each situation based on its own merits. Before the shutter release, decide what you like and dislike about the scene, and then do whatever is necessary to maximize the positives and minimize the negatives. The latter includes both controlling the camera and exposure conditions and consideration of what might be needed in post processing in order to achieve your objective.

[Marpel]

All good comments.

The Luminous Landscape site (luminous-landscape.com) has a recent article about ETTR that you may find of interest (brief reference in Bob's post). It is under "What's New" and is titled "Optimizing Exposure" (scroll down the page a bit). I would put a link directly to the article, but I don't know how to do that. It is relatively brief but holds all the salient points, including the relationship between bit depth, noise and ETTR. The author also states in a caveat, that "this approach only applies to raw files - not in-camera jpegs" but he doesn't provide any supporting reasons.

For those who are really interested in the ETTR topic, there is also a couple forum threads on the same site (Discussion Forums > About this site - both at least 5 pages in length), generated as a result of the article. Although there is a bit of an argument over why/how ETTR benefits an image, all seem to agree that it does have some benefit. I believe one of the threads has some images which show noise comparisons related to ETTR/non ETTR. I read through both threads and learned a lot but as is common with these conversations on technical issues (a number of the Forum participants are well-known leaders in all things digital/photography and talk technical quite often), some of the stuff was way over my head!!!

Marv

[doug]

First ... Thanks, Marv, for taking my earlier post as intended.

I stumbled upon the Luminous Landscape article about ETTR two or three years ago and found it to be the first thing about shooting digital (other than sharpening) that was totally at odds with the principles that I learned about shooting film. Based on that article I began to expose that way whenever the scene left enough headroom to do so without blowing highlights.

I am now shooting Raw plus JPEG. But, for many situations (i.e. images not intended for framed enlargements or magazine covers), the JPEG is satisfactory. But certainly not directly out-of-camera since EVERY image benefits from some sharpening (I don't have sharpening turned up in my JPEG settings to avoid the risk of oversharpening.) Because I am ALWAYS going to be doing some editing (even if just sharpening), it didn't seem to be a big problem to have to also "bring down" the excessive exposure in an image exposed per ETTR principles.

I never before noticed that the author of the Luminous Landscape article specified Raw. I just simplistically reasoned that if the benefit of ETTR is to maximize the amount of data captured, it must be beneficial to a JPEG as well as Raw. Perhaps the benefit is not as great as with Raw. But how, I reasoned, could capturing more data not be better than capturing less?

But then again, I don't begin to understand the affect (effect?) of different bit depths. So perhaps that's it. Perhaps, there is something about JPEG bit depth that says "trying to capture more data is a fools errand." "Something about bit depth makes it impossible to more fully utilize the bits available in the top of the dynamic range."

As you said previously, perhaps Den or someone else with special expertise will weigh in.

Doug

[den]

Ref: "Camera Exposure Metering Methods" http://www.ncplus.net/~birchbay/tutoria ... ure/01.htm

My understanding of the LL ETTR references as applicable to PWP5's RAW Dialog...

(1) For "JPEG only", use Method 1...

The camera's exposure meter preset to "Spot" with "+2EV Compensation" on a scene's whitest spot is required because non-linear expansion [gamma] is being applied to the linearly captured RAW data by the in-camera processer which is not user accessible... ...This will potentially shift the scene's whitest spot closer to the 255 tone without clipping than using the camera's AutoExposure... ...thereby increasing the signal to noise ratio of the captured scene which is the objective of ETTR regardless of recorded 8 or 16 bit depths... ...of course this assumes that the scene's dynamic range is less than the camera sensor's.

(2) For "RAW", use Method 2 and...

Open the resulting RAW file in PWP's RAW Dialog and before making other changes... Select the Gray tab and: (a) Magnify the histogram to its highest resolution; (b) set Dyn. Range to "No Change"; (c) move the Gamma slider left to 1.00; (d) move the Exposure slider right until the histogram's right most vertical is barely to the left of the right most grid line [100%] tone; and (d) return the Gamma slider to its original setting [usually 2.20]...

Proceed with other RAW Dialog preference edits [perhaps at least change Dyn. Range to "Full Range" or "Medium"] and/or select OK to use other PWP transforms and tools...

[doug]

Thanks for weighing in, Den.
I interpret your comment as saying that the benefit of ETTR with a JPEG capture may not be as much as with Raw, but there may be a small benefit as indicated in the comment: "it will potentially shift the scene's whitest spot closer to the 255 tone without clipping ...... thereby increasing the signal to noise ratio of the captured scene."

Doug