« Another Print Review: The Nokia's For Real | Main | Nikon D800 Samples Online »

Wednesday, 14 March 2012



Your argument draws a parallel between throwing out half of the Raw data with the tiny amount used for a white balance correction? Okey-dokey.

Btw, did someone here actually argue that absolutely no information must be sacrificed, even assuming that any information could be actually lost if one keeps the the Raw digital negative?

You are wielding the Argument of Extreme Logical Consequences like a bloody six pound mace. As justice requires for such crimes, you must now shoot in vivid jpg with all results double-processed in Photomatix at maximum contrast settings for one full year. And no primes, either.

I don't think the idea is that you should never discard data on the way to the finished product (be that a print or a JPEG for the web or whatever). I think the idea is that you don't want to discard data from your resources. When I adjust the white balance and the tone curve and so on, I'm losing data (vs. my raw file), but if I change my mind later it's easy enough to go back to the RAW file and recover the data I discarded. The data is gone from the product, but it's not gone forever.

In contrast with this, if I shoot JPEG, the data that I could've gotten by shooting RAW is gone forever. It's not just missing from a particular finished product: it's something I can never get back.

That's a very different complaint, and I think it's a more charitable reading of the ‘but you're throwing out data line’. And it's a complaint that most of your remarks above don't address.

Mind you, I'm not saying it's a good complaint. There are lots of cases in photograph (and in life) where irreparably discarding data is unavoidable or desirable. But that's a different issue from the issue of when it's appropriate to only present some of the data one has, which is what you seem to be talking about above.

Bravo on the balanced post. I'll never shoot JPG myself but I am not against it for the reasons you mentioned. However, as my comments on Ken's post indicate, I just don't think his rationales, especially one on processing power, are quite correct.

It all comes down to convenience and whether it's good enough. For me and my workflow, JPG actually will not save me any time so whether it's good enough is not even an issue. For my wife, it's much less convenience and JPG is definitely good enough, so she shoots JPG.


How does one determine the native white balance of a sensor? Is this information normally provided by the manufacturer?

Do you refer back to days of professionals using film, slide in particular, in order to hold this up as a defining goal of good professional photography, or because the digital workflow is better than offloading the responsibility for processing?

Of those who let others develop their film, how many would you say were artists and how many photojournalists?

There is a maxim that a good photographer must, to some degree, also be a good editor. I think that remains the case whether you're selecting from a desk-load of contact prints or from some Lightroom-a-like.

You may not be wrong that the process of making RAW data look interesting as a JPEG is lossy, whether it happens in-camera or afterwards; the flip-side of the argument is that it is painful to regret *not* having got as much data at a scene as you could have done - whether that be the number of shots overall or the way they were taken individually. Then again, I suspect *revisiting* doesn't feature highly in some people's workflow.

Your distinction between data and information is interesting and useful, but it does not alter the fact that with raw editing, one can regain whatever data have been "lost" in the process, because it isn't really lost, but merely set aside. This is not true of in-camera jpegs, which do discard data permanently, and thus allow for a narrower latitude in post-processing adjustments. I can accept that the camera's jpeg rendition may be perfectly acceptable, outstanding even, but why take a chance on that when raw editing is so simple and can be reduced to a single click, if one so chooses?

Ctien: I was under the impression that LR is "non-destructive" in that it does not destroy any "data" that was delivered from the camera (raw or JPEG). Unlike PS data destruction algorithms (using certain functions) LR adjustments are "commands". Additionally I was under the impression that if one uses a "non-destructive" workflow in PS (i.e. using adjustment layers) that the original data (raw or JPED or TIFF) is not changed. I understand the idea of a lossless algorithm (e.g. certain compression algorithms) but was unaware that this occurred in image processing software (e.g. LR or PS).

Secondly, what is the content of the second image (in the water)? I am intrigued. Is it a school of sea otters?

thanks and regards!

Bravo Ctein for the civility and genteel disagreement! I shoot in RAW for many reasons, not the least of which is a ENJOY post processing! For some I imagine it is a labor for some. I liked darkroom work back in the day as much as shooting, so it stands to reason I should feel this. Not wrong, just different.

The notion that there is a difference between 'non destructive' and 'destructive' editing is nonsense. Just create a copy of the file in question, and edit away to your hearts content - your original data/information is thus perfectly preserved and undisturbed.

I always use jpeg with my Olympus m4/3 camera and I'm very happy with the results. I adjust the camera settings and all is well. I used to use raw but gave it up years ago when cameras got better. jpeg saves me much time and storage space and gives me excellent images.

I'm glad some people are happy with raw; it's just not for me anymore.

The thing is, to me, dealing with a bunch of jpegs isn't really any easier than importing raws into Lightroom. Heck, one could always come up with a raw import setting and never adjust the picture once imported into the raw converter.

To me, digital cameras have a relatively short life cycle, so I don't want to rely on the camera to make processing decisions for me, if I'm likely to change cameras every couple years, and/or I use several cameras at once. Using a raw converter allows me to keep results relatively consistent across different cameras and use the same "film" with all cameras, rather than changing "films" every time I change cameras, as with jpeg.

Well, the RAW format abuse* fits perfectly into "more control" mantra. We need more buttons, we need customizable buttons, we want more information in VF, we want faster camera, we want 128 AF points, we want 6400 segment color metering, we want a perfect camera, we want a perfect lens, we want all data ... But this is not photography.

*By abuse I mean that probably less than 10% of good pictures in the world really benefit from using this format. People tend to focus on possibilities because this is the only thing one can buy. Everything else is a gift from God and hard work.


I agree 100 percent. but


Dear folks,

Forgive the lack of individual replies; I knew when I brought up the subject of information vs. data it would be difficult to explain clearly. Truth is I've put off writing a whole column about this for more than a year because I haven't figured out how to say it clearly enough.

First point I want to make is that there are, in fact, a lot of photographers who do believe that throwing out data is an inherent vice. They are not talking about preserving options or reusability; honest to God, they really do believe more data is inherently better. I read their letters and comments all the time. I know some of you think I set up a straw man there to knock it down, but trust me, I didn't. Sad but true, there are an amazing number of these folks.

Now that said, I'm not arguing extreme consequences. It doesn't matter if you're destroying 5% of the data or 50% of the data; so long as you're killing data that DOESN'T MATTER TO THE PHOTOGRAPH YOU CARE ABOUT (note emphasis, please!). Furthermore, it is always necessary to destroy data to improve the information content of the photograph, and sometimes the amount you destroy isn't at all obvious. By way of yet another example, judicious (and appropriate) use of noise reduction and deconvolution sharpening methods makes a substantial fraction of photographs look significantly better. The amount of data those two techniques clobber can indeed exceed 50%. The thing is, it's data you didn't care about or didn't even want to know about (like the changes in the number of photons from pixel to pixel due to random statistical fluctuations).

My point stands-- anyone who says X is bad because it throws out data is kinda missing the point of photography.

pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 

It seems that many arguments start with the premise that to shoot JPEG "leaves something behind." I don't buy that. RAW needs to be processed. RAW doesn't display and it doesn't print...it's not an image at all until it's processed.

So the question becomes: how best to process the RAW file. Should we leave it to the processor designed by the chip or camera manufacturer or should we invest in 3rd party software and either leave it to them or attempt to do a better job through out own manipulation. This is surprisingly similar in my mind to the debunked concept that how hard we worked to get the image matters to anyone except ourselves. It doesn't. Nor does how hard we processed mean anything to others. "It's in the pudding."

I'm not likely to go back to last years' 5 star images to see if I can do better with the newest software upgrade, or to see if I can make the 4s's into 5's etc. As long as I'm able to gets out and shoot that's what I'll do.

Frankly the engines built into most cameras are just so good that arguments for RAW ring hollow, at least for me.

Dear Michael,

Those are seals schooling or clumping or whatever it is that seals do for fun, off the Santa Cruz pier.

Things like Lightroom and adjustment layers do not destroy your original data, but as Derek neatly pointed out, you can achieve exactly the same degree of preservation just by making a duplicate of the original file and leaving it untouched. The real reason for using those tools is that they preserve many of your workflow changes in a form where you can go back and mess with them later if you want to. But, and this is a really important but, the rendering that you look at from all of those non-destructive changes is no different than if you applied them destructively to the original file: there is data loss from the original unmanipulated file to the finished photograph that you are looking at.

I hope that makes this clear. It is a subtle matter.

pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 

Dear Tim,

Only a very small percentage of slide film photographers processed their own film, and it tended to be either hobbyist-dilettantes or nutters like me. Such people do exist; I even knew a photographer who processed his own Kodachrome. For real! But the vast majority of well known “art” photographers who use slide film (in that you can include the great color landscape photographers, a group with which I have some familiarity) did not bother with the processing of it. It had nothing to do with how lofty their artistic goals were, it was how they preferred to work.

Among those who did process their film, only a small fraction of those engaged in process variants that produced results any different from what a proper commercial lab would've given them.

pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 

It seems to me that there has been a shift in the focus of the creative process of photography in moving from "chemical" to digital. With slides and film, the process of capturing the image was far more important because of the limited opportunity for editing or manipulation at the print/viewing stage (whatever form that was). For slides shown by projection, there was no opportunity to edit or manipulate short of retouching the slides themselves, something left only to very serious photographers indeed. Moreover, apart perhaps from bracketing, only a few images in a short period time were possible (studio photography excepted). Hence, most of the creative process was directed towards creating the best image at the point of capture.

With digital photography, the process of capturing the image has become less important. If in doubt, I can rattle off literally hundreds of frames with different variations in composition, exposure, etc. in less than two minutes (try that with film, even with a bulk loader!). The creation of the image then moves to the software editing and the manipulation processes chosen for the particular viewing medium (generally, screen or print).

Shooting RAW is merely a means of trying to maximise the opportunities available at the editing and manipulation stage of the digital process. Because digital photography makes the editing and manipulation process so easy, and because the focus of the creative process is on the production of the image in the editing and manipulation process (rather than its capture), there is naturally a tendency towards using RAW rather than JPEG.

That doesn't make RAW or JPEG better or worse, merely a creative choice for photographers with the knowledge to make it and abilities to benefit from it.

But for photographers without that knowledge or ability, the choice makes far less difference. If all that is required is a quick snapshot from a digicam that will be printed directly from the card at the local camera store, JPEG makes far more sense than RAW.

The photograph either works or it doesn't. Everything else is posturing.

Heh, heh, heh! Lovely article, Ctein.



Photography, I would argue, is just as much about throwing data away as it is recording it. A 50mm lens throws away huge amounts of data vs. a 25mm lens (let's not even talk about a 85mm!), but you don't see people making the same argument that they do against JPEGs, and suggesting we all use fish-eyes. But exclusion, yes, even throwing away, that's what we do. Throwing away stuff to emphasize other stuff. In framing, composition and processing. Each phase of the photographic process is stripping away another layer of that which is (if we do our jobs well) unnecessary or extraneous.

I can't afford the printer HCB used, but Olympus has a pretty good one, living right in the camera. (I know, it's processing and not printing, but you get my point!)

I recently tried going back to jpeg on occasion, but found it really wasn't worth it for me. it came down to the trade off between time saved and information lost. the number of "keepers" I get is a pretty low % of my overall shots, and I dont mind investing some time in processing the good images to get the best result. and I found that even with jpeg I was doing some minor adjustments, so it wasn't just "set it and forget it". in the end I'd rather have the flexibility of the raw file. a good photograph is worth a few minutes of my time in post.

Photography has always been a process of throwing information away at every step and many photographers would expend a lot of effort and money in an effort to retain as much information as they could in the earlier steps in order to have more to throw away later.

For what it's worth back around 1980 the scene at Duggal Color Projects every evening would be a whole bunch of photographers agonizing over their clip tests deciding whether to push or pull their E6 processing.

It was always "Rush the odd bag a half stop pull, hold the balance, and don't mount them till I get here" then run off to have dinner and hope they looked good so you could clean the studio before midnight and go see a band.

"You're all tired of hearing it over and over again, but it's true.  The negative is the score the print is the performance."

That depends a lot on the type of photography you are talking about. Ansel was making “pretty pictures"; there are plenty of powerful photographers whose prints are not as important as the subject matter they are showing. In those cases, the idea is the score and performance at the same time. (I'm not advocating that you discount craft and technical skill. I'm just saying that, sometimes, those things are secondary and not terribly important.)

Tanaka's text was thought provoking, I enjoyed reading it, I would like to read more thought provoking texts from him, but what response do /you/ expect when provoking?

I shot JPG the first 2.5 years shooting digital. Now I only shoot RAW. Why? Because I really enjoy working with the RAW file, especially with Lightroom. Second reason is, frankly, I'm afraid to go back to JPG.

Isn't think issue being a tad overanalysed? I didn't realise it was so hard. If I'm taking photos of some old cups and saucers to sell on eBay then use a jpeg. If I think the photo I'm about to take will hang on a wall or get published in a good book then I'll shoot raw.

We choose what 'information' we wish to exclude from the moment we pick up a camera. Why pick up the camera now, instead of 5 seconds ago? That information from 5 seconds ago now isn't recoverable!

I have a practical rather than dogmatic approach to raw files.

I shoot raw + jpeg on my D90, and jpeg only on my GF1.

Most of the time the jpegs are close enough - I edit them and upload to flickr and make prints for small books.

But sometimes I need to use the raw file - to make quick colour corrections, or rescue shadow/highlights.
Also I return to raw when I'm making larger one off prints.

I found that editing GF1 raw files was a zero sum game:
- Editing was slow - using the bundled software - and often resulted in files of lower quality than those processed from the camera.
- The raw files are quite "thin" with little headroom - very little extra information in highlights for example.
I was filling my hard drives with large files that were slow to edit and weren't much better than those from the camera. So after a while I stopped shooting raw on the GF1.

On the other hand the D90's raw files have "more in them" so for me its definitely worth keeping them.

I always shoot Raw/RAW/raw (however you choose to type it). There are several notions about data equaling visual information (in particular where the greatest volume of information is, brightest vs darkest pixels). In the end we all reduce the amount of data in producing a print or an image for the web whether you do it in camera using the default settings or manually in Lightroom/Photoshop/etc. It is called editing and editing involves removal. That is the definition of editing. For me (everyone else is free to choose otherwise) having 12 or 14 bits of data to work with gives me more flexibility to work on an image than starting with only 8. It's that simple.

I guess by the time this is posted many other people might have made the same point. Having a master copy of a jpeg, and producing multiple versions depending on intended use is nothing more or less than the road to workflow madness.

The difference between Raw and JPG has nothing to do with quality. Raw is just a lot easier for people with anything more than the simplest of workflows. Programs like LIghtroom even make it a cinch to produce multiple versions of the same image repeatedly without touching the original data or creating, storing and finding multiple copies. For example when different print sizes or formats are called for.

In addition of course, even if you keep a master copy of your out-of-camera jpeg, that image already has had decisions made about it that cannot be corrected.

And in addition to addition, I've never really used out of camera jpegs, but wouldn't you need to have at least one or two different camera settings to cater for different images? And then you run into even more workflow complications... shit! I forgot and shot the whole damn wedding at warp factor 9

I'm sorry, but Ctein is missing a critical point with his white balance example. The white balanced photo he presents is an interpretation of the original data, not necessarily a loss of the data. With a raw file the data is still there to be reinterpreted if you change your mind. With a JPEG file the original data is gone forever, you are stuck with the interpretation.

Imagine Ansel Adams for a moment. He re-evaluated and re-thought his prints over and over. If he were to throw away his original negatives and work from a print of the negative, such reinterpretations would be immediately restricted by the actual loss of data that occurred.

There certainly is a place for JPEG in the world. Just as there's a place for MP3 and MPEG video coming down your cable pipe. But I think it is exceedingly incorrect to regard these things as optimal data sets for anything other than what they're used for: lowest common denominator distribution.

The simple question everyone has to ask is whether they mind living in a compressed or data reduced world or not. For many things, sure, not a problem. For other things it becomes a real problem.

Like Chuck Norris taking on Bruce Lee, this is a debate I no longer take part but here goes. Any betting man knows that putting the odds in your favour is good strategy.

Bruce Lee
1958 Crown Colony Cha-Cha Championship

Chuck Norris
Six consecutive years as the middleweight world karate champion.

Black Belt Magazines fighter of the year 1969. Triple Crown winner for most contests won 1969

Put the odds in your favour, choose Chuck-I mean RAW.

I say that as a big Bruce Lee Fan

Assuming I understand the arguments correctly, wouldn't it be fair to say that it boils down to two statements: 1) Know what the end use of photograph is likely to be. 2) Choose your tools accordingly.

It also seems to me that the dividing line follows the same logic that divides shooters from darkroom workers (in the film era).

That's just my impression, but I'd love to hear further comments from people who have worked in both the film/darkroom and digital/photoshop formats.

Cartier-Bresson lied. He cared a lot about the print.
Cartier-Bresson was different from his public image. He was his own spin doctor, and a very clever one that is.

Certainly nothing wrong with a good JPG. It just takes more work in the field. My main reason for shooting RAW is invariably I find I or the camera didn't get the white balance right. I don't like being bothered to even think about settings like that when I shoot, so I leave it until later. If I have a ton of images to post process I get one right as far as white balance for the lighting they were taken in and copy this to them all in LR. Saves time and most images come out good. Then I change the ones that this didn't work for.

That's one of the liberating things about color digital. No more need to muck with color meters and color correction filters. JPG shooting brings back some of that work, albeit in a different manner.

Dear Thom,

No, I really didn't miss any point. You failed to get mine.

Please reread what I actually wrote, in the article and comments. I am not arguing RAW vs JPEG., I am discrediting some very specific (mistaken) arguments put forth.

There really are people believing that loss of data is a bad thing. They're not talking about fungibility or reusability; they think less data is inherently worse.

pax / Ctein

Well let's be specific.


Data represents a state of nature that is compromised by uncertainty and errors (noise). In photography some noise sources can be minimized and others can not.

The state of nature we want to record is the real, but unknown, amplitude for the visible light that reached each sensor while the shutter was open. Because of noise the best we can do is estimate those amplitudes. So the information in a raw file is an organized matrix of amplitude estimates for each sensor site.

However the data is information without context. The information in the data has potential, but by itself the data is useless to humans. This means data is not knowledge. Before it is rendered, raw data can not impart knowledge. We don't transfer data to the viewer, we transfer knowledge.

Information Content

If a small amount of light hits the sensor does the raw file contain less or more information than a raw file where the optimum amount of light hits the sensor?

Both have the same amount of information (assuming both were measure competently). Both are a matrix of amplitude estimates that differ only in signal-to-noise ratio. Noise has information content. It's just not information of interest to humans. But noise is an estimate for a state of nature. So far modeling noise is not part of photography. But modeling noise is very important in other areas such as diagnosing disease from MRI images. It's obvious the image with the optimum amount of light is superior for communicating with viewers. This is how knowledge become important.

If you delete amplitudes estimates from the matrix does the information content decrease? Yes, it does. However the discarded information may or may not be useful to humans. So the decrease in information content may impact our ability to communicate with the viewer or it may have no impact at all.


A photograph communicates knowledge to the viewer.

Our task is to progress from an estimate for the state of nature to a state of knowledge. A rendered image imparts a state of knowledge humans can contemplate. The knowledge content of the images rendered from the two raw file discussed earlier is very different. The knowledge content from an image with discarded data may or may not be different.

Applying our prior experience to the data is how we create a state of knowledge for the viewer. No image is viewed without relying upon the prior experience of the person who rendered the image from the original data. By prior experience I mean everything someone knows about how to communicate to the viewer with maximum efficiency and impact.

Information with context creates a state of knowledge becuse our prior experience gives the data meaning. Sometimes viewers need to know as much as possible about what we saw or felt when they view the rendered image. Other times this is unnecessary.

In-camera jpeg rendering relies heavily upon the prior experience of the programmers who designed the compression algorithms. Their prior experience may or may not be appropriate for the data in a given raw file. The programmers never analyzed the data you recorded. They have no idea if you operated your camera properly. And they certainly know nothing about your aesthetic goals. The programmers made the best bet they could, but it's just a bet the compression will retain all the data necessary to transfer the most amount of information to the viewer.

Discarding data after it's collected increases risk in order to achieve convenience. Different people accept different levels of risk.

There is no free ride. You can be extremely thoughtful and careful before the data is recorded by using the optimum parameters (exposure, WB, etc) and enjoy the convenience of jpeg compression with confidence. Or you can do the best you can at recording the data when time is limited (a different kind of convenience) and tolerate the inconveniences associated with out-of-camera raw rendering.

A completely different compromise is to make convenience the highest priority before and after recording the data and accept the risk that some of your images will suffer. Today's cameras are extremely competent at recording and rendering so this compromise makes sense to many photographers.

Nice reading..Thank-you.

How do I find out the native white balance of my sensor?

It is what I choose to present regardless.


"You're all tired of hearing it over and over again, but it's true. The negative is the score the print is the performance."

True enough-but Ansel was talking about classical music and classical landscape photography.

Ask Charlie Parker about his music or Robert Frank about his pictures
and it might be another story.

The comments to this entry are closed.