« Sometimes the Pictures Are Good Anyway | Main | Fun Fact About DogDogs »

Saturday, 19 July 2014


Feed You can follow this conversation by subscribing to the comment feed for this post.

One reason for having a jpg engine in the camera is that people tend to like to see their photos in the back screen. Raw images are not amenable to be seen directly (for example, see http://upload.wikimedia.org/wikipedia/commons/d/de/Normal_and_Bayer_Filter_(120px-Colorful_spring_garden).png ). Even if you only capture in raw format, you still have to process the image to show it to the photographer - and at that point you have a jpeg engine in the camera, for better or worse.

And actually one of the issues I have had in the past with some cameras is that they use the jpeg that is automatically embedded inside the raw for preview. This jpeg tends to have less resolution than the raw file as it is just for quick preview purposes, but that brings out the issue of zooming to 1:1 for focus check. In the Canon 350D (Rebel XT?) you could not use this feature for accurate focus check, as the embedded jpeg was so small. The Fuji X-M1 has the same problem, but it has a workaround: if you shoot in both raw+large jpeg, the preview skips the embedded jpeg in the raw file and uses the large jpeg instead.

It would be another matter if you want to build a film-like digital camera that does not show previews of the captured photograph; then you could drop the jpeg engine.

". . . it's always seemed a bit odd to me that higher-end cameras don't just ditch JPEG processing altogether."

Then you wouldn't be able to view images on the camera's screen, given that RAW files are just data, not images, and cannot be seen as images until they are converted into image files, such as JPEGs or some other image format.

It would be like the film days--you'd have no idea what, or if, you captured until you got back home and processed it.

Pretty sure that the camera has to generate a jpg in order for it to be shown on screen. Raw isn't an image file, it is a data file. I've always thought of a raw file as analogous to a latent image on film. Not a perfect comparison but useful.


while I agree with you that Raw is the way to go, I encourage you to use that Sigma Raw converter on a day's worth of photographs, then you'll have your answer. That Raw converter is the most awkward and painfully slow piece of software I have ever used.

Regarding the comments suggesting a camera needs JPEG files for display purposes - this isn't so. And to say that RAW files are 'just data' and JPEG files are an 'image format' is a completely false dichotomy. Both file types are just data, and both formats are an image format if processed correctly! In both cases, for a camera (or any computer) to represent the the data as an image, a certain amount of processing from one sequence of bytes to another sequence of bytes has to occur. JPEG is compressed and RAW (usually) stores individual red/green/blue values, whereas raster displays (e.g. LCDs) invariably require an uncompressed bitmap format.

There is a practical implementation difference, however, in that JPEG being a long-time standardised format allows firmware developers to reuse well-optimised code for image display. Usually, this will be enough of an incentive for the camera to require a JPEG of some sort, along with the fact that the longer read times for larger RAW files and the potentially processor intensive de-mosaicing algorithms may hamper playback performance.

Mike, you ask, “Why not assume anyone using a DP2 Quattro will also be using a RAW converter?”

Maybe because Sigma knows how bad Sigma Photo Pro software is said to be?

The proprietary JPEG engines in ALL of the higher end cameras are actually something of a headache because Nikon, Canon, Fuji, Olympus, et. al. all seem to think that the little knobs they give you in the UI to tweak their hardware JPEG engines are some kind of "added value" ... so only the crappy Nikon, Canon, Fuji, Olympus, etc desktop software is allowed to read those settings and match them when you do the full on RAW conversion later.

Given that the first-party conversion software is almost universally crappy, no rational human would actually want to use it for anything. But, the good software (LR, Photoshop, etc) can't match the in-camera engine. The only rational move then is to never use the settings for anything. I set up every new camera to a fairly neutral state that attempts to match the default conversion settings for Lightroom, but then I never touch them again. Ever.

I'm not really sure who uses the in-camera knobs. People who use mostly straight JPEG I suppose. There must be a few.

Why the hell are they comparing the DP 1 merrill to the DP2 Quattro? Complete apples and oranges. Different lenses etc.

So they overexposed the shot, thus pushing the red beyond the gamut of srgb and they are blaming the camera?
Or rather, they overexposed for the jpeg but the raw was fine and they think that is a problem?

My guess is that they are using tungsten lighting for their still life setup, although I can't find any mention of the lightning setup. If the red is problematic on many cameras, that points to consistent user error in the lighting setup, probably with too red of light, and expecting color balance in either the JPEG engine or the raw developer to fix it,

Complaining about the srgb colorspace jpeg output of a camera like this is like complaining that 4x5 transparencies wont fit in a slide projector.

There is nothing "necessary" about jpeg, people , myself included, were making and displaying full color digital images for about ten years before JPEG became widely used.

I personally like jpeg images as part of my standard workflow. I look through the in-camera jpgs to decide which pictures deserve further attention. I suppose I could change my workflow, but it seems very convenient to me.

I assume there is also a demand for in-camera jpeg processing among professionals who want something they can send out quickly. This must be especially true of news photographers, where anything more than the in-camera jpeg raises questions of photographic manipulation. Since news photography is an important market for high-end cameras, that makes a good in-camera jpeg pipeline commercially vital.

What I meant to say but inexplicably didn't was that since it needs to produce a jpg to view anway, it might as well make it look decent. I'm pretty sure photographers have the ability to shoot the smallest jpgalong with raw. I tend to ignore jpegs in general but they are useful for quick posting and sharing.

That processed RAW files are OK is very important information that I did not notice when reading the IR materials and I thank you for that clarification. You are correct that virtually everyone buying this camera will be using RAW formats, so the OOC out-of-gamut issue is a minor annoyance at worst. And, no electrons are killed to make the JPEG - electrons have a half-life that exceeds the probable life span of the universe.

But why does IR still review camera using their JPEG output??

Adding a JPG engine does not cost the companies much in terms of money or time. It is a matter of adding an existing small programme to the firm ware. But it works as a major selling point.
It is not every photographer who wants to shoot RAW and only RAW. Many, especially professionals(!) do not use RAW.
Many amateurs buy high end cameras not for the high picture quality or the RAW capability, but for the pride of possession of a technologically advanced device. They do not care for RAW.
In such a situation no manufacturer would dare to cut that feature out of their camera.
Ranjit Grover India

What I find confusing about the IR results is that Lloyd Chambers (who so far has had a less than enthusiastic reaction to the Quattro based on his testing) despite being a strong advocate of the Merrill series has repeatedly stated that the jpgs from the Quattro are the best he's ever seen. Lloyd is not one to be effusive.

I believe that the universe is constructed in such a way that there can be multiple incompatible truths. Though Physics experiments on for example light and virtual particles bear this out the easiest way to experience this universal phenomenon is to read lots of camera reviews and fora comments on the web.

The RAW file processed with Photo Pro 6 looks a lot better though.

Nikon's new D810 test samples do exhibit the same behaviour at 1600 ISO.
Very disappointing, it looks like a firmware update for the 800 successor is badly needed.

Why have an in-camera JPEG engine at all in this camera?
I guess it's because the raw files have to be processed almost exclusively with Sigma Photo Pro, which is cursed by many. A few other alternative choices remain (Iridient for Mac people, essentially), but the "big ones" (Capture One, Lightroom/Camera Raw...) are sadly out of the equation.

I first thought that the raw image is completely unacceptable, until I check the original link. In a twist of irony, the resizing of the image to be inline among the text seems to destroy the details.

Nevertheless, the raw results of this camera are very impressive, too bad the usability seems to be horrendous.

I am not impressed with the IR review. Why not use a DP2M if you want to be realistic about the comparative process? I would recommend readers visit Lloyd Chambers' site or other sites for comparison with a DP2M.

To de-obfuscate my earlier comment, I think there is something screwy about IR 's lighting setup, perhaps unfiltered tungstun, that is causing blown out reds in a lot of the tests.

Doesn't the sigma raw processing software have a batch mode to convert a few hundred files all at once? That would make its clumsiness moot (in both senses) would it not? Just convert them into 32 bit tiff files an move on from there with tone mapping and other corrections.

The sigma camera clearly can record a wider gamut than the rest of the software stack, and in optimizing for high quality in most cases there are edge cases where the software fails.

Interesting reading about gamut here

Sigma could fix this by applying aggressive tone mapping , but that would operate up a whole other set of issues.

Just to clear up the whole "raw plus jpeg" issue. If you shoot "raw" without the jpeg, a high res jpeg is still embedded in the raw. Not just with this cameras but with almost all others - fuji's being the exception, to use a lower res jpeg in some of their raws.

Google 'Instant JPEG from RAW'. Yep, it's freeware. Win or Mac. Once installed, it's just a single right click on the folder with your raws and viola! It'll extract all of those lovely jpegs in a fraction of a second. Less write time during original capture. No need to fire up the raw converter for proofing.

I think it is too early to give a verdict about the DP2 Quattro. I have this hunch that there are bugs in the Photo Pro software (version 6.04 is no available) that may have an impact on the dp2 quattro files...

The key to look at the new DP1/2/3Q is that it is a different film. Ignore it is an upgrade. Pointless to compare even. I buy DP1/2/3M as a frozen V ISO 50 slide for future use. For the latest V ISO 100 slide, I wait for it to settle down before I got it. Film (with a camera come for free).

For the thing discussed about Jpeg+Raw, at least one feature was turned off if you using the Jpeg+Raw in M3P, the audio recording memo to the photo. There is some other odds if you use that Jpeg+Raw option I recalled and hence I am not using Jpeg+Raw in my DP1/2/3M just in case. (However, if you do not choose standard profile, it is not available later in Sigma software. Audio or Standard profile Jpeg it seems.)

In fact other than battery and wait 7 seconds for the review issue (but you can get more batteries and you do have the auto preview if you do not +/- 0.7), I do not find Sigma slow. It is quite ok. Buffer is a bit like my Nikon 7100. Limited but within limit it is fine. Can take photo in the street (and in fact I deliberately take a few road repairing road in Canada even, ;-P).

Quite interested to Q1/2/3M especially its in-and-outs e.g. what will be turned off if you use the Jpeg+Raw option.

However, I would wait. Get used to it in using Sigma in any case. Housetrained. :-)

Yep, there is a different look between published examples of the Quattro and the Merrill cameras. "Different film types" is the way to think of the two cameras. I have the three DP Merrills and don't plan to get a Quattro unless the update to the wide angle lens is significantly better than the Merrill equivalent. For me, Bayer and Merrill are two good film choices, I don't know if I want to fiddle with a third yet. I am more likely to play with Iridient Developer RAW converter vs. SPP 5.5.

The comments to this entry are closed.



Blog powered by Typepad
Member since 06/2007