jpeg vs. raw

[Turtle]

Speed, convenience, small size and great results for simple images = JPEG
Flexibility, dynamic range and the ability to wring out the last drop = RAW

I don't think there is much more to it and it wil presumably always remain this way as any JPEG engine is by definition automated.

 

[Ronald M]

Done raw + jpeg on two different sessions of preschool kids. did the raws one time, jpeg another. PS is set to receive JPEGS. I made every effort to get WB correct in camera.

With no doubt, there is no difference in work with this procedure. Now if you accept what the camera provides and do not think dodge/burn can improve your pictures, use JPEG. In 50+ years, I never did pic that did not benefit from some correction. That correction in ACR is the same for JPEG or raw , so I see no benefit from JPEG. The time expended was the same in both cases.

 

[Jamie Pillers]

RAW is a lossless format; JPEG is lossy, with the manufacturer's baked-in processing algorithms. Why throw away data that you might want in future, if you want to re-interpret the image in pp?

Why? Because then I can re-experience the good old film days, when film manufacturers locked in the look and allowed me almost no wiggle room in pp! 🙂

 

[willie_901]

There seems to be some confusion about raw and jpeg in this thread. I believe each of us can choose our own workflow and I really could care less how people meet their goals. At the same time, people should make decisions using as much information as possible along combined with their experience.

When the exposure and white balance parameters are perfect much of the information content of the raw data is redundant. Printing or computer screen viewing can not use all of the information the sensor records. Except for commercial use(some clients require tiff formatted images) out of camera jpeg recorded with perfect exposure and WB settings are not technically inferior to jpegs rendered from raw data. Raw, jpeg and tiff have exactly the same number of pixels. Scaling images to increase pixel density has to create data that was never recorded in the first place. When the degree of jpeg compression is appropriate for the image detail level, an in-camera jpeg can scale as well as a jpeg rendered by third-party software.

Sometimes photographers make exposure mistakes (cameras never make exposure mistakes because a camera can't think, it has limited prior experience at best and it has no way to know what the photographer wants to achieve). The effects of exposure errors can be minimized when an experienced photographer manipulates the raw data during post processing. These manipulations are no different than processing film to push or pull exposure, affect contrast, or, dodging and burning and paper selection during analog printing. Both digital and analog post-processing can be challenging and tedious... especially when the exposure errors are large. The same is true for white balance errors. Often exposure and WB errors are small and the in-camera jpeg image can be fully optimized quickly during post processing.

Sometimes the contrast of the subject exceeds the dynamic range of the camera system. In this case optimum exposure becomes critical. Now the risk of failure by discarding the raw data in-camera becomes significant. The in-camera compression destroys important data needed to make the most of the camera's inherent signal to noise ratio and dynamic range. Recovering highlights and pulling shadow regions is complicated by hue twisting. Shadow areas loose detail. Correcting for these problems requires tedious selective processing of the raw data. But the only way to make full use of whatever dynamic range you have is to skillfully manipulate the raw image. In-camera jpeg rendering parameters are available to make use of the camera's dynamic range. The parameter guesses are often adequate and the loss of flexibility compared to post-processing the raw data is not important (unless it is).

Unfortunately it is rare to have a scene where the light has one color temperature. This means neither auto nor manual WB parameter selection can render accurate colors. The only recourse is to selectively modify the color balance to match the color temperature in different regions of the frame. When the differences are small post-processing an in-camera jpeg works well. Otherwise, the using raw data works much better.

The aesthetic implications of raw or jpeg are also important. The camera firmware renders the jpeg based on parameters that are set in the menu system by the photographer or presets using parameter values chosen by camera company employees and consultants to generate a certain aesthetic look. Some camera firmware allows a high degree of flexibility in jpeg rendering parameters and sets of customized parameters can be saved for use in different circumstances. The factory default parameters are interesting. These are decided by strangers who have no idea what is being photographed or what the light is like and how the photographer choose to expose the image. In both cases the rendering parameters are just guesses. Some guesses are very good and some are not. The problem is, when the guess is wrong your ability to improve the rendering aesthetics is limited. The lossy jpeg compression removes the data you need for unlimited flexibility to render the image. However, the better the guess for the in-camera jpeg rendering parameters, the more effective post-production image modification becomes. More often than not, minor adjustments during post processing of jpegs produce pleasing results. Some raw rendering software applies the all of the in-camera raw processing parameters. All commercial raw rendering software uses the camera's WB parameter. The initial(automatic)rendering of the raw image during post processing is only a guess too. But unlike an in-camera jpeg, there is no technical or aesthetic disadvantage to radically change the rendering parameters. This flexibility comes at a cost and for some of us the most significant cost is time.

 

[Peter_wrote:]

thanks for the informative post, willie_901.


is jpg better than raw? neither -nor. the one is a ready picture and the other just the base for one. two different pair of shoes.
so the real question is, are you able with your pp skills and the time you are willing to spend to produce a better picture than the automatism of the camera?
so it really depends on the own priorities and capabilities, what is better for someone.
looking at the ooc pictures of some cameras like the ricohs or the leica x2 i personally wouldn't even spend a minute on thinking about shooting raw, if i had them.
even with my canon, whose output i don't really like, i shoot jpg. i am mainly into analog stuff, and the reason to shoot digital is for me that i want quick results and not much work.
i sometimes forced myself to shoot raw, but in fact i was always too lazy to develop them. so i am just careful with exposure and wb while shooting, and that's it.

It [Tiff] has most of the disadvantages of JPEG (inability to adjust white balance or recover blown highlights) with none of the advantages (small file size so you don't fill your memory card too fast).

that surprises me. i always thought, the reason why it's difficult e.g. to recover blown highlights with JPG is the 8-bit restriction. so i was the opinion, that with a 16-bit Tiff file there is also much more potential to recover highlights, or doing other manipulations without banding, tonal breaks...
can anybody enlight me?

 

[Dwig]

...
that surprises me. i always thought, the reason why it's difficult e.g. to recover blown highlights with JPG is the 8-bit restriction. so i was the opinion, that with a 16-bit Tiff file there is also much more potential to recover highlights, or doing other manipulations without banding, tonal breaks...
can anybody enlight me?

Your thinking is correct. TIFF may be little better than a JPEG if it is an 8bpp TIFF where its only possible advantage is that it avoids the quality loss inherent in JPEG's compression methods.

On the other hand, if the TIFF is a 16bpp file there will be significant added adjustment ability. It still lacks the color and exposure adjustment flexibility of a RAW but it surpasses any 8bpp format.

 

[RichC]

I prefer to shoot in Raw as that gives me the option to play around with the file if I need to. I look at the Raw file as my digital negative, and avoid shooting JPGs. It is entirely a personal choice, of course, as others have said: if JPGs give you the image quality you're happy with, then why bother with Raw?

That said, some people are doing JPG a disservice. Although it doesn't have the greatly enhanced range of Raw adjustments, there's still an awful lot you can do with JPGs - granted it requires more work and skill.

I hope Alex doesn't mind, but here's what 10 minutes in Photoshop can do with his very underexposed JPG - given more time and a larger image, I could improve it far more (including removing the blue-purple colour cast, and improving the contrast caused partly by lens flare).

I've attempted to create a natural-looking image, so the highlights are deliberately high key, since Alex shot against the light.

(For reference, his JPG and Raw images are below.)

Here's a crappy example. The first shot is OOC jpeg, apologies for the crappy exposure, but it is the only one in a few bracket shots in which the sky is not completely blown. Try as you might to pp the jpg, I know I cannot get anything out of it.

Here's the same shot processed from RAW, it's not perfect (like the color cast due to underexposure) but as least it's a start...

 

[aleksanderpolo]

You certainly have better skill pping the jpeg than I have. That said, if one need to pp, I see no advantage using jpeg other than saving the dirt cheap hard drive space. 🙂

 

[BlackXList]

Well I guess 2013 is the year I stop putting off getting my head around RAW

 

[Damaso]

A debate, at this point, really? I don't think so. Those who want to shoot either format do so. I shoot RAW. If someone out there feels that jpeg only is the way forward why should I bother trying to disabuse them of that notion?

 

[Peter_wrote:]

Your thinking is correct. TIFF may be little better than a JPEG if it is an 8bpp TIFF where its only possible advantage is that it avoids the quality loss inherent in JPEG's compression methods.

On the other hand, if the TIFF is a 16bpp file there will be significant added adjustment ability. It still lacks the color and exposure adjustment flexibility of a RAW but it surpasses any 8bpp format.

thanks, dwig! so at least i wasn't completely wrong in saving my scan files as 16-bit tiff...

why does raw have still more flexibility for exposure adjustment?

 

[RichC]

why does raw have still more flexibility for exposure adjustment?

Think of Raw as a stack of possible images, from which you extract one as a single-image file such as JPG.

That's a very simplified simile, but it gives you the idea of how Raw works.

Essentially, a Raw file is a heap of data from the camera sensor that allows ranges of image information to be extracted and/or modified. Thus, the exposure information in a Raw file is not just for a single level of exposure but permits a range of exposure to be extracted (with accurate-ish mapping of tonality in part of this range).

In contrast, a JPG (or TIF) image is exactly what you see: a collection of tiny dots on a two-dimensional surface - pixels.

So, Raw files have more in common with data files such as spreadsheets or databases than with "traditional" image files such as JPG.

 

[Peter_wrote:]

but raw also just stores a concrete amount of information...

so when tiff has less flexibility, there must have been some information lost, from converting from raw to tiff. but how could that be, when you use 16bit uncompressed tiff? i somehow can "feel" ("feel" because i have too less knowledge about this stuff), that there is some colour information lost. because the bayer pattern is summed up in one pixel then...
but i still don't understand, why there is less flexibility for exposure adjustment...

 

[aleksanderpolo]

Perhaps someone with more technical knowledge can answer you, I think that tiff already have the exposure, white balance baked in so it is more limited than what you can do with RAW, here's something I found:

http://www.thephoblographer.com/2011/09/26/are-tiffs-and-raws-really-the-same-thing/

but raw also just stores a concrete amount of information...

so when tiff has less flexibility, there must have been some information lost, from converting from raw to tiff. but how could that be, when you use 16bit uncompressed tiff? i somehow can "feel" ("feel" because i have too less knowledge about this stuff), that there is some colour information lost. because the bayer pattern is summed up in one pixel then...
but i still don't understand, why there is less flexibility for exposure adjustment...

 

[kbg32]

This whole argument is so frickin' ridiculous. I have been having a battle with some guy over in Turkey over this very thing. He says - "If you can not take photographs which are not good in original, then its your technical understanding of photography, which
is poor" BALDERDASH!

See here - http://www.rangefinderforum.com/forums/showthread.php?threadid=127829

 

[raid]

Based on the posted discussions here, I will from now on use jpg fine for family photos and will use DNG+jpg fine otherwise.

 

[kbg32]

Use whatever gives you the results and expresses your vision best. End of story.

 

[kbg32]

Based on the posted discussions here, I will from now on use jpg fine for family photos and will use DNG+jpg fine otherwise.

BTW Raid, RAW is great for family as well as "otherwise"

 

[RichC]

but raw also just stores a concrete amount of information...

so when tiff has less flexibility, there must have been some information lost, from converting from raw to tiff. but how could that be, when you use 16bit uncompressed tiff? i somehow can "feel" ("feel" because i have too less knowledge about this stuff), that there is some colour information lost. because the bayer pattern is summed up in one pixel then...
but i still don't understand, why there is less flexibility for exposure adjustment...

Yes, Raw has a finite amount of information - but far more than a TIF. As I said, think of a Raw file as a stack of images but a TIF or JPG as a single image. This pile has images that vary in their settings such as colour balance and exposure. And rather than being able only to take one image from this pile, you can combine settings from anywhere in this pile, so you can can create a single photo with any combination of exposure or colour balance that an image in the pile has - this single image is your TIF. (Again, I''ll stress this is extremely oversimplified.)

So, once you create a TIF (or JPG) file from a Raw file, you've lost a lot of information - the TIF file contains just some of the data in the Raw file. Once you have a TIF file, you no longer have access to the different settings in the Raw file - the pile of images.

8 and 16 bit with regard to image files refer simply to the number of colours: 8-bit images allow 256 colours per pixel, whereas 16 bit allows 65,536 colours.

A good way of thinking about 8 bit vs 16 bit in this conversation is not as amounts of information but as colour accuracy: 16-bit images are a lot more accurate with regard to colour. That said, in many instances that accuracy isn't needed as the colour in 8-bit images is often very close to that captured by the camera anyway (e.g. because of the limited range of colours in the scene - a landscape may be mostly green) so the extra colours in a 16-bit image may not all be needed or so similar to the 8-bit ones as to be indiscernible.

You can of course change the colours in the TIF file - but this is limited. Think of modifying a TIF as akin to adjusting an old-fashioned silver print being developed with an enlarger, using dodging and burning, etc.

 

[kbg32]

A quick visual comparison/explanation -

http://www.slrlounge.com/raw-vs-jpeg-jpg-the-ultimate-visual-guide