RAW or JPG

Antonio Correia · August 8, 2006

Suddenly I ask myself:
Is is worth, for me, to shoot RAW all the time ?
Which of these pictures was shot on RAW on camera and transformed to JPG?
And which was born JPG ?
Guess ...

A

B

thebigsky · August 8, 2006

Did you do any PP on the RAW image or just convert it straight to JPG?

Antonio Correia · August 8, 2006

thebigsky wrote:

Did you do any PP on the RAW image or just convert it straight to JPG?

You were fast !

I photoshop both files.

Nikolai · August 8, 2006

Antonio Correia wrote:

Suddenly I ask myself:
Is is worth, for me, to shoot RAW all the time ?

I think only you can answer that question...

As to me, I answered it long time ago: RAW only, unless we're talking some ultrafast event (bikeracing being such an example), where you may want to sacrifice the end quality for the number of shots taken during a very short amount of time..
HTH

Antonio Correia · August 8, 2006

Nikolai wrote:

I think only you can answer that question...

As to me, I answered it long time ago: RAW only, unless we're talking some ultrafast event (bikeracing being such an example), where you may want to sacrifice the end quality for the number of shots taken during a very short amount of time..
HTH

Sorry Nikolai but I am not quite following you, I think.
You mean:
Shoot only raw when we want a great quality.
If we are shooting kite-surf shoot jpg. Less quality but fast saves in the CF.
I re-read and I think this is your point.

But here, and that is if I am interpreting correctly, lies my point:
For the post in the net and prints not larger than 20*30 cms (7.9*11,8 inches) do I have to go throught larger files, slower computer and work flow, duplicating the files, space on disk, etc....

Does it worth ?

wxwax · August 8, 2006

It's not important for us to be able to tell the difference.

It's important for you to know whether the additional flexibility and control offered by shooting in RAW will help you or not. If it doesn't, then you don't need RAW. But if you find yourself wishing you could correct a shot and can't, then maybe RAW would be handy for you.

Only you know. What we think doesn't matter.

Nikolai · August 8, 2006

Antonio,

Antonio Correia wrote:

But here, and that is if I am interpreting correctly, lies my point:
For the post in the net and prints not larger than 20*30 cms (7.9*11,8 inches) do I have to go throught larger files, slower computer and work flow, duplicating the files, space on disk, etc....

Does it worth ?

Once again, my friend, this is the question only you can answer for yourself...
Before dSLR I was pretty happy with JPEGs. I simply always copied them twice into two different drives, never-ever touched ones I dubbed "originals".
Now with DSLR my RAWs are my originals, so I only copy them once (backup is a different issue:-) and post-process in place. Since I *must* post-process them before I can give them to somebody or to upload, it forces me to look at them and often cull out the baddies, thus improving general average quality of the crop. For me, I find the benefits of the RAW workflow outweighing their bulkiness and the necessity to always process them.
I must note that I have a rather fast "gaming-grade" machine and enough of the HDD space. Your mileage may wary...

Nikolai · August 8, 2006

Succinctly put!

wxwax wrote:

It's not important for us to be able to tell the difference.

It's important for you to know whether the additional flexibility and control offered by shooting in RAW will help you or not. If it doesn't, then you don't need RAW. But if you find yourself wishing you could correct a shot and can't, then maybe RAW would be handy for you.

Only you know. What we think doesn't matter.

What Sid said

pathfinder · August 8, 2006

Like Nik and waxy I tend to shoot RAW routinely, unless there is a compelling reason not to.

That is my choice and I don't think the RAW processing really slows me down that much more than shooting jpgs, because even my jpgs need some tweaking to satisfy me.

I was a darkroom rat when I was younger, so maybe I just can't really give up the dark room.

If you prefer your images to be processed by the local drugstore, jpgs will probably do just fine for you. Folks looking at them may not see much difference, but I CAN see a significant difference in my RAW shots and my jpgs. I rarely shoot jpgs.

If I was a journalist or a sports shooter who was churning 500 frames a day that needed to be sold immediately, I would rather shoot jpgs then. But I'm not.

spudjer · August 8, 2006

A

Nikolai wrote:

I think only you can answer that question...

As to me, I answered it long time ago: RAW only, unless we're talking some ultrafast event (bikeracing being such an example), where you may want to sacrifice the end quality for the number of shots taken during a very short amount of time..
HTH

I agree with Darth!

I always shoot RAW for a superb "original" with all the control I demand. Jpegs only if you twist my arm... ever shoot 35mm for a client because they were going to reproduce them small in a brochure and some images worked so well they wanted to do full page magazine ads or billboards and 35mm wasn't good enough for reproduction? I have. Since then, I have always shot in largest format possible... and in digital world it is RAW.

Rock and Raw Antonio.... peace

jfriend · August 9, 2006

RAW vs. JPEG is a personal choice of tools, no correct answer

Antonio Correia wrote:

Suddenly I ask myself:
Is is worth, for me, to shoot RAW all the time ?
Which of these pictures was shot on RAW on camera and transformed to JPG?
And which was born JPG ?
Guess ...

A B

First of all, I've chastised a few folks on dpreview for making RAW vs. JPEG a religious argument with only one correct answer so I want to make sure we don't do that here. RAW is a tool that is available to you. You have to weigh the usefulness of that tool and use it if it gives you a better result for the cost/work that you have to put in. It's no different than deciding whether to buy a separate noise reduction program vs. the built-in noise reduction in CS2. Evaluate the tools available and decide which one works for you within your cost boundaries. If the answer for you is JPEG instead of RAW, then that's the best choice for you and I am fine with that.

For me, I know that there are shooting conditions where I can get better results from RAW (messed up or widely varying white balance, wide dynamic range, accidental exposure mess up, intense colors with potential channel clipping problems, high or low contrast scenes). Further, I've gotten down the learning curve with Bridge/ACR that I spend less time in PP with RAW than I would with JPEG (largely because of the advantages of multiple image changes and non-destructive editing) and I mostly shoot events of 200-300 photos. The only downside I really feel with RAW is that it takes more storage on the card and on my hard disk. To me, that's a tradeoff worth paying for. I just buy bigger storage.

thebigsky · August 9, 2006

How about taking this approach, when you shoot JPEG your camera applies some PP to the image, depending on the model it's usally a bit of sharpening, contrast, saturation etc.

You could shoot all your images in RAW and use your computer to batch convert your images to JPEG applying the above settings as chosen by you.

Now you have your JPEG images but you also have the security of a RAW file should you ever need it.

Having changed to RAW I can't imagine going back, and I regret deeply that some of my early images weren't shot in RAW as my exposures were invariably a little out and I could now recover those mistakes, plus I don't like some of the PP that was applied by the camera but now I'm stuck with it.

The downside of RAW is that every image needs PP and when new to this you find youself fiddling endlessly with every permutation, paranoid that you're not getting the best result. However you soon develop your own workflow and this processing time speeds up.

Charlie

Stustaff · August 9, 2006

A

Now that I do like thebigsky, never got it that straight in my head!
How do I apply a setting to all my RAW's that ouputs a sharpened etc JPEG?

I have CS2 on a Mac

because if I can do that and then use the RAW for the 2 or 3 BEST images or any that need a major fix but have most of them in JPEG it would be great.

Any suggestions?

thebigsky · August 9, 2006

Stustaff wrote:

Now that I do like thebigsky, never got it that straight in my head!
How do I apply a setting to all my RAW's that ouputs a sharpened etc JPEG?

I have CS2 on a Mac

because if I can do that and then use the RAW for the 2 or 3 BEST images or any that need a major fix but have most of them in JPEG it would be great.

Any suggestions?

I should point out I don't actually employ this method, I shoot all my images in RAW now and import them into Lightroom.

I found this link for batch converting in CS2 but there are others probably more qualified to advise. By the way many cameras will save both a JPEG and RAW file for you when shooting, though you're going to need a fast and large CF card if you choose this route.

Charlie

P.S. Have you tried Lightroom, it's currently a free beta download, I really like it.

Stustaff · August 9, 2006

A

Thanks,

problem with jpeg in camera is the camera processing is then done! if I create a jpeg from a Raw I can then alter WB etc so hopefully get a better JPEG!

Lightroom - read a bit about it but at the moment my Main PC is a PC! and my mac is just an ibook so not really up to running lightroom.

I'll take a look at that link cheers, I seem to remember something about doing Batch conversions through Bridge!?

thebigsky · August 9, 2006

Lightroom is now available for the PC.

dancorder · August 9, 2006

A

My $0.02:

I'm planning on shooting RAW from now on for three reasons:
1) Bibble Pro. It's a RAW converter but it also lets me do the most common edits (curves, cropping, sharpening, noise reduction) directly. It's actually quicker for me to do my edits in Bibble than in the GIMP (no PS for me at the moment) unless I want to do something special.
2) Exposure latitude. I'm not a particularly experienced photographer yet so the ability to recover some shadow or highlight detail that would be lost in a JPG is useful to me
3) I have lots of storage and I might want to blow up a picture really big in the future. Then again I might not, but I don't feel I'm losing anything by shooting RAW.

mercphoto · August 9, 2006

First of all, EVERYONE shoots RAW. The only question is which converter do you use? For some they use the converter in the camera. This means the camera spits out a JPG. And you can usually tweak various conversion parameters in the camera to affect the final JPG. But once its converted there is no going back. Others either do not like the in-camera results, or want to be able to choose various conversion parameters on an image-by-image basis, or (let's admit it) cannot get a proper exposure and white balance to save their life, so they use a RAW converter on a computer rather than in the camera. All those are valid reasons to shoot RAW (even the last one).

When I had the original Digital Rebel I found its auto-white-balance feature to be lacking and I shot RAW more often than I wanted just to get around color casts. The 20D was nearly flawless on AWB, plus its in-camera JPG's were very, very good. I seldom shot RAW with a 20D unless the lighting situation was particularly difficult.

I now have a 1D Mark II. I find the AWB to be not quite as good as the 20D. However these days I'm shooting a lot of night racing. I'm not always happy with the flash metering so I am now shooting mostly RAW so that I can adjust exposure, white balance and shadow point as needed.

claudermilk · August 9, 2006

wxwax wrote:

It's not important for us to be able to tell the difference.

It's important for you to know whether the additional flexibility and control offered by shooting in RAW will help you or not. If it doesn't, then you don't need RAW. But if you find yourself wishing you could correct a shot and can't, then maybe RAW would be handy for you.

Only you know. What we think doesn't matter.

The most accurate, to-the-point answer so far.

Me, I now go both ways.

Usually I shoot RAW as I prefer the latitude for much of my shooting as it't in demanding lighting conditions & the control-freak in me prefers to reserve final decisions. However, when doing quick shooting in daylight, and when the camera gets passed to someone else for when I cannot be the loose nut behind the viewfinder, JPG is easier. It just depends on the needs of the situation; sometimes shooting RAW makes a huge difference, sometimes not at all.

Nikolai · August 9, 2006

One way for me...

claudermilk wrote:

Me, I now go both ways. Usually I shoot RAW as I prefer the latitude for much of my shooting as it't in demanding lighting conditions & the control-freak in me prefers to reserve final decisions. However, when doing quick shooting in daylight, and when the camera gets passed to someone else for when I cannot be the loose nut behind the viewfinder, JPG is easier. It just depends on the needs of the situation; sometimes shooting RAW makes a huge difference, sometimes not at all.

While I totally agree that sometimes (and maybe even often) there is not much difference between out-of-the-camera jpeg and a quickly processed raw file, I - personally - find it easier to have just one (RAW-based) workflow rather than two different ones, one for RAWs and one for JPEGs...

Just MHO and my personal preference, of course

:

rutt · August 9, 2006

You are getting a lot of great advice here, so normally I wouldn't weigh in. But I've been rethinking this very issue and exactly what raw is good for and what out of camera jpegs are good for, so here goes.

First, what exactly is a raw image? It's an exact recording of the state of the camera's sensor at the moment the shutter activated. No post processing at all is applied. It just gives a level for each pixel on the sensor. A number of camera settings have no impact at all on raw images. Whate balance is applied as a post processing step by the camera, so this setting doesn't effect the raw image (well, it gets recorded in the raw image, but that's it.) Similarly for contrast and saturation. Also ISO. ISO only effects the meter which determines shutter speed and aperture and also how the image is processed. But shoot in manual mode and ignore the meter and ISO doesn't matter to your raws.

What's wrong with raw?

It's big. Even the finest quality jpegs (essentially lossless) are quite a bit smaller, maybe 1/4 the size. Big translates to slow when you want to shoot a lot of frames quickly (as at a sporting event.) The buffer fills up faster and then it has to empty to the card and it's generally a drag. Also when it's time to preview and edit, more has to be read from the camera and/or card and that can be awkward.
It's not post processed. Look at a raw image in ACR and turn off all the automatic stuff, exposure, shadows, contrast &etc. Choose a flat curve. No sharpening. Choose a random color balance. The image will be flat. 0 pop. If that's what came out of the camera, people who didn't know better would complain. A lot.

So when the camera makes jpegs, it does some things that are good for 90% of the images that people actually shoot. It applies an S curve to open up the midtones at the expense of the highlights and shadows. Typically this curve is designed to make some part of the image a light point and some part a dark point. The camera makes this choice based on some algorithm. That means that it will plug a shadow and blow a highlight. That's what you want if it chooses the right highlight to blow and the right shadow to plug. Not what you want if it chooses to blow or plug something you care about. Similarly for sharpening and noise reduction.

Now in most DSLRs and all professional DSLRs, it's possible to set things up so no sharpening is applied to jpegs and very little contrast or saturation added. Jpegs like this offer almost as much flexibility as raw images if they are correctly exposed. ISO is still not ignored when generating the jpegs which means some detail is lost in the shadows and/or highlights.

So? Well, the truth will set you free. How much post processing are you planning on? Shooting raw really doesn't add much time if you were already planning on some manual post processing and it can open up quite a few options. Manual sharpening in particular can almost always improve on what the camera does. And once the an image has been sharpened, it's basically impossible to resharpen effectively. Color balance is really no big deal. It's very easy to fix with curves once you get the hang of it. Contrast and exposure can really be a big deal. But on balance, if you can turn off in camera sharpening, contrast and saturation enhancement, in camera jpegs can work as a starting point for post processing much of the time. Butt I just hate it those few times the camera lost something in making the jpeg that I wanted. One good shot ruined out of 50; how much would you pay for a camera that did better?

I like to shoot RAW + medium quality jpeg. It's still slower than just jpeg, but the medium quality jpegs don't add much to total size and don't slow things down much. They are very useful for previews and proof sheets and are ready to upload to smugmug.

Sorry for the long post. As I said, I've been thinking about this.

Nikolai · August 9, 2006

Great post, John!
I only didn't quite get this one point of yours:

rutt wrote:

ISO only effects the meter which determines shutter speed and aperture and also how the image is processed. But shoot in manual mode and ignore the meter and ISO doesn't matter to your raws.

I mean, I understand what you're saying, but it seems to me that ISO (i.e. light sensitivity) setting is important enough parameter to think about.
I mean, even if you keep the shutter speed and f/stop the same in full manual mode, changing the ISO *will change* the content of the RAW file.

Am I missing something?

Thanks!

rutt · August 9, 2006

It's taken me a while to get this, but the ISO setting doesn't actually change the state of the sensor at the time of exposure. With a given shutter and aperture setting, the exact same number of photons strike each pixel on the sensor no matter how the ISO is set. The raw image just records this value, essentially the number of photon strikes. It makes a lot of sense to use ISO to interpret this afterwards, whether in the camera of in the raw converter, so that happens. But if you are willing to change the raw converter settings, you can get the same image back no matter how the ISO is set. Prove it to yourself.

Nikolai wrote:

I mean, even if you keep the shutter speed and f/stop the same in full manual mode, changing the ISO *will change* the content of the RAW file.

Am I missing something?

Thanks!

Nikolai · August 9, 2006

John,

rutt wrote:

It's taken me a while to get this, but the ISO setting doesn't actually change the state of the sensor at the time of exposure. With a given shutter and aperture setting, the exact same number of photons strike each pixel on the sensor no matter how the ISO is set. The raw image just records this value, essentially the number of photon strikes. It makes a lot of sense to use ISO to interpret this afterwards, whether in the camera of in the raw converter, so that happens. But if you are willing to change the raw converter settings, you can get the same image back no matter how the ISO is set. Prove it to yourself.

It sounds really strange to me, so I am going to check it out tonight.
I'm gonna set the camera on a tripod, lock shutter and f/stop in manual and then take a series of shots changing only the ISO. Then I'm going to bring the results into ACR, uncheck all the auto stuff, set curve to liear and see if I can arrive to identical end-result images from ALL of them by manipulating only with the exposure slider (which supposed to be a master one, correct).

If I don't lose any highlights or shadows - you win:): , but, frankly, I highly doubt this will happen

Does this plan make sense in "proving the point", or you were talking about something different?

Thanks!

rutt · August 9, 2006

That's just the experiment you should do. To be honest I haven't done it, so I am interested in your results. My thoughts on this are based on what I learned in my Computational Photography course.

Really, though, the camera can change the physics of the sensor. The same number of photons are going to strike each pixel, not matter what the ISO setting. It can only change the interpretation of it. It may be that precision is lost at lower ISO settings. I'm not sure what kind of numbers are used to represent the data. (Fixed, floating, how many bits.) Low light images are essentially noisy because less light means fewer photons strike each sensor and it becomes error becomes statistically significant.

Nikolai wrote:

It sounds really strange to me, so I am going to check it out tonight.
I'm gonna set the camera on a tripod, lock shutter and f/stop in manual and then take a series of shots changing only the ISO. Then I'm going to bring the results into ACR, uncheck all the auto stuff, set curve to liear and see if I can arrive to identical end-result images from ALL of them by manipulating only with the exposure slider (which supposed to be a master one, correct).

If I don't lose any highlights or shadows - you win:): , but, frankly, I highly doubt this will happen

Does this plan make sense in "proving the point", or you were talking about something different?

Thanks!

Nikolai · August 9, 2006

John,

rutt wrote:

That's just the experiment you should do. To be honest I haven't done it, so I am interested in your results. My thoughts on this are based on what I learned in my Computational Photography course.

Really, though, the camera can change the physics of the sensor. The same number of photons are going to strike each pixel, not matter what the ISO setting. It can only change the interpretation of it. It may be that precision is lost at lower ISO settings. I'm not sure what kind of numbers are used to represent the data. (Fixed, floating, how many bits.) Low light images are essentially noisy because less light means fewer photons strike each sensor and it becomes error becomes statistically significant.

I'm glad to hear we're on the same page about this test applicability.

Let me just tell you now, in advance, what I'm thinking would happen.
I think that ISO level actually affects how sensor records the incoming light. Roughly speaking, if, say, at ISO 100 it would take 100 photons to increase the pixel's value one step, at ISO 200 it should only take 50 of them (the actual numbers of photons are unimportant for this discussion, only their relative values are).
That been said, I speculate that if I "zero in" shutter and f/stop at ISO 400 and then go in both directions, thus getting 100-200-400-800-1600 (and 3200, just for the sake of it), I should end up losing some shadows at 100 and some highlight at 1600.
There is one point in your favor:-) in this particular test, though: going both ways from 400 only asks for 2 full f/stops adjustments. And we all know that RAW can easily handle that amount of correction without necessarily breaking a sweat.
So, to make it representative, I'm also going to perform two other series.
In the first I start with ISO 100 and go all the way up to 3200.
In the second I'll start with ISO 3200 and go down to 100.
This way we'll get not 2, but 5 full stops in each direction, which, to my knwoledge, ACR cannot handle as gracefully as 2.
I predict we'll lose highlights when going from 100 to 3200 and we'll lose shadows when going backwards.
We shall see, won't we?

rutt · August 9, 2006

Yeah, Nik, we're on exactly the same page here. Real true raw would actually record the real true number of photon hits and I'd win. But perhaps the representation isn't good enough for this. Some sort of floating point representation would work great here, I'd think because it would allow exact description of shadows with just a few pixel hits and adapt well to higher light situations where the differences are expressed as much greater numbers of pixel hits. Essentially, this is how good high dynamic range representations work. Of course, you don't want floating point exactly because you don't need negative numbers or non integers. But that exponent is just what you want.

Antonio Correia · August 9, 2006

I want to thank all for the participation in this discussion.
It is very interesting and I hope it will continue with the experiences of the two wellknown photographers present in this thread.

I must learn to work with the window of raw conversion in CS2.
Now, I look but I don't see.

mercphoto · August 9, 2006

rutt wrote:

It's taken me a while to get this, but the ISO setting doesn't actually change the state of the sensor at the time of exposure. With a given shutter and aperture setting, the exact same number of photons strike each pixel on the sensor no matter how the ISO is set. The raw image just records this value, essentially the number of photon strikes.

I agree up until that last sentence. The ISO is basically a gain on the sense amplifier coming off the sensor. ISO 100 is a 1:1 gain. ISO 200 is 2:1 gain. Etc. So the RAW file really will change with a different ISO setting.

Nikolai · August 9, 2006

Bill,

mercphoto wrote:

So the RAW file really will change with a different ISO setting.

That's what I think, too.
We'll get experimental confirmation in a few hours

Nikolai · August 9, 2006

Test results
Well, I didn't even have to go to +/- 5 f-stops

ISO 400 +/- 2 stops shows enough difference:

If Rutt's theory about "ideal raw" were correct, adjusting ISO in camera and then adjusting exposure setting "in reverse" in post (ACR in my case) should have brought the histogram (and the image itself) back to the original values.
However, it does not look so. Adjusted histograms, although close, are not matching the original one, especially at the end of the spectrum.
In the images this was reflected as a strong cast.

Just in case anybody interested, I have the entire set of images I was talking about earlier:

100, 200, 400, 800, 1600 (all at 1/640s at f/10)
100, 200, 400, 800, 1600, 3200 (all at 1/30s at f/22)
100, 200, 400, 800, 1600, 3200 (all at 1/1000s at f/22)

the highlighted item denoting the anchor shot (for which the series was correctly exposed) and the rest just with the ISO changed. I was shooting in RAW + small coarse JPEG. Both (converted raw and jpeg, both scaled down to 800x600) are uploaded here: http://nik.smugmug.com/gallery/1758459

Caveat: we have not one, but TWO test subjects: camera sensor hardware and post-processing software, ACR 3.4 in this case. I'm not claiming ACR is perfect, but I kinda hope it does a good job with exposure adjustment.

Anyway, for all practical purposes, I believe this test proves that in-camera ISO settings does affect the content of the RAW file in a way that cannot be compensated via available software in post.

Quad erat demonstrandum.

PS
John, you owe me a

RAW or JPG

Which is RAW 16 votes

Comments