Raw Processing v. Photoshop

Duffy Pratt

In his 5th edition of Professional Photoshop, Dan Margulis devotes a chapter to Adobe Camera Raw. He shows how Raw capture is superior to any JPG is a large class of images.

He also says that the current adjustment tools in ACR are not worth the time investment, and that he has not found any image where Photoshop could not do as well as ACR in making adjustments. He also shows that there are many images where Photoshop does better than ACR. What he means is that by making tonal or color adjustments in Camera Raw, you often are doing stuff that you could do just as easily in Photoshop, and that on some images, you are actually making subsequent corrections more difficult.

As a result, he recommends that, at least at this point, its probably better to simply shut off all of ACRs auto adjustments, and then move the unadjusted file into PS for correction. I suspect this recommendation will be controvestial. I'm wondering if anyone here agrees with it. If so, why? If not, why not? And do the other camera modules have more robust, or fully featured correction systems than ACR?

Duffy

Andy

Duffy Pratt wrote:

In his 5th edition of Professional Photoshop, Dan Margulis devotes a chapter to Adobe Camera Raw. He shows how Raw capture is superior to any JPG is a large class of images.

He also says that the current adjustment tools in ACR are not worth the time investment, and that he has not found any image where Photoshop could not do as well as ACR in making adjustments. He also shows that there are many images where Photoshop does better than ACR. What he means is that by making tonal or color adjustments in Camera Raw, you often are doing stuff that you could do just as easily in Photoshop, and that on some images, you are actually making subsequent corrections more difficult.

As a result, he recommends that, at least at this point, its probably better to simply shut off all of ACRs auto adjustments, and then move the unadjusted file into PS for correction. I suspect this recommendation will be controvestial. I'm wondering if anyone here agrees with it. If so, why? If not, why not? And do the other camera modules have more robust, or fully featured correction systems than ACR?

Duffy

Bruce Fraser, in Real World Camera Raw makes a compelling case for doing a lot of adjustments in ACR. Having learned to do it quickly and efficiently in Photoshop already, plus having all the other "stuff" that photoshop has, I use Photoshop for my exposure and fine color adjustments.

I use ACR for white balance, and exposure, possibly vignetting correction, and that's about it.

colourbox

Duffy Pratt wrote:

As a result, he recommends that, at least at this point, its probably better to simply shut off all of ACRs auto adjustments, and then move the unadjusted file into PS for correction. I suspect this recommendation will be controvestial. I'm wondering if anyone here agrees with it. If so, why? If not, why not?

What about highlight recovery and chromatic aberration correction. In ACR they are one- or two-slider operations. In Photoshop, I'm not sure it's that easy. Especially the highlight recovery, which I understand is a rather sophisticated algorithm. Does Dan go into how to do those quickly in Photoshop?

The other reason to use ACR is that if you correct in ACR, your corrections are saved in a tiny metadata file. In Photoshop, your corrections are usually a stack of space-consuming adjustment layers and masks. And then there's cropping. If you decide to go back to the original Raw and convert it again, in ACR the crop and rotation can be saved with the raw file and you can always change the crop and rotation and recover areas you cropped out. In Photoshop, once you've cropped something it's not easy to replicate that same crop and rotation precisely if you have to start over.

For setting black point, white point, and curves, Dan is right. It's just as easy to bring in an uncorrected raw and add a Curves adjustment layer. That's what I do with scans. For white balance, Dan is probably right because he is Dan, but I am not that good at it and find it much easier to do white balancing in ACR than in Photoshop.

dogwood

Duffy Pratt wrote:

And do the other camera modules have more robust, or fully featured correction systems than ACR?

My two-cents here:

ACR is fast and fairly reliable. But I'm a Canon shooter and I personally think the DPP RAW conversions are better than the ACR convertions. That's just my opinion, of course, but I find the "faithful" style in DPP to be good enough I don't need a curves adjustment in PSCS2 afterwards. But that said-- I generally use ACR in day to day convertions just because it's faster and less work and reliable enough for day to day conversions.

But again... for files where I want the best quality conversion, I'll use DPP.

BillyVerden

dogwood wrote:

My two-cents here:

ACR is fast and fairly reliable. But I'm a Canon shooter and I personally think the DPP RAW conversions are better than the ACR convertions. That's just my opinion, of course, but I find the "faithful" style in DPP to be good enough I don't need a curves adjustment in PSCS2 afterwards.
But again... for files where I want the best quality conversion, I'll use DPP.

I agree, i use DPP for WB and overall exposure. Its very fast and seems to be less destructive than ACR or any other converters i have tried. I also use "faithful" in DPP. One last thing is that DPP also saves your changes and lets you see before and after side by side comparisons as you make changes. DPP then PS thats my method.

Duffy Pratt

I just checked the manuals that came with my 20D, and there is no info on how to use DPP. Do you have any suggestions about where I can go to learn the software? I am in a sort of odd situation, in that I use Elements 3 for my filing and for much of my printing, and use PS CS for corrections. If I could make DPP fit into that clunky set-up without too much hassle, I will probably start using it.

Duffy

BillyVerden

Here's a start: DPP Tutorial

claudermilk

Yup, I definitely disagree with his recommendation. Why? Edits in a RAW converter are non-destructive as they are simply instructions on how to modify the way it converts from linera RAW data to the final RGB file. In PS, however, each edit you do is destructive. Yeah layers mitigates this, but I still think doing as much as you can as early as you can in the workflow the better. The other bonus is already mentioned: the converter instructions takes up less room than a multi-layer PSD or TIFF file.

I also don't think ACR is the best tool for the job. But that is personal preference & an unwinnable argument.

Scott_T

I think part of this recommendation might be particular to workflow preferences.

To me I get dramatically better results, using Nikon NX to do the raw conversion, saving as tiff then going into CS2 for additional editing. When I was using ACR to do raw conversions the images just suffered compared to Nikon NX...

I do not like having this 2 stop process, but the difference it too big to ignore....

Regards,
Scott

rutt

This is Chapter 16, What Comes Around, Goes Around.

1. I loved this chapter when I read it and so did the other beta readers. I got great results with flowers using the ideas from this chapter. But they can't be taken out of context. The chapter is one of a series 15,16, 17, and especially 18 dealing with representing out-of-gamut colors and contrasts.
2. I knew this would be controversial when I read it and was greatly looking forward to a prolonged flame-fest. I just hope it doesn't last as long as the stupid 8 vs 16 bit thing.
3. You have to be careful with Dan's controversial positions. He is almost never wrong, but the full statement is often more complex than the controversial summary. In this case, I think Dan is actually saying:

   Dan Margulis paraphrased by Rutt wrote:

   I, Dan Margulis, can produce a better image with the tools in Photoshop, than you, whoever you are, can produce with the tools in ACR.

   Anyone who has watched Dan work, much less competed with him, will be very hesitant about contradicting that statement.
4. But there is a broader point, one that I've been making here on dgrin for a few years. ACR is great, but it's no substitute for Photoshop. One plausible way to use it is to choose settings which result in a conversion which loses no information (or at least no information which you will eventually want.) After that, shadow/highlight, curves, blending, etc can do anything ACR could have done BECAUSE NO INFORMATION WAS LOST (my premise).
5. Even Dan agrees that color balance is a strength of ACR. I especially like it for professional lighting (theater lights) where the color temperature is known, but using the dropper on a known neutral point also rocks.
6. Hey, Duffy, why not write the chapter summary? You've got a fine flame-fest thread going here. Once you do write the summary, you can edit it into the initial post of this thread, Andy can rename it, and I'll link into the master reading group table of content.

Duffy Pratt

claudermilk wrote:

Yup, I definitely disagree with his recommendation. Why? Edits in a RAW converter are non-destructive as they are simply instructions on how to modify the way it converts from linera RAW data to the final RGB file. In PS, however, each edit you do is destructive. Yeah layers mitigates this, but I still think doing as much as you can as early as you can in the workflow the better. The other bonus is already mentioned: the converter instructions takes up less room than a multi-layer PSD or TIFF file.

Assume, for the sake of argument, that you are saving your original RAW files. If the corrections you make afterward improve the picture, why do you care if they are destructive or not? I've never understood the fetish for not bruising pixels. Anything that I need to save, at any time, I can save. And if the picture is better, I don't care in the slightest if I've hurt the data.

On the second point, if I use Digital Photo Professional, I will have to convert the RAW to a TIFF before I can open it in Photoshop anyways. So, at least with that module, the space saved by the RAW instructions then gets lost upon conversion. I don't know if the other modules act similarly. I think not.

Duffy

Duffy Pratt

I don't think that's what Dan is saying. If so, big deal. I'm pretty sure Dan could get a better image using just Camera Raw than I could get using Photoshop. Instead, I think he's saying that he can always or almost always get as good or better an image out of just PS, than he can out of Camera Raw pre-processing plus PS. So, instead of comparing Margulis to me (which is sort of like thinking of a boxing match between Muhammed Ali and Woody Allen), I think he is pitting himself v. himself, which makes for a much fairer fight.

Your broader point is the one that interests me. I've done some exploring of DPP now, and I can't see the usefulness of any of it, beyond getting as full a conversion as possible. Not only does it not have the PS flexibility, but unless I'm missing something, it almost completely lacks the metrics. The best you seem to be able to do is to set one neutral point for white balance. I guess that's fine, if there is only one neutral point.

I don't use any RAW converter now. I shoot JPGs. The reason I'm interested is because I'm just about to switch over from RAW to JPG and I am trying to get a handle on the best way to go about it. I could write a summary of the chapter, but I've still got 12 chapters to go to put it into context (for example, the amazing Moose correction went over my head in some details). But I do write the summary, it will have some natural deficiencies because I don't have Camera Raw and I'm a total noob when it comes to RAW modules.

BTW, what do you think of 16.4 D, his final correction of the cast on the couple in the schoolroom? He convinced me that the shirt and hair should be neutral. But what about the lips? Her lips look a sickly shade of magenta, which could be close if she had terrible taste in lipstick. But his lips are almost the same, and everything from his cheekbones down looks really magenta to me as well, as do the bags under her eyes.

Duffy

dogwood

BillyVerden wrote:

Here's a start: DPP Tutorial

Cool-- thanks. Hadn't seen that one and the Canon DPP pdf hasn't been a whole lot more useful (for me anyway) than just playing with the software.

rutt

I know Dan doesn't think that starting with jpegs is always as good as starting with raw because the camera applies a stupid S curve which loses information sometimes that Dan might have used. In ACR, you can use a linear curve, which Dan likes. And I know Dan thinks that simple color balance is often quick, easy, and effective in ACR.

What he doesn't like is corrections in ACR which are "junior" versions of their PS conuterparts. Curves without separate control over the individual channels. Levels-like commands as opposed to curves. Sharpening without all the control that USM and layer blending yields.

I know the words I put in Dan's mouth weren't actually right, but it sure was fun to do.

Duffy Pratt wrote:

I don't think that's what Dan is saying. If so, big deal. I'm pretty sure Dan could get a better image using just Camera Raw than I could get using Photoshop. Instead, I think he's saying that he can always or almost always get as good or better an image out of just PS, than he can out of Camera Raw pre-processing plus PS. So, instead of comparing Margulis to me (which is sort of like thinking of a boxing match between Muhammed Ali and Woody Allen), I think he is pitting himself v. himself, which makes for a much fairer fight.

Duffy Pratt

I know the words I put in Dan's mouth weren't actually right, but it sure was fun to do.

Well, you know him personally, so you have a bit more license there. My closest personal contact with him was an exchange of e-mails I had. He made a claim in one of his books that the Finger Lakes produces some of the best Reislings in the world. I went on vacation there last summer, so I asked him which ones I should try. To my surprise, he sent me a lengthy reply. Also, as it turns out his taste in wines is about on a par with his taste in color correction. Someday, I will take one of his classes.

I got the distinct impression that he was a bit more accepting of the white balance sliders in ACR than he was of any of the other tools. I can easily see why setting master endpoints or master curves is likely to be a bad idea. And I have no idea why people sharpen early in the RAW process. But the ability to change the white balance seems good to me. I just don't understand the lack of any metrics (at least in DPP). If I could get a readout of the LAB values, then I would have a much better idea what I was doing with the sliders and color wheel they give. Right now, it looks like you have to eyeball it, and I don't think thats a very good way to fix casts.

Duffy

jfriend

How I think of ACR vs. PS

rutt wrote:

I know Dan doesn't think that starting with jpegs is always as good as starting with raw because the camera applies a stupid S curve which loses information sometimes that Dan might have used. In ACR, you can use a linear curve, which Dan likes. And I know Dan thinks that simple color balance is often quick, easy, and effective in ACR.

What he doesn't like is corrections in ACR which are "junior" versions of their PS conuterparts. Curves without separate control over the individual channels. Levels-like commands as opposed to curves. Sharpening without all the control that USM and layer blending yields.

I know the words I put in Dan's mouth weren't actually right, but it sure was fun to do.

The way I think about ACR and Photoshop is this:

1) If I have a ton of images to do and I'm not trying to make them perfect (e.g. 800 soccer season photos), I'm going to use the mass processing tools in ACR to make them as good as I can in ACR. I might hand tweak a few in PS just because they intrigue me, but I'm really valuing the workflow efficiency of working in the RAW tools on large numbers of images.

2) If I have a single image and my goal is to make it the best it can possibly be, then I want to use the RAW tool to give me the best starting point for PS and make the most advantage of the RAW data, but then as soon as I've gotten a good starting point, I hand it off to PS where the tools are just richer and give you more control.

To that end, I make sure I've got the channels centered where I need them - most dynamic range I can get, nothing clippped I might want, approximate white balance to get the channels centered appropriately, any highlight recovery that I need, no ACR curves applied, etc.... I then apply any of the tools in ACR that are just better or easier in ACR than they are in PS like vignetting or CA fixes. Then, I take the still raw, unfinished image into PS. It's now got the best starting point I can give it for the more advanced tools in PS. There I'll use curves, advanced sharpening, more advanced, non-global color correction, etc...

claudermilk

Duffy Pratt wrote:

Assume, for the sake of argument, that you are saving your original RAW files. If the corrections you make afterward improve the picture, why do you care if they are destructive or not? I've never understood the fetish for not bruising pixels. Anything that I need to save, at any time, I can save. And if the picture is better, I don't care in the slightest if I've hurt the data.

On the second point, if I use Digital Photo Professional, I will have to convert the RAW to a TIFF before I can open it in Photoshop anyways. So, at least with that module, the space saved by the RAW instructions then gets lost upon conversion. I don't know if the other modules act similarly. I think not.

Duffy

I do in fact save my RAWs. What if I decide to go back later and process the image again using an improved tool, a newly-learned technique, or a totally different look? I have done all of the above. In retaining the RAW, I started at the same point as before without a reduced data set. That's my reasoning; if it works for someone else great, if not it sure won't hurt my feelings. Horses for courses.

My converter of choice is Bibble, which required that I save to TIFF before opening PS, though there is a plugin version that takes over for ACR; two reasons I don't use it: a) buggy performance in the recent past, and b) reduced batch processing.

I already stated that I don't care for ACR--from the subsequent discussion it sounds like the interpretation of the original statement is that you are better off using the more robust tools in PS rather than the reduced-capability ones in ACR. I'll go with that qualified statement; as a blanket statement as originally presented, I disagreed.

monkeydeus

I don't think it's just opinion: it's a fact that canon keeps it's color tags proprietary. This means DPP is the only app that has access to the true and full color information of your picture. Anything else, is just an approximation, although in some cases a very good one. IMHO, bibble blows ACR out of the water, and once you look at the same pic in DPP, it's almost impossible to use ACR for anything other than scratch conversions. I've gotten weird artifacting and banding out of ACR, and thought several of my lenses were junk until I started using DPP.

To Duffy's comments about not caring about destructibility: if you prefer a jpeg or tiff it's your prerogative, but RAW gives you the ultimate flexibility to go back and start all over later if you want to: why would anyone make an argument against that? You can say it's not for you, but you can't say that's a bad thing.

dogwood wrote:

My two-cents here:

ACR is fast and fairly reliable. But I'm a Canon shooter and I personally think the DPP RAW conversions are better than the ACR convertions. That's just my opinion, of course, but I find the "faithful" style in DPP to be good enough I don't need a curves adjustment in PSCS2 afterwards. But that said-- I generally use ACR in day to day convertions just because it's faster and less work and reliable enough for day to day conversions.

But again... for files where I want the best quality conversion, I'll use DPP.

monkeydeus

Not to be a DPP apologist, because the lack of parametric adjustments on the histogram and tone curve bother me a little too, but the improved color and sharpness for me mitigate the lack of fine control. Plus, one can argue, do you want to rely on the computer's telling you that your blacks are at 8,8,8, or do you want to rely on your eye telling you it's balanced well? But, if you want a _little_ more control, you can use the histogram on the JPEG tab to show you how close you are to clipping in luminance or any of the individual channels.

Duffy Pratt wrote:

I don't think that's what Dan is saying. If so, big deal.

Your broader point is the one that interests me. I've done some exploring of DPP now, and I can't see the usefulness of any of it, beyond getting as full a conversion as possible. Not only does it not have the PS flexibility, but unless I'm missing something, it almost completely lacks the metrics. The best you seem to be able to do is to set one neutral point for white balance. I guess that's fine, if there is only one neutral point.

Duffy Pratt

monkeydeus wrote:

To Duffy's comments about not caring about destructibility: if you prefer a jpeg or tiff it's your prerogative, but RAW gives you the ultimate flexibility to go back and start all over later if you want to: why would anyone make an argument against that? You can say it's not for you, but you can't say that's a bad thing.

Of course, I save the originals of all my keepers. So there is no argument about that. I agree everyone should be able to go back to square one, and saving the original lets you do that. It's the idea that you shouldn't be harming any pixels in the color correction process that I don't understand.

Duffy

Duffy Pratt

monkeydeus wrote:

Plus, one can argue, do you want to rely on the computer's telling you that your blacks are at 8,8,8, or do you want to rely on your eye telling you it's balanced well?

Ideally I want both. But for neutrality, especially when dealing with subtle casts, I would trust the numbers over my eyes, if I was absolutely faced with a choice. It doesn't take much staring at an image on a monitor for chromatic adaption to set in. So, your eyes could be telling you that something is neutral, while the numbers show that its off a bit. In that case, if I had some other reason to distrust the numbers, I would step away from the monitor for awhile so I could get a fresh look to make sure that my eyes were not deceiving me.

Duffy

claudermilk

Duffy Pratt wrote:

... It's the idea that you shouldn't be harming any pixels in the color correction process that I don't understand.

Duffy

I don't look at it as "harming pixels" as much as retaining as much data as I can for as long as I can. Is that such a bad thing?

monkeydeus

I hate getting sucked in to these debates But, if the "don't harm the pixels" thing isn't working for you, as claudermilk said, think of it as retaining control. If you shoot JPEG, sharpening and white balance decisions are irreversibly made for you. You can approximate WB adjustments on a JPEG in PS, but mathematically you just cannot replicate the same white balance shifting on converted data that you can on RAW data, as I understand it. So, you'd have a much harder time doing the faux-HDR thing of exposing a pic multiple ways to get the sky and shadows right for example. I'm not trying to convert anyone, but just because the advantages don't work for you, it doesn't mean that there aren't in fact advantages.

Duffy Pratt

I haven't argued once against shooting in RAW. Whatever file you start with, lets suppose there are two methods. One gives a better final result. The other destroys pixels or data. Which would you choose?

If taking the time to do non-destructive edits will lead to better results, then I'm all for it. But I'd like someone to actually show me the better results, instead of arguing from the mathematics.

Duffy

claudermilk

But, if it takes no more time than a destructive edit?

While this situation is not guaranteed, it does happen to be how most of my workflow is. I use Bibble Pro & most of my adjusting takes place there instead of PS. Much of it the same operations as used to be done in PS. No extra time & non-destructive. IMHO that's a win-win.

I think it's obvious we'd all choose the method that gives us a better final result. My point is that trying to maintain as much of the original data as long as possible gives you more room to get to that final result. So now, I thnk we're just running is circles here. :skippy