Reading mode: Light Dark

RAW converters: they do 2 things only

Go to last post
A

Andy61557

Senior Member
Messages
1,348
Reaction score
0
Location
ZW
Finally, when memory cards and portable storage (Wolverine 40GB for $180) became reasonable, I can afford to shoot RAW. I was never against it in principal, just couldn't stand having only one third of memory capacity vs. JPEG.

Now about all those RAW converters. The first thing they do is interpolate the 4 'pixels' of the sensor, or it's 'red-green-blue-green' group, into 4 pixels of the image to be saved. Depending on the converter's algorithm they do it with different sharpness and moire-related results. The other thing they do in this process is applying the white ballance: changing weights of the red, green and blue components. This is all. Then they just create a 16-bit per pixel per color image from the 12-bit raw data. The highest 4 bits reman as zeros. This image can be saved as is or converted into 8-tib format.

All other adjustments, such as exposure compenstaion, brightness and contrast, sharpness, etc., can be done in Photoshop with Curves and Levels. If you watch the histogram in Adobe Raw during exposure compensation, you will notice that it's nothing more than adjusting the curves.

What I'm trying to say is: Adobe Raw is just as good as any other, if you leave all adjustments to the Photoshop.

Now about my RAW workflow. The pro guys day Adobe Raw has none, but maybe they just want to do more in it that they really have to.

1. Copy a batch of RAW files into a directory. Browse the files in CS File Browser and delete the 'garbage'.

2. Batch-convert all files into a 'flat' RAW format: 0 for all parameters (exposure, brightness, ets.). Leave white balance "as shot" or apply it in batch for groups of shots, maybe one-by one.

3. Save all shots as 16-bit TIFF's. This is what gives you the 'extra stops' vs. 8-bit JPEG files. Actually you work with 12 bits, but the computers can't handle a 'byte and a half', they need two.
4. Edit all TIFF's for exposure, sharpness, crop and everyting else.

5. Convert all files into 8-bit TIFF's. This is your final image, not for any further adjustments. They are about 3 times smaller than 16-bit images, not 2. I guess the LZW algorithm doesn't work that well here.

6. Copy all RAW files into Archive and burn a DVD (2 copies) when enough files are accumulated.

Now please tell me what you think. Just please try to give an explanation for all your thoughts based on your technical knowledge. Not just "you can adjust exposure in Photoshop!". Explain how 'exposure compensation' in Adobe Raw, or any other converter, is diffrerent from Curves.
Hugs and kisses to all
--
http://www.pbase.com/andybelov
 
I like this, someone talking sense about RAW files for a change.
--
jase
 
I believe you're going to get a few more responses on this but they do other things as, that's why they differ. Thomas Knoll stated that the new raw converter in CS2 actually processed some of the 'problem' lines found 20D boosted shadows out of the picture, so there's more to this than meets the eye.
 
Andy said:
All other adjustments, such as exposure compenstaion, brightness
and contrast, sharpness, etc., can be done in Photoshop with Curves
and Levels. If you watch the histogram in Adobe Raw during exposure
compensation, you will notice that it's nothing more than adjusting
the curves.
Click to expand...
That is obviously true, however there are definite advantages to doing as much as you can at the RAW-conversion stage vs. later in the process. The RAW data is linear and the curve used in the conversion is not like any curve you're likely to ever use in Photoshop. It does many of the same things but additionaly it remaps the data from a linear gamma of 1.0 to Mac or Win gamma (1.8 or 2.2.) There is a tremendous abundance of data at the bright end and very little data at the dark end. So, at the very least it is highly desireable to adjust overall tonality and contrast here instead of later in Photoshop - the process is much less destructive and avoids possible posterization if you manipulate the tonality later.

Don't be mislead by the representation of curves in the user interface--they are deliberately misrepresented to give you something to work with that is more familiar-looking.

You may want to have a look at Bruce Fraser's "old" Adobe Camera RAW book or the new version devoted to ACR 3.x.

Or take a look at some of his posts in the Rob Galbraith forum such as this one: http://forums.robgalbraith.com/showflat.php?Cat=0&Number=346233&an=0&page=0#346233
 
Please, just compare a blown JPEG and the converted RAW file, a yellowish or blueish WB gone JPEG and the converted RAW, a noisy 1600 ISO JPEG and the converted RAW.

And try the "fill light" adjustment in RSE, just give it a try.

I love CS2 and my 20D but there's no CPU, VLSI or machine able to do a better job on the RAW material than you, human.

Ludo
http://ludo.smugmug.com
Andy said:
Finally, when memory cards and portable storage (Wolverine 40GB for
$180) became reasonable, I can afford to shoot RAW. I was never
against it in principal, just couldn't stand having only one third
of memory capacity vs. JPEG.
Now about all those RAW converters. The first thing they do is
interpolate the 4 'pixels' of the sensor, or it's
'red-green-blue-green' group, into 4 pixels of the image to be
saved. Depending on the converter's algorithm they do it with
different sharpness and moire-related results. The other thing they
do in this process is applying the white ballance: changing weights
of the red, green and blue components. This is all. Then they just
create a 16-bit per pixel per color image from the 12-bit raw data.
The highest 4 bits reman as zeros. This image can be saved as is or
converted into 8-tib format.
All other adjustments, such as exposure compenstaion, brightness
and contrast, sharpness, etc., can be done in Photoshop with Curves
and Levels. If you watch the histogram in Adobe Raw during exposure
compensation, you will notice that it's nothing more than adjusting
the curves.
What I'm trying to say is: Adobe Raw is just as good as any other,
if you leave all adjustments to the Photoshop.
Now about my RAW workflow. The pro guys day Adobe Raw has none, but
maybe they just want to do more in it that they really have to.
1. Copy a batch of RAW files into a directory. Browse the files in
CS File Browser and delete the 'garbage'.
2. Batch-convert all files into a 'flat' RAW format: 0 for all
parameters (exposure, brightness, ets.). Leave white balance "as
shot" or apply it in batch for groups of shots, maybe one-by one.
3. Save all shots as 16-bit TIFF's. This is what gives you the
'extra stops' vs. 8-bit JPEG files. Actually you work with 12 bits,
but the computers can't handle a 'byte and a half', they need two.
4. Edit all TIFF's for exposure, sharpness, crop and everyting else.
5. Convert all files into 8-bit TIFF's. This is your final image,
not for any further adjustments. They are about 3 times smaller
than 16-bit images, not 2. I guess the LZW algorithm doesn't work
that well here.
6. Copy all RAW files into Archive and burn a DVD (2 copies) when
enough files are accumulated.
Now please tell me what you think. Just please try to give an
explanation for all your thoughts based on your technical
knowledge. Not just "you can adjust exposure in Photoshop!".
Explain how 'exposure compensation' in Adobe Raw, or any other
converter, is diffrerent from Curves.
Hugs and kisses to all
--
http://www.pbase.com/andybelov
Click to expand...
 
If you convert to a linear 16 bit tiff format, then there isn't a data loss issue. No gamma curve.
 
I agree with the above posters comment on blown highlights (I think). The raw converter can take a more intelligent approach to fixed the problem than Photoshop normally would.

One question, why the 8 bit tiff vs. a jpeg?
 
I've been trying to work with some RAW images, but I have not been able to get the sharpness to look right either in ACR or in PS. I just haven't been able to find the recipe to even make the raw look as good as the JPEG capture both.

What's normally done for sharpness?
 
"Gamma

The brightness of the midtone values. The values produced by a monitor from black to white are nonlinear--if you graph the values, they form a curve, not a straight line. The gamma value defines the slope of that curve halfway between black and white. Gamma adjustment compensates for the nonlinear tonal reproduction of output devices such as monitor tubes."
I don't see how it's different from Curves.

Guys, think about it. What is an image? It's an array of pixels, where each pixel has only three things: it's red, green and blue values. That's it. There is nothing more to change. The only other components is the pixel's coordinates. Adding this to the color adjustments allow you to change sharpness.

Linear vs. non-linear RAW convertion means nothing more than applying cureves during the conversions. Yes, the sensor has different sensitivities in the dark and light areas and the curves are used to compensate for this.

But it doesn't matter when the curves are applied. The image data is in the same 12-bits, only in TIFF format it's padded up to 16-bits, not changed.

And there are no hidden procedures during the curve convertion, there can't be. All the curves do is tell the converter what input values need to be changed to output. If the red is 5 on input, it must be changed to 7. If it's 201, it must be 210, etc. Even the mane of this is very simple: look up table.
I am a professional programmer and work with image processing for many years.

--
http://www.pbase.com/andybelov
 
Don Glenn said:
If you convert to a linear 16 bit tiff format, then there isn't a
data loss issue. No gamma curve.
Click to expand...
Almost correct. A RAW converter can apply all transformations at once, avoiding saving intermediate steps to 16-bit format and thus accumulating round-off error (albeit small). Also, if you convert to a linear file, you need to then apply gamma as part of applying your custom tone curve in PS. This type of curve is hard to create manually on your own. There have been many threads here on processing linear data, ultimately resulting in those methods being abandoned by those initially championing them.

One nice thing is that RAW converters typically provide a selection of default tone curves selectable via drop-down menu (C1 provides an enhanced shadows "canned" curve, for example), which can get you closer to your final result with less effort. Slider controls that control the toe and shoulder of the tone curve, such as in RawShooter, are also useful and quicker to use compared with dragging points on an X-Y plot representing the curve.

David
 
About highlights. Just like I've said: no sense in stating something without expalining. Blown highlights are due to 8-bit vs. 16-bit (12-bit) limitation, not the 'RAW intelligense'.
8-bit TIFF is lossless, JPEG gives you artifacts and degrades after adjustments.
--
http://www.pbase.com/andybelov
 
Since many operations are best performed in the linear regime (white balance, initial "capture" sharpening), postprocessing everything in PS requires starting with a linear conversion. Yes, all subsequent steps such as application of gamma and a tone response curve can be performed in PS. However, small deviations in these curves can result in major visible differences, making it tough to create good-looking curves just by dragging dots up and down on a curve-editing tool. It helps that RAW converters apply tone transformations for you, offering tweaks such as a selection of such curves as well as sliders to control shadow and highlight contrast.

I've tried starting with a truly linear file (such as spit out by the Canon SDK) and manipulating it to a good-looking image using PS. It can be a major effort.

David
 
I think Adobe is trying to be nice to those who wants to do all the processing in one place. You open your file in Adobe Raw, adjust everyting and get the fine you need. I don't see any other reason for that.

Everywhere you look, people say that you have much more control over sharpness in Photoshop.
--
http://www.pbase.com/andybelov
 
Try RSE, it features a double approach to sharpness through the "detail" and "sharpness" adjustments. Very nice software.

Ludo
AussieDawg said:
I've been trying to work with some RAW images, but I have not been
able to get the sharpness to look right either in ACR or in PS. I
just haven't been able to find the recipe to even make the raw look
as good as the JPEG capture both.

What's normally done for sharpness?
Click to expand...
 
AussieDawg said:
I've been trying to work with some RAW images, but I have not been
able to get the sharpness to look right either in ACR or in PS. I
just haven't been able to find the recipe to even make the raw look
as good as the JPEG capture both.

What's normally done for sharpness?
Click to expand...
I used to shoot JPG and process in CS using Fred Miranda's CSPro.

Now I shoot RAW and process in CS2 using ACR. My setting is zero for sharpness in ACR. I then sharpen using Smart Sharpen in CS2. Smart Sharpen is amazing and eliminates the problem of over sharpened halos.

--
Gary Coombs, W9VJ
http://GaryCoombs.com
My Profile contains my Equipment List

A good photograph is knowing where to stand. -Ansel Adams
 
You know, you really don't need any RAW converter or Photoshop at all. All you need is a pen and a stack of paper and you can calculate all the conversions / adjustments manually yourself! After all, a pixel is nothing more than a numerical value, and all conversions / adjustments are nothing more than a set of transformations on these numerical values!

Ok, that's a bit of an exaggerated example. The point, though, is that just because all methods can give you the same result does not mean all methods are equally good or usable. As previous posters have already pointed out, RAW converters allow you to make adjustments to the image's linear data before it is gamma-corrected. Sure, you can do the same thing in Photoshop by first converting the image linearly, perform adjustments, then gamma-correct. But it's much easier and intuitive to do it directly in the RAW converter. So why not take advantage of it?

Incidentally, there is I believe one feature in ACR that is not readily duplicable in Photoshop. When you have an image with clipped highlight in one channel and you perform a negative exposure compensation, ACR actually attempts to recover the clipped channel by examining the un-clipped channels as well as the pixels surrounding the clipped pixel(s). Obviously, this doesn't work if all of your channels are clipped.

Andy
 
Thanks. I'm not aware of Smart Sharpen. How do you access it? I don't have CS2 available now to check it out.
 
I was assuming a linear icc profile was in use. I wouldn't think it would be possible otherwise.

Anyway, I don't work in 16 bit tiff. I take a approach and apply more changes during conversion, saving the file to jpeg.
 
I wouldn't bet on it at this point, but I can't offer much proof either.

Some of the converters appear to change the demosaic process for blown pixels. This way, it would rely more on the neighboring pixels then it normally would.
.
 
Don Glenn said:
If you convert to a linear 16 bit tiff format, then there isn't a
data loss issue. No gamma curve.
Click to expand...
First of all, processing a linear tiff in PS (and I have done it) is a royal pain in the a* . The good RAW converters do it quicker and produce better results. There are some uses for linear tiff conversions (blending exposures is one of these since the highlights have tons of data) but as an everyday technique for all RAWs it would be unnecesarily clunky for no better (and usualy wors) results.

Secondly there are many, many things going on "under the hood" of the better RAW converters such as ACR, C1, etc. sch as some slight sharpening, noise reduction, de-banding, etc. over and above the user controls. There are subtle differences between the best output from ACR, C1, DPP, EVU, RSE that are reflective of those other functions.
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top