Sharpening, etc during RAW conversion

[RobertP]

I'd like to hear some views on the merits of adjusting images during the RAW-TIFF conversion vs. doing so afterwards, working with TIFFs in Photoshop. I understand that some people think you should do as much adjusting as possible during conversion because you're working with 10 bits (per channel?) at this stage, as opposed to 8 bits after conversion. But is this difference significant when it comes to image quality? That is, is it really worth using the relatively crude adjustment tools provided by the conversion utility, rather than Photoshop's far more flexible and sophisticated functions?

How does the above argument apply to the sharpness function of the conversion utility? Can you get a little extra sharpness by choosing "High sharpness" during the conversion without the sharpening artefacts associated with Photoshop's Unsharp Mask? I've tried setting the conversion sharpness to high, but in truth, the effect is so minimal as to seem worthless when compared to Unsharp Mask.

[Pekka]

It would be of course nice to do it all in Photoshop, but you can not export a 36-bit TIFF or even 30-bit TIFF from G1's Twain driver. If you work in Twain driver, you'll work on 10 bits per channel so level truncation and colorspace conversion is done after sharpening, WB and levels. That is a good thing. If you export to 24 bits, you've lost original data and have no way to restore it.

With low sharpness in RAW conversion you'll end up with an image that is almost soft and smooth as D-30's. I think it's a good thing to be able to do such TIFFs when you go for that big master print. But it's great to have a "negative" and a choice to "develop" it. I do export often in normal sharpness because it's good enough for my purpose (to see/for 50% size net gallery). Achieving strong and good sharpness that does not affect noise in PS is easiest with low sharpness original and Ultra-Sharpen Pro 3 in my (current) opinion. It creates masks for unsharp mask sharpening. US pro is freeware but needs a full version of PS.

http://photography-on-the.net/forum/showthread.php?t=95 were sharpened with US pro 3.

Twain dialog's low contrast is good for making skintones and highlights less aggressive and bringing out shadow detail. Twain driver contrast settings can of course be simulated with levels/curves in Photoshop, but again, you work in 24 bits in PS and 30 bits in Twain dialog.

Same goes for saturation, easy in PS, but the extra dynamic range makes Twain dialog more effective.

I'd say do in RAW->TIFF as much as you can and then edit in PS.

I'd like to see a 36-bit CRW import module for PS, that would solve the whole 24 vs. 30 bit dynamic range issue.

[Andrei]

I've read again one extremly useful article about Dynamic range, 24 or 30 or 36 and so on:

http://www.scantips.com/basics14.html


Below is citing (rather big, but interesting):

----------
Said again because it's the main point in a difficult subject: Because 10 bits can contain 4 times the range and number of steps of tones that 8 bits can, 10 bits makes a fuller range easier, and can also extend the range into deeper areas. In a 30 bit scanner, the hard work of handling logarithmic data has the advantage of working with 10 bits of data, instead of 8 bits. The improvement is in the extended Dynamic Range, i.e., the ability to differentiate deep black tones.

Some scanners can return more than 24 bits to the output port, to be used for histogram and curve processing in external software. However, there is little point of this if the scanner can do it first. The truth is that very little image programs can handle 30 or 36 bit images. PhotoShop 4.0 can, but can only do histogram and curves, and then the image must be converted to 24 bits for any additional processing. And that job is better done in the scanner. It is the reason the scanner does it. Scanner firmware is the hard way to do it, but it is best, so it is provided. And most image programs or file formats can't handle more than 24 bits either. Frankly, it's not important, whatever would you do with such an image anyway?

The benefit of 30 or 36 bits is only in the internal processing in the scanner, to be able to get a "better 24 bits", and we we are going to get to this in just a few more paragraphs. The main point of this extra range internally within the scanner is that it allows the 30 bit scanner to capture a higher range of image density for its internal gamma processing. It is often said that the 30 bit scanner returns the "best" 24 bits, as if it simply discards a bit or two of noise here and there. That is an extremely misleading and simplistic way to look at it. It's more correct to say it returns better 24 bit values (especially at the black end).
---------
So, this is what a 30 bit scanner does and how it does it. 30 bits do not affect colors. It does NOT deliver better color in the RGB sense, it cannot, it returns only 24 bits. It does deliver better detail in shadow tones, if such deep dark detail exists in the image. The 30 bits remain internal, and are not returned to the scanner driver
----------------------------

Conclusion: to get good 24 bit for post-processing in PS, use 30-bit pre-processing in Twain driver.

[RobertP]

Thanks, Pekka and Andrei. So you see a perceptible quality difference between images in which contrast/saturation/white balance has been done prior to conversion vs. after (in Photoshop)? What's the difference--more artefacts in the latter?

Regarding Pekka's current sharpening strategy, I'd love to try it, but as a Mac user, I don't have access to Ultra Sharpen Pro. Is it simply a filter that applies some pre-chosen settings to Photoshop's Unsharp Mask? Could I duplicate the effect by carefully choosing my own Unsharp Mask settings, or does the filter do something I can't do myself in Photoshop?

Thanks again for the advice.

[Andrei]

To be honest, I don't see visible degrade in quality using JPEG. I've been working with SuperFine and I'm still happy. But I have to admit, that quality of picture in shadows is better (at least it's easier to preserve more details) using CRW as pre-processing before printing.

So happened, that for the last days I got Pekka interested in PS, but Pekka got me interested in CRW conversion Now I see how another expenses are coming : CD-RW, 1GB Microdrive ....

RobertP wrote:
Thanks, Pekka and Andrei. So you see a perceptible quality difference between images in which contrast/saturation/white balance has been done prior to conversion vs. after (in Photoshop)? What's the difference--more artefacts in the latter?

Regarding Pekka's current sharpening strategy, I'd love to try it, but as a Mac user, I don't have access to Ultra Sharpen Pro. Is it simply a filter that applies some pre-chosen settings to Photoshop's Unsharp Mask? Could I duplicate the effect by carefully choosing my own Unsharp Mask settings, or does the filter do something I can't do myself in Photoshop?

Thanks again for the advice.

[Mike K]

" Achieving strong and good sharpness that does not affect noise in PS is easiest with low sharpness original and Ultra-Sharpen Pro 3 in my (current) opinion. It creates masks for unsharp mask sharpening. US pro is freeware but needs a full version of PS."

Does US Pro3 work with Photoshop Elements? It does seem to support plug ins, is that sufficient? Do you have a URL ?

[jglodge]

Yes US3 does work with Elements haven't figgered out how to get the keyboard shortcut to work but it works through File/automate

[Sherman]

I must admit that this discussion make me a bit confused.

Isn`t is ok to do like this.

First I convert the RAW pictue with Yarc (only yarc.exe). Then I open the tiff with PSP7 and sharpen it with KTP Equalizer.

or...

open the RAW in PSP7 via the Twain driver. PSP then convert it to psp format. I can choose between 8 bit or 16 bit. Have always used the default 8 bit. Should I use 16 Bit?
From there I save it as tiff.

/Sherman