Digital BW, The Print

For those of you converting Canon D30 images to black-and-white, what is your 
procedure?  Do you convert immediately and then make all adjustments in 
Photoshop to the black-and-white image?  Or, do you make some adjustments to 
the color image, convert to black-and-white, and make further adjustments?

Helene

> For those of you converting Canon D30 images to black-and-white, what is
your 
> procedure?  Do you convert immediately and then make all adjustments in
> Photoshop to the black-and-white image?  Or, do you make some adjustments to
> the color image, convert to black-and-white, and make further adjustments?
> 
> Helene

I shoot in RAW mode, convert to 16 bit linear tiff with Canon's Image
Converter app, bring the file into Photoshop 6 as is (no color management),
assign a custom D30 linear profile, and then convert to my working space
(Adobe RGB).  I find that using the linear tiff conversion gives much better
control over highlights.  This part of the workflow is similar to that
developed by Fred Miranda (http://www.fredmiranda.com/LPbatch/index.html ).
If the file is destined for B&W printing then the next step would be
conversion to 16 bit greyscale (either using the channel mixer or the
Lightness channel in LAB mode).  Any required levels or curves adjustments
and sharpening are made in 16 bit mode before converting to 8 bit for
printing.

on 2/12/02 8:05 PM, Carl Schofield wrote:

> If the file is destined for B&W printing then the next step would be
> conversion to 16 bit greyscale (either using the channel mixer or the
> Lightness channel in LAB mode).  Any required levels or curves adjustments
> and sharpening are made in 16 bit mode before converting to 8 bit for
> printing.  


FWIW, depending on your output, converting to 8-bit for printing may not be
necessary. The Piezo driver ver6 (I'm told) will actually utilize all
16-bits of data, and even if printing through the Epson driver you can just
send the 16-bit file to print and either Photoshop or the Epson driver will
downsample the data on the fly leaving your file unscathed.

It may not make a difference in print, but since you went to the trouble to
stay in the high bit mode throughout, might as well keep it there. Plus it
saves you a step.

Todd

> on 2/12/02 8:05 PM, Carl Schofield wrote:
> 
>> If the file is destined for B&W printing then the next step would be
>> conversion to 16 bit greyscale (either using the channel mixer or the
>> Lightness channel in LAB mode).  Any required levels or curves adjustments
>> and sharpening are made in 16 bit mode before converting to 8 bit for
>> printing.  
> 
> 
> FWIW, depending on your output, converting to 8-bit for printing may not be
> necessary. The Piezo driver ver6 (I'm told) will actually utilize all
> 16-bits of data, and even if printing through the Epson driver you can just
> send the 16-bit file to print and either Photoshop or the Epson driver will
> downsample the data on the fly leaving your file unscathed.
> 
> It may not make a difference in print, but since you went to the trouble to
> stay in the high bit mode throughout, might as well keep it there. Plus it
> saves you a step.
> 
> Todd

Thanks for the tip Todd.  I didn't know that this was possible.  I'm using
MIS VM inks with Paul Roark's curves and have been converting back to RGB
after the 8 bit conversion.  I'm wondering if I could just skip the
greyscale conversion at the end of the workflow I described previously, stay
in 16 bit RGB mode after using the channel mixer monochrome conversion,
apply Paul's curves, and then just send the 16 bit file to the printer.

on 2/13/02 8:40 AM, Carl Schofield wrote:

> Thanks for the tip Todd.  I didn't know that this was possible.  I'm using
> MIS VM inks with Paul Roark's curves and have been converting back to RGB
> after the 8 bit conversion.  I'm wondering if I could just skip the
> greyscale conversion at the end of the workflow I described previously, stay
> in 16 bit RGB mode after using the channel mixer monochrome conversion,
> apply Paul's curves, and then just send the 16 bit file to the printer.

Yup!

First off, you don't need to convert to grayscale (after your chmx) and then
back to RGB. The chmx operation just made all three channels the same
anyway, you'd only need to convert to grayscale for a different workflow,
like Piezo. Just stay in RGB.

Then as to applying Paul's curves to your 16-bit data, there is a pro and
con to this.

Pro: Paul's curves are pretty radical so applying them to 16-bit data should
be much less destructive to the image than doing the same to 8-bit data.
(though we see from experience it typically works fine with 8-bit data,
which says a lot about how large of edits CAN be applied to 8-bit data. Some
people are very conservative about that, but day to day use shows it ain't a
big deal.) Anyway, yes, it allows you to stay in a 16-bit workflow for all
it's worth. And perhaps in the future even the Epson drivers will utilize
16-bit data and you'll be well positioned to take advantage of that.

Con: That curve set is now "one" with your image, so there'll be no changing
your mind about print tone later. In 8-bit mode you can just put the curves
on an adjustment layer and change your mind over and over - not so in
16-bit. The solution is just to duplicate the file so that you keep one
virgin and curve your duplicates.

I say try it all ways. Dupe your 16-bit file twice, then store the original
away safely. To one dupe add your curves directly in 16-bit mode and send
that to the printer directly. To the other, convert it down to 8-bit, apply
Paul's curves as an adjustment layer, than print. Compare.

Whether applying Paul's curves in 16-bit mode vs 8-bit will be advantageous
may depend upon how sound your 8-bit file is to begin with. In your workflow
it should be fine, but for someone who works their file a lot in 8-bit mode
first, so that the integrity of the image is hanging by a thread, Paul's
curves *could* push it over the edge. BTW, I would think the same is true
with Piezo profiles (or any other profiles), though I'm not truly certain of
how they work.

Todd

Carl,

Yes, you can stay in RGB after channel mixer.  I have been keeping files in 
16 bits.  I apply Paul's curves at the very end, right before printing -- no 
need to go down to 8 bits.

My original question, however, did not relate to how to convert to grayscale. 
 I wanted to know if people do *any* editing at all on the original D30 color 
file (Jerry said he does).  Or, if they go directly to grayscale and do all 
the editing there.

I see a difference and I would guess it's because of working on color vs. 
gray.  I wondered if anyone else has noticed this and have they decided one 
way was better than the other.

Helene 


> 
> 
> > on 2/12/02 8:05 PM, Carl Schofield wrote:
> > 
> >> If the file is destined for B&W printing then the next step would be
> >> conversion to 16 bit greyscale (either using the channel mixer or the
> >> Lightness channel in LAB mode).  Any required levels or curves 
> adjustments
> >> and sharpening are made in 16 bit mode before converting to 8 bit for
> >> printing.  
> > 
> > 
> > FWIW, depending on your output, converting to 8-bit for printing may not 
> be
> > necessary. The Piezo driver ver6 (I'm told) will actually utilize all
> > 16-bits of data, and even if printing through the Epson driver you can 
> just
> > send the 16-bit file to print and either Photoshop or the Epson driver 
> will
> > downsample the data on the fly leaving your file unscathed.
> > 
> > It may not make a difference in print, but since you went to the trouble 
> to
> > stay in the high bit mode throughout, might as well keep it there. Plus 
> it
> > saves you a step.
> > 
> > Todd
> 
> Thanks for the tip Todd.  I didn't know that this was possible.  I'm using
> MIS VM inks with Paul Roark's curves and have been converting back to RGB
> after the 8 bit conversion.  I'm wondering if I could just skip the
> greyscale conversion at the end of the workflow I described previously, 
> stay
> in 16 bit RGB mode after using the channel mixer monochrome conversion,
> apply Paul's curves, and then just send the 16 bit file to the printer. 
> 
> 
> 
> 



[Non-text portions of this message have been removed]

--- In DigitalBlackandWhiteThePrint@y..., Todd Flashner 
<tflash@e...> wrote:
> on 2/13/02 8:40 AM, Carl Schofield wrote:
> 
> > Thanks for the tip Todd.  I didn't know that this was possible.  
I'm using
> > MIS VM inks with Paul Roark's curves and have been 
converting back to RGB
> > after the 8 bit conversion.  I'm wondering if I could just skip 
the
> > greyscale conversion at the end of the workflow I described 
previously, stay
> > in 16 bit RGB mode after using the channel mixer 
monochrome conversion,
> > apply Paul's curves, and then just send the 16 bit file to the 
printer.
> 
> Yup!
> 
> First off, you don't need to convert to grayscale (after your chmx) 
and then
> back to RGB. The chmx operation just made all three channels 
the same
> anyway, you'd only need to convert to grayscale for a different 
workflow,
> like Piezo. Just stay in RGB.
> 
> Then as to applying Paul's curves to your 16-bit data, there is a 
pro and
> con to this.
> 
> Pro: Paul's curves are pretty radical so applying them to 16-bit 
data should
> be much less destructive to the image than doing the same to 
8-bit data.
> (though we see from experience it typically works fine with 8-bit 
data,
> which says a lot about how large of edits CAN be applied to 
8-bit data. Some
> people are very conservative about that, but day to day use 
shows it ain't a
> big deal.) Anyway, yes, it allows you to stay in a 16-bit workflow 
for all
> it's worth. And perhaps in the future even the Epson drivers will 
utilize
> 16-bit data and you'll be well positioned to take advantage of 
that.
> 
> Con: That curve set is now "one" with your image, so there'll be 
no changing
> your mind about print tone later. In 8-bit mode you can just put 
the curves
> on an adjustment layer and change your mind over and over - 
not so in
> 16-bit. The solution is just to duplicate the file so that you keep 
one
> virgin and curve your duplicates.
> 
> I say try it all ways. Dupe your 16-bit file twice, then store the 
original
> away safely. To one dupe add your curves directly in 16-bit 
mode and send
> that to the printer directly. To the other, convert it down to 8-bit, 
apply
> Paul's curves as an adjustment layer, than print. Compare.
> 
> Whether applying Paul's curves in 16-bit mode vs 8-bit will be 
advantageous
> may depend upon how sound your 8-bit file is to begin with. In 
your workflow
> it should be fine, but for someone who works their file a lot in 
8-bit mode
> first, so that the integrity of the image is hanging by a thread, 
Paul's
> curves *could* push it over the edge. BTW, I would think the 
same is true
> with Piezo profiles (or any other profiles), though I'm not truly 
certain of
> how they work.
> 
> Todd
I've found that I can stay in 16 bit RGB workflow, all the way from 
untagged linear tiff to wide gamut to Adobe and finally to srgb 
just before applying Paul's curves and sending the 16 bit file to 
the printer.  Prints are beautiful.  Now using Pekka Saarinen's 
new linear workflow action for processing the linear tiff files.  
Includes a "chooser" function for greyscale conversion offering 3 
channel mixer options and the conversions are placed as 
previews on the history pallette.  The beauty of this approach is 
that you can easily step back to a different greyscale conversion 
or the original rgb after printing - no need for 8 bit conversion and 
adjustment layers.  Pekka's workflow actions are here:
http://photography-on-the.net/D30/linear/