Luminous Landscape Forum

Raw & Post Processing, Printing => Adobe Lightroom Q&A => Topic started by: PBC on November 06, 2012, 04:59:36 pm

Title: Color Profile Question with Round Trip Editing with Photoshop
Post by: PBC on November 06, 2012, 04:59:36 pm
I am trying to get my head around this, so sorry if the question doesn't make sense....

The workflow I am currently playing with is: -

a) LR to process the RAW file
b) "Edit In" Photoshop where I "finish" the image including converting to a different Profile (DCAM 3 from Joe Holmes), then assigning one of Joe's profile variants to the image.
c) When finished in Photoshop I save back into LR.
d) In LR, I may make the odd tweak to the image, but nothing much
e) Image will then be Printed or Exported for the Web from LR

The question is, how does LR handle the DCAM 3 Color Space and variant?  Does it respect it for the Print or Export, or does it convert back to Melissa RGB changing what I did within Photoshop?

Or does this just make no sense given the way LR works ?

Appreciate your thoughts

Thanks

Phil

 
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: digitaldog on November 06, 2012, 07:49:12 pm
For export, you're reprocessing so you'll go through the internal color space. For print I believe it will honor the color space of the rendered data (that's how LR and Photoshop also match as they should).
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: eliedinur on November 07, 2012, 04:38:00 am
If exporting you can put it back into DCAM 3 but then you would have to go back to PS again for the Varient because you can't assign in LR.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: Simon Garrett on November 07, 2012, 06:19:42 am
I've read the articles on the Joseph Holmes site (albeit quickly), but I think I'm missing something. 

What is the benefit of assigning his profiles to an image? 

Clearly assigning a profile (as opposed to converting to a profile) alters the colour of the image.  Under what circumstances is this a good idea?
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: Tim Lookingbill on November 07, 2012, 08:25:10 am
I've read the articles on the Joseph Holmes site (albeit quickly), but I think I'm missing something. 

What is the benefit of assigning his profiles to an image? 

Clearly assigning a profile (as opposed to converting to a profile) alters the colour of the image.  Under what circumstances is this a good idea?

Joe's profiles as are some others are engineered to provide better (or characteristically different) overall saturation and definition appearance when assigning to an image that's been processed into another color space slightly smaller over using Photoshop's Hue/Sat tools.

I've done the same technique many years ago restoring bad scans from a consumer grade flatbed scanner except using Joe's free version of Ekta Space RGB by assigning and converting between AdobeRGB and sRGB. I got some odd hue combos that made the image look pleasing. It was crude and took a lot of trial and error but it worked only not on all scans.

Joe's method is a lot more planned out and sophisticated over my early attempts. I've only seen his method used on high quality medium format scans, so I'm not sure what to expect applying it to digital camera based Raw image processing.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: sandymc on November 07, 2012, 08:45:28 am
The question is, how does LR handle the DCAM 3 Color Space and variant?  Does it respect it for the Print or Export, or does it convert back to Melissa RGB changing what I did within Photoshop?

Assuming (and so far as I am aware this is the case) that DCAM 3 is a "smaller" space than ProPhoto, it will make no difference whether or not LR converts to Melissa. It would make a difference if LR assigned Melissa, but as far as I am aware LR will always convert, not assign.

Sandy
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: deejjjaaaa on November 07, 2012, 09:10:42 am
Assuming (and so far as I am aware this is the case) that DCAM 3 is a "smaller" space than ProPhoto, it will make no difference whether or not LR converts to Melissa.

1) why melissargb ? internal colorspace for LR operations shall be with linear gamma

2) except precision with conversion in shadows, right ?
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: digitaldog on November 07, 2012, 09:43:44 am
1) why melissargb ? internal colorspace for LR operations shall be with linear gamma

Correct. Melissa RGB is one of the name of the color space that is used for the non soft proof numeric values and histogram and is indeed based on a sRGB TRC "gamma" using ProPhoto RGB primaries. The internal color space used for processing has no name. That space is also ProPhoto primaries but with a linear TRC.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: sandymc on November 07, 2012, 10:40:17 am
1) why melissargb ? internal colorspace for LR operations shall be with linear gamma

2) except precision with conversion in shadows, right ?

Melissa RGB was the question the OP asked. Internal LR ops are indeed (so far as I'm aware anyway) in a linear space. But it makes no difference - linear or sRGB tone curve or whatever else, as long as it's a convert, not assign, it won't make a difference.

Convert will not make a difference in the shadows either, except to the extent that implementation specifics such as rounding/15 bit precision in the engine impacts on you. Which should not be much.

Sandy

Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: digitaldog on November 07, 2012, 11:43:44 am
Melissa RGB is the name for a color space that only shows you values and a histogram. There's never any processing within the ACR/LR engine that uses it.

Any processing in Develop and other operations (Export to X RGB space) will convert the original data through the linear, ProPhoto primaries.

Assign has no role nor need in Lightroom. There is never any assignment of profiles. There is an assumption of untagged RGB (sRGB). Even then, LR doesn't assign or embed that into the master. If the data isn't sRGB, well you fix that elsewhere while doing something nasty to the bonehead who provided the untagged file into the operation <g>.

What the OP has to consider is the source RGB working space of rendered data being used within Lightroom, what and when LR will convert into another color space (this one quite wide). It can make a difference in the final output depending on what specific operations we're talking about.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: PBC on November 07, 2012, 02:25:55 pm
Thanks guys - slowly things are becoming clearer.

As to why I am doing this; my current workflow is to Export & convert to DCAM 3. In Photoshop I then assign a DCAM 3 variant as I like the saturation treatment that this can provide, which fits with my feel for my images .  i would then finish the file in Photoshop, save as a PSD, then resize, output sharpen using Photokit Sharpener and output to print or web.

Works well for me, but...

I also like the workflow that LR provides and I like the soft-proof facility and Print Module (much better than PS to me).  Given this, I have been trying to work out if I can remain in LR, just doing a return trip to Photoshop and then output the resulting TIFF from LR.

What I don't understand is what impact the assigning of a DCAM3 Variant would have on the LR processes; hence the question.

Appreciate all advice and pointers

Phil

Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: deejjjaaaa on November 07, 2012, 02:35:34 pm
Convert will not make a difference in the shadows either, except to the extent that implementation specifics such as rounding/15 bit precision in the engine impacts on you. Which should not be much.
there is a theory that in shadows it does... more so if we are talking about several roundtrips through ACR/LR
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: Simon Garrett on November 07, 2012, 02:41:24 pm
Joe's profiles as are some others are engineered to provide better (or characteristically different) overall saturation and definition appearance when assigning to an image that's been processed into another color space slightly smaller over using Photoshop's Hue/Sat tools.

I've done the same technique many years ago restoring bad scans from a consumer grade flatbed scanner except using Joe's free version of Ekta Space RGB by assigning and converting between AdobeRGB and sRGB. I got some odd hue combos that made the image look pleasing. It was crude and took a lot of trial and error but it worked only not on all scans.

Joe's method is a lot more planned out and sophisticated over my early attempts. I've only seen his method used on high quality medium format scans, so I'm not sure what to expect applying it to digital camera based Raw image processing.
I've read Joseph's articles again, and it seems that the main purpose of his "special" working spaces is to get optimum gamut and TRC without going to 16 bit per channel. 

However, the idea of assigning "chroma variants" of his working spaces as a way of altering colour seems rather strange, or at least: it's using profiles in a way for which (IMHO) the profile mechanism wasn't designed. 

Assigning a profile is rather like taking a temperature value of 20C and rather than convert to 68F, you simply assign the Farenheit scale and say "the temperature is 20F".  It might achieve what you want, but it's rather subverting the temperature measurement, and it results in values that don't mean what they purport to mean. 

Personally I think that's a bit dodgy.  But I guess that if it works... 
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: digitaldog on November 07, 2012, 02:49:49 pm
Assigning a profile is rather like taking a temperature value of 20C and rather than convert to 68F, you simply assign the Farenheit scale and say "the temperature is 20F".  It might achieve what you want, but it's rather subverting the temperature measurement, and it results in values that don't mean what they purport to mean. 

Personally I think that's a bit dodgy.  But I guess that if it works... 

It's a non destructive edit. Who's to say what the best working space assigned to a pile of numbers really is. We don't think of editing images this way but depending on the edit, this is fair game. You could load an RGB working space profile in Photoshop's Color Settings/Custom and edit the chromaticity values, save it out as an ICC profile and assign it to an image.

I'd rather sort such small visual edits in Lightroom. Then just land in ProPhoto RGB, 16-bit. 
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: Simon Garrett on November 07, 2012, 04:00:35 pm
Good point, Andrew.  I think that like you I'd rather make changes to colour in Lightroom (or Photoshop) than by assigning carefully crafted colour spaces. 
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: sandymc on November 07, 2012, 11:50:47 pm
there is a theory that in shadows it does... more so if we are talking about several roundtrips through ACR/LR

Well, there are situations under which it could. E.g., many older apps use an approximation to the sRGB tone curve rather than the correct curve with the "kink" in the shadows. If you round tripped multiple times between LR and one of those in a space with a sRGB tone curve, then you could get a change in the shadows. But in this case the OP is asking about Adobe <-> Adobe round tripping, and so far as I am aware Adobe use the same math in all their products.

Sandy
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: samueljohnchia on November 08, 2012, 04:41:11 am
Phil and Sandy, do note that Photoshop still uses a simplified version of the tone curve in Joe's profile. Instead of the 1024 precision point curve, Photoshop approximates it as gamma1.96. That what Photoshop uses. Photoshop also uses the simplified gamma 2.2 tone curve for sRGB.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: sandymc on November 08, 2012, 04:56:39 am
Phil and Sandy, do note that Photoshop still uses a simplified version of the tone curve in Joe's profile. Instead of the 1024 precision point curve, Photoshop approximates it as gamma1.96. That what Photoshop uses. Photoshop also uses the simplified gamma 2.2 tone curve for sRGB.

Really  ???

Might I ask your source for that? My understanding is that modern versions of PhotoShop and LR fully respect profiles. (Note that it doesn't matter whether the internal "Melissa RGB" uses sRGB or the approximation; all that matters is whether the profile in a file that's read in is respected.

But again, when round tripping,  it wouldn't actually make a difference as long as both LR and PS were using the same curve. Problem comes about when one uses the approximation, and one use the correct curve.

Sandy
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: samueljohnchia on November 08, 2012, 05:25:57 am
Sandy, please correct me if I am wrong. The version of the profile that Photoshop understands and uses is what you see when you 1.select the profile in the "working spaces RGB" menu, then select 'custom' in the same menu. It shows gamma 2.2 for sRGB, and gamma 1.96 for Joe's profiles. It seems only to be so for working space profiles, photoshop will implement a simplified version. Is there a method for one to verify that gamma 2.2 or the original tone curve of sRGB is used when implemented as a working space profile in photoshop?
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: samueljohnchia on November 08, 2012, 05:37:04 am
I retract my previous statement that Photoshop implements a simplified gamma, instead of honoring the internal tone curve of working space profiles. I have verified this using synthetic grayscale gradients created in documents with gamma 2.2 and sRGB, and also gamma 1.96 and Joe's profiles. Sorry for the confusion.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: digitaldog on November 08, 2012, 09:38:46 am
The sRGB profile does have a true TRC and not a gamma curve and Photoshop honors this. In Color Settings, when you futz around with the gamma settings, you'll probably not get this TRC. IOW, typing 2.2 gamma should provide this while the actual sRGB profile has that funny bump down in the shadows.

If there's some old software that accesses the sRGB profile as designed, and uses a simple 2.2 gamma curve, well you probably don't want to be using that product assuming it can even run on a modern OS.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: deejjjaaaa on November 08, 2012, 10:10:15 am
Well, there are situations under which it could. E.g., many older apps use an approximation to the sRGB tone curve rather than the correct curve with the "kink" in the shadows. If you round tripped multiple times between LR and one of those in a space with a sRGB tone curve, then you could get a change in the shadows. But in this case the OP is asking about Adobe <-> Adobe round tripping, and so far as I am aware Adobe use the same math in all their products.

Sandy

yes, but is that math precise enough to handle roundtrips where different colorspaces coordinates and different curves are used, as we are talking about recalculations between prophoto coordinates w/ g1 inside and different colorspace/curve outside

PS: and I am not sure that math is the same either... it is ACR/LR vs PS... different code, isn't it ?
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: samueljohnchia on November 08, 2012, 10:19:03 am
Andrew, you are right. I do apologise if my post confused anyone. I am intrigued though, since Photoshop calculates alpha channels of the current document based on your gray gamma settings in Photoshop, which means that one cannot use any special tone curve that is not a gamma curve, or the  curves Photoshop uses for dot gain. I wonder how the dot gain curves are calculated.
Title: Re: Color Profile Question with Round Trip Editing with Photoshop
Post by: sandymc on November 08, 2012, 10:39:42 am
yes, but is that math precise enough to handle roundtrips where different colorspaces coordinates and different curves are used, as we are talking about recalculations between prophoto coordinates w/ g1 inside and different colorspace/curve outside

PS: and I am not sure that math is the same either... it is ACR/LR vs PS... different code, isn't it ?

Yes, I did say "except to the extent that implementation specifics such as rounding/15 bit precision in the engine impacts on you"  :)  But I'd be surprised if there was a visible difference without doing a LOT of round tripping.

Personally, I've never seen any difference between how LR and PS handle tagged images. I'd think, but have no evidence other than that ACR and LR share common code for raw processing, that the color management code was very similar.

Sandy