Identifying and removing a cast

PE
Posted By
phoney.email
Feb 21, 2004
Views
1386
Replies
13
Status
Closed
I’m looking for a reliable method for *identifying* (and removing) a cast.

The traditional method of setting the black, white and gray points, is not always applicable as not every image has a black, white and gray point! Clipping histograms is also unreliable because in some cases it can actually introduce a cast rather then removing it.

Besides, in spite of all color profiling etc. I just don’t trust my eyes because human perception is very "flexible" and malleable depending on the context. For example, I can create several "corrected" versions of the same image by setting B, W & G points (depending on where I click) and yet they all look "OK" in their own way… Considering I have my complete slide, negative and photo collection to scan, I can’t spend hours on each image tweaking ad nauseam and yet never being completely satisfied with the result.

So, what I’m looking for is an *objective* method that would take me (the unreliable step) out of the equation. Also, I want to streamline the procedure so I can finish all the scanning before the end of this millennium…

Even though I’ve been at this for a few months already and have read a lot, I’m still a relative newbie, so assume nothing… As they say: "Nothing is foolproof because fools are so ingenious!" ;o)

Thanks!

Don.

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

B
bhilton665
Feb 21, 2004
From: (Don)

I’m looking for a reliable method for *identifying* (and removing) a cast.

There’s some good info on "working by the numbers" in Dan Margulis’ books … you can download this chapter for free and get a sense of what he’s about … he claims he can teach a color blind man how to do color correction 🙂

http://www.ledet.com/margulis/PP7_Ch02_ByTheNumbers.pdf … in particular pg 9/27 (9 in this pdf, 27 in the book), 19/37 and 21/39. Somewhere in the book he has a listing of color values for common colors.
PE
phoney.email
Feb 22, 2004
On 21 Feb 2004 15:04:14 GMT, (Bill Hilton)
wrote:

I’m looking for a reliable method for *identifying* (and removing) a cast.

There’s some good info on "working by the numbers" in Dan Margulis’ books … you can download this chapter for free and get a sense of what he’s about … he claims he can teach a color blind man how to do color correction 🙂
http://www.ledet.com/margulis/PP7_Ch02_ByTheNumbers.pdf … in particular pg 9/27 (9 in this pdf, 27 in the book), 19/37 and 21/39. Somewhere in the book he has a listing of color values for common colors.

Thanks Bill!

I’ve studied the chapter but didn’t really learn anything new. It still boils down to guesswork and "circumstantial evidence". Also, the use of CMYK rather than RGB is a bit awkward. If anything, I find that – at least at the removal stage – LAB appears more convenient.

To be fair, this is all based on a single chapter and therefore not meant as criticism.

Don.
CG
Colin G Edwards
Feb 22, 2004
Besides, in spite of all color profiling etc. I just don’t trust my eyes because human perception is very "flexible" and malleable depending on the context. For example, I can create several "corrected" versions of the same image by setting B, W & G points (depending on where I click) and yet they all look "OK" in their own way… Considering I have my complete slide, negative and photo collection to scan, I can’t spend hours on each image tweaking ad nauseam and yet never being completely satisfied with the result.
So, what I’m looking for is an *objective* method that would take me (the unreliable step) out of the equation. Also, I want to streamline the procedure so I can finish all the scanning before the end of this millennium…

I’m not sure there is a right answer. You make the relevant points yourself: ‘human perception is very "flexible" and malleable depending on the context’ and if you make several different corrections ‘they all look "OK" in their own way…’. Surely, given these points, it’s impossible to find one *objective* method which will appeal to all viewers for all images? Or even just to your own eye?

Maybe the Auto Colour feature is the step you want (you don’t say which version of PS you’re using)? You could do all your scanning, then set up Auto Colour as you want it, then apply it blindly to each image in your collection from the File Browser. I don’t honestly think that you’ll be pleased with all the results though. If all your images are of the same subject, and have been taken in a studio with consistent lighting conditions, it’ll work. If not, how can you avoid looking at each one and deciding what adjustments it needs?
PE
phoney.email
Feb 22, 2004
On Sun, 22 Feb 2004 19:37:02 +0800, Derek Fountain
wrote:

So, what I’m looking for is an *objective* method that would take me (the unreliable step) out of the equation. Also, I want to streamline the procedure so I can finish all the scanning before the end of this millennium…

I’m not sure there is a right answer. You make the relevant points yourself: ‘human perception is very "flexible" and malleable depending on the context’ and if you make several different corrections ‘they all look "OK" in their own way…’. Surely, given these points, it’s impossible to find one *objective* method which will appeal to all viewers for all images? Or even just to your own eye?

What I’m getting at is that if an object is of a certain color in real life I’d like to see that very color in the image (even though, sometimes, we humans may not like the result and prefer an "imperfect" image – which is why several different solutions all may look "good").

To illustrate what I mean here’s an example. We can determine the brightest point in an image in two extreme ways:

1. Well, there is a definite possibility of a firm maybe that this kind-of, sort-of looks like it could be the brightest point in the image. Perhaps?

2. Or, use the threshold tool and then there is no doubt which is the brightest point.

Now, we may not designate that as the white point because it’s a – what do they call it – a "specular" or a "catchlight", but at least there is no doubt which is the brightest point in the image, and this was arrived at with an unbiased and repeatable method.

I was just wondering if there was a similar objective method to identify a cast. But I think you’re right and there is probably no clear answer otherwise it would have been automated long ago (i.e. the rule would be translated into program code).

Maybe the Auto Colour feature is the step you want (you don’t say which version of PS you’re using)? You could do all your scanning, then set up Auto Colour as you want it, then apply it blindly to each image in your collection from the File Browser. I don’t honestly think that you’ll be pleased with all the results though. If all your images are of the same subject, and have been taken in a studio with consistent lighting conditions, it’ll work. If not, how can you avoid looking at each one and deciding what adjustments it needs?

I am prepared, and indeed expect, to do some post-processing, but I’d like the cast removal to be done automagically (read, objectively) so at least I don’t have to start at the foot of the mountain each time but have a level playing field – to mix a few metaphors… ;o)

I’m using version 6, and Auto Color is just too blunt. Like the documentation says, sometimes it actually introduces a cast.

Don.
AQ
Aaron Queenan
Feb 22, 2004
"Derek Fountain" {} wrote in message
} } Besides, in spite of all color profiling etc. I just don’t trust my } } eyes because human perception is very "flexible" and malleable } } depending on the context. For example, I can create several } } "corrected" versions of the same image by setting B, W & G points } } (depending on where I click) and yet they all look "OK" in their own } } way… Considering I have my complete slide, negative and photo } } collection to scan, I can’t spend hours on each image tweaking ad } } nauseam and yet never being completely satisfied with the result. } }
} } So, what I’m looking for is an *objective* method that would take me } } (the unreliable step) out of the equation. Also, I want to streamline } } the procedure so I can finish all the scanning before the end of this } } millennium…
}
} I’m not sure there is a right answer. You make the relevant points yourself:
} ‘human perception is very "flexible" and malleable depending on the } context’ and if you make several different corrections ‘they all look "OK" } in their own way…’. Surely, given these points, it’s impossible to find } one *objective* method which will appeal to all viewers for all images? Or } even just to your own eye?
}
} Maybe the Auto Colour feature is the step you want (you don’t say which } version of PS you’re using)? You could do all your scanning, then set up } Auto Colour as you want it, then apply it blindly to each image in your } collection from the File Browser. I don’t honestly think that you’ll be } pleased with all the results though. If all your images are of the same } subject, and have been taken in a studio with consistent lighting } conditions, it’ll work. If not, how can you avoid looking at each one and } deciding what adjustments it needs?

What would be ideal is a filter and autoexposure system similar to that used by the minilabs. Most minilab prints are, while not as good as they could be, at least acceptable to most people. In other words, the same as the difference between a minilab and professional-quality individual prints.

Does anyone know of software that does this well? The Auto Colour command in Photoshop only seems to do a marginal job on portrait photos, and doesn’t seem to cope with scans from negatives where the colour is oversaturated. Ideally, the filter should just add adjustment layers to the photo, and not touch the image itself.

Thanks,
Aaron Queenan.
WE
Wino Evertz
Feb 22, 2004
"Aaron Queenan" schreef in
bericht
"Derek Fountain" {} wrote in message
} } Besides, in spite of all color profiling etc. I just don’t trust my } } eyes because human perception is very "flexible" and malleable } } depending on the context. For example, I can create several } } "corrected" versions of the same image by setting B, W & G points } } (depending on where I click) and yet they all look "OK" in their own } } way… Considering I have my complete slide, negative and photo } } collection to scan, I can’t spend hours on each image tweaking ad } } nauseam and yet never being completely satisfied with the result. } }
} } So, what I’m looking for is an *objective* method that would take me } } (the unreliable step) out of the equation. Also, I want to streamline } } the procedure so I can finish all the scanning before the end of this } } millennium…
}
} I’m not sure there is a right answer. You make the relevant points yourself:
} ‘human perception is very "flexible" and malleable depending on the } context’ and if you make several different corrections ‘they all look
"OK"
} in their own way…’. Surely, given these points, it’s impossible to
find
} one *objective* method which will appeal to all viewers for all images?
Or
} even just to your own eye?
}
} Maybe the Auto Colour feature is the step you want (you don’t say which } version of PS you’re using)? You could do all your scanning, then set up } Auto Colour as you want it, then apply it blindly to each image in your } collection from the File Browser. I don’t honestly think that you’ll be } pleased with all the results though. If all your images are of the same } subject, and have been taken in a studio with consistent lighting } conditions, it’ll work. If not, how can you avoid looking at each one
and
} deciding what adjustments it needs?

What would be ideal is a filter and autoexposure system similar to that
used
by the minilabs. Most minilab prints are, while not as good as they could be, at least acceptable to most people. In other words, the same as the difference between a minilab and professional-quality individual prints.
Does anyone know of software that does this well? The Auto Colour command in Photoshop only seems to do a marginal job on portrait photos, and
doesn’t
seem to cope with scans from negatives where the colour is oversaturated. Ideally, the filter should just add adjustment layers to the photo, and
not
touch the image itself.

Thanks,
Aaron Queenan.
1. Teststrip from Vivid Details
2. ICorrect from Pictographics
3. DCE-tools (tool: Collercast correction)

All 3 give different results and use a different philosophy.

1. The first is the most professional and can be used as a colour analyser like in the good old colour-darkroom. The "Methamorphosis"-method (Click on the butterfly-icon in the filter) can give surprising good results in really difficult pictures.
2. The second uses automatically a good auto-level algorithm (much better as PS!) and can be very helpfull to starters. It uses referent colour-picking for black/white, skintones, sky and plant-green. There is a risk of over-doing!
3. Is a shareware-tool and it sometimes really can help to find a nearly invisible colourcast.

Wino Evertz
WE
Wino Evertz
Feb 22, 2004
"Wino Evertz" schreef in bericht
"Aaron Queenan" schreef in
bericht
"Derek Fountain" {} wrote in message
} } Besides, in spite of all color profiling etc. I just don’t trust my } } eyes because human perception is very "flexible" and malleable } } depending on the context. For example, I can create several } } "corrected" versions of the same image by setting B, W & G points } } (depending on where I click) and yet they all look "OK" in their own } } way… Considering I have my complete slide, negative and photo } } collection to scan, I can’t spend hours on each image tweaking ad } } nauseam and yet never being completely satisfied with the result. } }
} } So, what I’m looking for is an *objective* method that would take me } } (the unreliable step) out of the equation. Also, I want to
streamline
} } the procedure so I can finish all the scanning before the end of
this
} } millennium…
}
} I’m not sure there is a right answer. You make the relevant points yourself:
} ‘human perception is very "flexible" and malleable depending on the } context’ and if you make several different corrections ‘they all look
"OK"
} in their own way…’. Surely, given these points, it’s impossible to
find
} one *objective* method which will appeal to all viewers for all
images?
Or
} even just to your own eye?
}
} Maybe the Auto Colour feature is the step you want (you don’t say
which
} version of PS you’re using)? You could do all your scanning, then set
up
} Auto Colour as you want it, then apply it blindly to each image in
your
} collection from the File Browser. I don’t honestly think that you’ll
be
} pleased with all the results though. If all your images are of the
same
} subject, and have been taken in a studio with consistent lighting } conditions, it’ll work. If not, how can you avoid looking at each one
and
} deciding what adjustments it needs?

What would be ideal is a filter and autoexposure system similar to that
used
by the minilabs. Most minilab prints are, while not as good as they
could
be, at least acceptable to most people. In other words, the same as the difference between a minilab and professional-quality individual prints.
Does anyone know of software that does this well? The Auto Colour
command
in Photoshop only seems to do a marginal job on portrait photos, and
doesn’t
seem to cope with scans from negatives where the colour is
oversaturated.
Ideally, the filter should just add adjustment layers to the photo, and
not
touch the image itself.

Thanks,
Aaron Queenan.
1. Teststrip from Vivid Details
2. ICorrect from Pictographics
3. DCE-tools (tool: Collercast correction)

All 3 give different results and use a different philosophy.
1. The first is the most professional and can be used as a colour analyser like in the good old colour-darkroom. The "Methamorphosis"-method (Click
on
the butterfly-icon in the filter) can give surprising good results in
really
difficult pictures.
2. The second uses automatically a good auto-level algorithm (much better
as
PS!) and can be very helpfull to starters. It uses referent colour-picking for black/white, skintones, sky and plant-green. There is a risk of over-doing!
3. Is a shareware-tool and it sometimes really can help to find a nearly invisible colourcast.

Wino Evertz
Teststrip is now called 20/20 in an improved version. Have not used that. You can download a demo.

Wino
AQ
Aaron Queenan
Feb 22, 2004
"Wino Evertz" {} wrote in message
}
} "Aaron Queenan" {} schreef in
} bericht } } What would be ideal is a filter and autoexposure system similar to that } used
} } by the minilabs. Most minilab prints are, while not as good as they could
} } be, at least acceptable to most people. In other words, the same as the } } difference between a minilab and professional-quality individual prints. } }
} } Does anyone know of software that does this well? The Auto Colour command
} } in Photoshop only seems to do a marginal job on portrait photos, and } doesn’t
} } seem to cope with scans from negatives where the colour is oversaturated.
} } Ideally, the filter should just add adjustment layers to the photo, and } not
} } touch the image itself.
} }
} } Thanks,
} } Aaron Queenan.
} }
} 1. Teststrip from Vivid Details
} 2. ICorrect from Pictographics
} 3. DCE-tools (tool: Collercast correction)
}
} All 3 give different results and use a different philosophy. }
} 1. The first is the most professional and can be used as a colour analyser } like in the good old colour-darkroom. The "Methamorphosis"-method (Click on
} the butterfly-icon in the filter) can give surprising good results in really
} difficult pictures.
} 2. The second uses automatically a good auto-level algorithm (much better as
} PS!) and can be very helpfull to starters. It uses referent colour-picking } for black/white, skintones, sky and plant-green. There is a risk of } over-doing!
} 3. Is a shareware-tool and it sometimes really can help to find a nearly } invisible colourcast.
}
} Wino Evertz

Thanks, I’ll give them a try!

Aaron.
PE
phoney.email
Feb 23, 2004
On Sun, 22 Feb 2004 20:28:44 +0100, "Wino Evertz" wrote:

1. Teststrip from Vivid Details
2. ICorrect from Pictographics
3. DCE-tools (tool: Collercast correction)

All 3 give different results and use a different philosophy.
1. The first is the most professional and can be used as a colour analyser like in the good old colour-darkroom. The "Methamorphosis"-method (Click on the butterfly-icon in the filter) can give surprising good results in really difficult pictures.
2. The second uses automatically a good auto-level algorithm (much better as PS!) and can be very helpfull to starters. It uses referent colour-picking for black/white, skintones, sky and plant-green. There is a risk of over-doing!
3. Is a shareware-tool and it sometimes really can help to find a nearly invisible colourcast.

Thanks, I’ll try them too! And the links are:

http://www.VividDetails.com/
http://www.picto.com/
http://www.mediachance.com/

Don.
BS
Bob Shomler
Feb 23, 2004
I am prepared, and indeed expect, to do some post-processing, but I’d like the cast removal to be done automagically (read, objectively) so at least I don’t have to start at the foot of the mountain each time but have a level playing field – to mix a few metaphors… ;o)
I’m using version 6, and Auto Color is just too blunt. Like the documentation says, sometimes it actually introduces a cast.
Don.

A couple of possibilities may be found in the last two Photoshop releases (cost of upgrading might be in same ballpark as purchasing another program or a plugin). Photoshop CS has a new Match Color command that has a Neutralize check box option; and Photoshop 7 introduced an Auto Color that, while probably marginal with its default settings, can be quite helpful in a "Semi-Auto Color" operation that Bruce Frazer describes in a Creativepro article: "Out of Gamut: Don’t Underestimate Photoshop’s Auto Color"

www.creativepro.com/story/feature/17164-1.html

Bob Shomler
www.shomler.com
PE
phoney.email
Feb 24, 2004
On Mon, 23 Feb 2004 07:43:23 -0800, Bob Shomler
wrote:

I am prepared, and indeed expect, to do some post-processing, but I’d like the cast removal to be done automagically (read, objectively) so at least I don’t have to start at the foot of the mountain each time but have a level playing field – to mix a few metaphors… ;o)
I’m using version 6, and Auto Color is just too blunt. Like the documentation says, sometimes it actually introduces a cast.
Don.

A couple of possibilities may be found in the last two Photoshop releases (cost of upgrading might be in same ballpark as purchasing another program or a plugin). Photoshop CS has a new Match Color command that has a Neutralize check box option; and Photoshop 7 introduced an Auto Color that, while probably marginal with its default settings, can be quite helpful in a "Semi-Auto Color" operation that Bruce Frazer describes in a Creativepro article: "Out of Gamut: Don’t Underestimate Photoshop’s Auto Color"

www.creativepro.com/story/feature/17164-1.html

Thanks Bob, I’ll chase up the link.

Don.
ME
Mike Engles
Feb 24, 2004
Don wrote:
On Sun, 22 Feb 2004 20:28:44 +0100, "Wino Evertz" wrote:

1. Teststrip from Vivid Details
2. ICorrect from Pictographics
3. DCE-tools (tool: Collercast correction)

All 3 give different results and use a different philosophy.
1. The first is the most professional and can be used as a colour analyser like in the good old colour-darkroom. The "Methamorphosis"-method (Click on the butterfly-icon in the filter) can give surprising good results in really difficult pictures.
2. The second uses automatically a good auto-level algorithm (much better as PS!) and can be very helpfull to starters. It uses referent colour-picking for black/white, skintones, sky and plant-green. There is a risk of over-doing!
3. Is a shareware-tool and it sometimes really can help to find a nearly invisible colourcast.

Thanks, I’ll try them too! And the links are:

http://www.VividDetails.com/
http://www.picto.com/
http://www.mediachance.com/

Don.

Hello

Try the demo of Color Mechanic. It is designed for removing casts. You have to do the work yourself, it is not automatic.

Mike Engles
PE
phoney.email
Feb 25, 2004
On Tue, 24 Feb 2004 23:15:42 +0000 (UTC), Mike Engles
wrote:

Don wrote:
On Sun, 22 Feb 2004 20:28:44 +0100, "Wino Evertz" wrote:

1. Teststrip from Vivid Details
2. ICorrect from Pictographics
3. DCE-tools (tool: Collercast correction)

Thanks, I’ll try them too! And the links are:

http://www.VividDetails.com/
http://www.picto.com/
http://www.mediachance.com/

Try the demo of Color Mechanic. It is designed for removing casts. You have to do the work yourself, it is not automatic.

Hi Mike!

I tried the 3 tools above but they are all a variation on, well, Photoshop’s Image/Adjust/Variations asking me to make a judgement.

As I wrote at the beginning the idea is to take me (the unreliable bit) out of the equation. That’s why I was asking for an objective way of doing this. Once the cast is out, that’s where I come in…

Anyway, I’ll give Color Mechanic a go.

Thanks!

Don.

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

Related Discussion Topics

Nice and short text about related topics in discussion sections