basic scanner questions

All talk about software, scanners, printers, digital backs, etc.
buze
Posts: 86
Joined: Sun Aug 20, 2006 10:31 am Etc/GMT-1+01:00
Location: Windsor, Berks
Contact:

Post by buze » Thu Sep 25, 2008 6:13 pm Etc/GMT-1+01:00

Just imagine color scanning as a matrix instead of as a 'curve'. Each hue will have a small potential percentage of error; some will have considerably more than others. Thats just for the sensor...

Then you factor in the small percentage or error of the emitting light, the small percentage of error attached to /each and every resistor and capacitor/ along the way (you cannot buy '0%' tolerance electronic components - even 'luxury' resistors are 1%), the small percentage of error that is probably implied into the scanner glass physical specification. and you end up with variation that /can/ be quite wild.
Oh, and make it so that temperature, humidity and age of each component also affects all of these factors in non-linear ways.

So the profiler scans a 'known' physical media (these IT87 targets) that have been measured pretty carefully with high quality spectrometry, in a controled environment; and the profiling software tries to figure a correction factor into each of the hues of your matrix, to get as close to possible as what they are 'supposed' to be.

Here is your profile !

Now, do the same for the screen and the printer!

User avatar
Thingy
Forum Hero
Posts: 578
Joined: Wed Aug 20, 2008 11:13 pm Etc/GMT-1+01:00
Location: Orpington, Kent

Re: basic scanner questions

Post by Thingy » Mon Aug 24, 2009 2:25 pm Etc/GMT-1+01:00

RE: Wolf Faust http://www.targets.coloraid.de/

I have finally ordered the Epson V750 scanner but would like advice on the Wolf Faust Scanner Calibration Targets which Joanna recommends. Being a greenhorn to using anything but an office scanner, could someone tell me precisely what these targets do and what do I need for scanning in 6x12cm & 5x4in Fuji films below on the Epson V750?

Velvia 50 & 100
Pro 160s
Acros (Neopan 100)

Can I use these profiles for printer, screen and scanner and if so how do I use them? :?: Also what are the IT 8.7/1, IT 8.7/2 and ISO 12641 standards and how do they affect me? :?:

Please.... plain English if possible! :oops: :oops: :oops: :blink:
Love is an Ebony mounted with a Cooke PS945.......

Joanna Carter
Founder
Posts: 1283
Joined: Wed Mar 29, 2006 7:26 pm Etc/GMT-1+01:00
Workshop Images: http://grandes-images.com/fr/Paysages/P ... _2009.html
Location: Plestin-les-Grèves, France
Contact:

Re: basic scanner questions

Post by Joanna Carter » Mon Aug 24, 2009 7:15 pm Etc/GMT-1+01:00

Thingy wrote:Please.... plain English if possible! :oops: :oops: :oops: :blink:
Mais bien sûr mon ami 8)
Thingy wrote:I have finally ordered the Epson V750 scanner but would like advice on the Wolf Faust Scanner Calibration Targets which Joanna recommends. Being a greenhorn to using anything but an office scanner, could someone tell me precisely what these targets do and what do I need for scanning in 6x12cm & 5x4in Fuji films below on the Epson V750?

Velvia 50 & 100
Pro 160s
Acros (Neopan 100)
You don't need anything in the way of scanner profiles for negative film, only for transparency.
Thingy wrote:Can I use these profiles for printer, screen and scanner and if so how do I use them? :?: Also what are the IT 8.7/1, IT 8.7/2 and ISO 12641 standards and how do they affect me? :?:
The Faust targets are only useful for scanning; you need other stuff for screen and printer.

IT 8.7/2 is a reflective target and is only really useful if you want to scan images printed on the paper the target is for.

IT 8.7/1 is a transmissive target; you will need one target to match each transparency emulsion you plan to scan. In your ecxample, you will need the V1 target for Velvia 50 and the N1 target for Velvia 100 or Astia 100F.

The whole idea of profiling is this:

1. A digital image is stored in your computer and should hold the definitive colour values and levels for that image; a screen profile ensures that your video card displays those colours faithfully, compensating for the differences in colour and brightness of your particular monitor.

2. A scanner profile ensures that, regardless of the non-standard colour and brightness of your scanner illumination, the file created from the scan will contain the corrected colours. A scanner profile is applied to the scanned image after it has been scanned, not during the scanning process; you will see a wierd colour cast on the scanned image after scanning but before applying the appropriate profile.

3. The printer profile takes the correct colours of the image file and ensures that any "mistranslation" of those colours, by the printer driver, is cancelled out to give a true rendering of the image.

Is profiling necessary?

Screen Profiling - this is unavoidable and has to be done if you want to stand any hope of seeing your images correctly displayed, not only in terms of the correct colours, but also having the correct black level and greyscale separation. However, all this ensures is that the colours in the file you are working on are displayed correctly, it has absolutely no effect on either scanning or printing.

Scanner Profiling - each transparency film transmits the light from the scanner in slightly different ways; the light in the scanner can also vary in its colour and brightness. You can get "acceptable" colours without a profile but it is much more difficult to get rid of colour casts without one.

Printer Profiling - Once again, you can get "acceptable" results without a profile but, to get as near as possible to what you are seeing on screen, you really need a profile. Now, you can use "generic" profiles, supplied by the paper manufacturers, or you can get yourself a profiling tool to create your own, more accurate, profiles for each paper and ink combination.

In my opinion, the most economical way of being able to profile all three things (screen, scanner and printer) used to be to buy the Monaco EZColor system which used to provide a "puck" for measuring the screen, along with software that could create profiles for screen, scanner and printer.

Having just checked on the Colour Confidence site, it would now appear that you may have to buy something like the eyeOne 2 kit for screen profiling, the Monaco EZColor software just to do the scanner, and something like the Spyder3Print to profile your printer. :? :(

Finally, you may want to consider getting a Better Scanning film holder from Doug Fisher, as the Epson supplied holders are quite inadequate for anything other than the quickest of previews, usually because the film gets held at the wrong height to be in focus.

BTW, I will be holding another Photoshop for LF Photographers workshop, around the end of October; you might like to consider coming along and finding out how all this stuff fits together :D
Reassure yourself - stroke an Ebony

timparkin
Forum Hero
Posts: 472
Joined: Tue Aug 21, 2007 1:40 pm Etc/GMT-1+01:00
Location: Leeds, West Yorkshire
Contact:

Re: basic scanner questions

Post by timparkin » Mon Aug 24, 2009 8:52 pm Etc/GMT-1+01:00

Joanna Carter wrote:
Thingy wrote:Please.... plain English if possible! :oops: :oops: :oops: :blink:
Finally, you may want to consider getting a Better Scanning film holder from Doug Fisher, as the Epson supplied holders are quite inadequate for anything other than the quickest of previews, usually because the film gets held at the wrong height to be in focus.
My tranny holder was at the optimum height straight out of the box. However it's simple to check if you don't mind a bit of diy. You only need an accuracy of about 0.5mm to get significant results. You can do this by cutting out some plastic/card squares and stacking them under the corners of the tranny holder. Once you are happy, glue them on. They aren't as sensitive to height as it sounds from the reviews, you probably only lose about 20% resolution on a badly calibrated scanner. With a 4x5 tranny that still makes for some big prints :-)

Tim
Waiting for the developing bill - 2 hours (and it's so small now!)

DJ
Site Admin
Posts: 360
Joined: Thu Sep 07, 2006 1:48 pm Etc/GMT-1+01:00
Location: Norfolk

Re: basic scanner questions

Post by DJ » Tue Aug 25, 2009 12:21 pm Etc/GMT-1+01:00

Joanna Carter wrote:
Thingy wrote:Also what are the IT 8.7/1, IT 8.7/2 and ISO 12641 standards and how do they affect me? :?:
IT 8.7/2 is a reflective target and is only really useful if you want to scan images printed on the paper the target is for.
IT 8.7/1 is a transmissive target;
The IT8 targets were originally designed for a visual interpretation of colour, and were never intended for any calibration or profiling purposes, as it turns out, they're quite useful in that respect as well, although not as much as a target specifically designed for it, such as Don Hutcheson's HCT target.
Joanna Carter wrote: you will need one target to match each transparency emulsion you plan to scan. In your ecxample, you will need the V1 target for Velvia 50 and the N1 target for Velvia 100 or Astia 100F.
I'd dispute that, sorry Joanna. The emulsion will make very little difference, you really only need one.

It's important to remember that it's the scanner you're profiling, not the film. The target has been reproduced on the film and then measured with a spectrophotometer, and it is against those measurements you are comparing, not the film characteristics. Technically you could produce a profile from a target printed on toilet paper if you really wanted... This is unlike printer profiling where you are profiling the paper as well, in scanner profiling, you're merely using a measured target.

The target will have certain points in colour-space, which are used to define the profile. The points defined on an IT8 ( or other ) target are pre-defined, and output onto the medium as best possible. Film emulsion will vary in it's ability to reproduce those, and will also be affected by the film development. The reason targets are produced on different emulsions is more because they can, rather than they should. Effectively, they're all trying to reproduce the same colour anyway! Using different emulsions may lead to the points on that target being slightly different to another emulsion, which at best may give the profile a bit of an "accent", but I believe the difference is marginal and completely imperceptible to the Mk1 Human Eyeball.
Joanna Carter wrote: 1. A digital image is stored in your computer and should hold the definitive colour values and levels for that image
Again, sorry, have to dispute that ( she's going to thump me at the next meeting... :wink: ). A digital image stored on a computer, stores no actual colour values at all. Nada. None. The values stored in a jpeg or tiff etc., (for the sake of argument, 8 bit) where we're used to seeing values of 0-255 are not colours, they are percentages of the colour a particular device can either capture or render. i.e. 255 red is 100% red. Of what red? camera red? emulsion red? scanner red? printer red? This is exactly the problem colour management was made to fix. Those "colour" values are abritrary, and have no point of reference. It's only when those percentages are linked with an ICC profile that we can correspond them to a real world colour.

Basically, each device has a different concept of colour, 100% red on a scanner may not be 100% red on a printer, and these values vary from model to model, and unit to unit. ICC profiles speak actual real colours. An ICC profile will map what a device thinks of as colour, so when a scanner says 100% red, you can look in the ICC profile and find out what the scanner thinks is 100% red, and map it to the nearest colour that the printer can produce.

In practice, you translate between these devices through a common denominator, your working space, ICC profiles use CiE LAB as the conduit between these colour spaces, as CiE LAB values can be attributed to a real world colour. As long as you translate colour in ( scanner/camera profiles ) and translate colour out ( monitor/printer profiles ) you have some assurance that your colours have been managed from end to end, and should correspond to the correct colour, or at least as well as your devices can reproduce it. Miss out a stage and you're effectively making large assumptions as to what those colour percentages mean.
Thingy wrote: Please.... plain English if possible!
All devices speak a different language when talking colour, and ICC profiles are the Babelfish :wink:

In a nutshell, that's pretty much what Colour Management is.

On an equipment note, I've tried most of them. I favour the Gretag Macbeth/X-Rite stuff over all others. Colorvision stuff (Spyder etc) I wouldn't touch with a barge-pole. I have Monaco EZ-Colour that came with my V750, and hated it. Personally I would go for an EyeOne system, i1Match is good software and can do monitors and scanners and printers. Have a look at the packages available.

Joanna Carter
Founder
Posts: 1283
Joined: Wed Mar 29, 2006 7:26 pm Etc/GMT-1+01:00
Workshop Images: http://grandes-images.com/fr/Paysages/P ... _2009.html
Location: Plestin-les-Grèves, France
Contact:

Re: basic scanner questions

Post by Joanna Carter » Tue Aug 25, 2009 6:54 pm Etc/GMT-1+01:00

DJ wrote:
Joanna Carter wrote: you will need one target to match each transparency emulsion you plan to scan. In your ecxample, you will need the V1 target for Velvia 50 and the N1 target for Velvia 100 or Astia 100F.
I'd dispute that, sorry Joanna. The emulsion will make very little difference, you really only need one.

It's important to remember that it's the scanner you're profiling, not the film. The target has been reproduced on the film and then measured with a spectrophotometer, and it is against those measurements you are comparing, not the film characteristics. Technically you could produce a profile from a target printed on toilet paper if you really wanted... This is unlike printer profiling where you are profiling the paper as well, in scanner profiling, you're merely using a measured target.
Sorry DJ, but from real world experience, with real targets, it does make a difference. Also, see Tim Parkin's blog for his experimentations.
DJ wrote:
Joanna Carter wrote: 1. A digital image is stored in your computer and should hold the definitive colour values and levels for that image
Again, sorry, have to dispute that ( she's going to thump me at the next meeting... :wink: ). A digital image stored on a computer, stores no actual colour values at all. Nada. None. The values stored in a jpeg or tiff etc., (for the sake of argument, 8 bit) where we're used to seeing values of 0-255 are not colours, they are percentages of the colour a particular device can either capture or render. i.e. 255 red is 100% red. Of what red? camera red? emulsion red? scanner red? printer red? This is exactly the problem colour management was made to fix. Those "colour" values are abritrary, and have no point of reference. It's only when those percentages are linked with an ICC profile that we can correspond them to a real world colour.
Once again :wink: how ever you describe the means of recording the colours, the only thing that remains constant is the record of the colour percentages, or whatever, in the file; and this record is independent of the device from which, or to which it is input/output. Yes, input or working space profiles may become part of that file, as well, but it is possible to store a scanned file with no profile information and no corrections made to allow for the inadequacies of the input device.

When I scan a piece of film, it arrives in Photoshop with no profile at all.

Now, the screen profile is already at work, translating the pixels to accomodate the vagueries of the graphics card and monitor. And, if I so wished, I could send the unprofiled scan file to the printer and the printer profile would translate the yucky, uncorrected colours to the printer; thus ensuring that the yucky colours on screen are reproduced faithfully on paper.

Once the file is in Photoshop, only then would I assign the appropriate profile for the film to the file, followed by converting the result to the working space profile, usually ProPhoto RGB.
DJ wrote:Basically, each device has a different concept of colour, 100% red on a scanner may not be 100% red on a printer, and these values vary from model to model, and unit to unit. ICC profiles speak actual real colours. An ICC profile will map what a device thinks of as colour, so when a scanner says 100% red, you can look in the ICC profile and find out what the scanner thinks is 100% red, and map it to the nearest colour that the printer can produce.

In practice, you translate between these devices through a common denominator, your working space, ICC profiles use CiE LAB as the conduit between these colour spaces, as CiE LAB values can be attributed to a real world colour. As long as you translate colour in ( scanner/camera profiles ) and translate colour out ( monitor/printer profiles ) you have some assurance that your colours have been managed from end to end, and should correspond to the correct colour, or at least as well as your devices can reproduce it. Miss out a stage and you're effectively making large assumptions as to what those colour percentages mean.
Agreed, it is important to ensure that the colour management is consistent throughout the workflow. I would just disagree that the profiles necessarily translate the values from the input device directly to the output device; As I mentioned, the scanner profile can be "optional", but the printer profile will still translate the unprofiled colours to the printer, just as the screen profile will faithfully display those colours to the screen.
DJ wrote:All devices speak a different language when talking colour, and ICC profiles are the Babelfish :wink:
That is a very good explanation.
DJ wrote:On an equipment note, I've tried most of them. I favour the Gretag Macbeth/X-Rite stuff over all others. Colorvision stuff (Spyder etc) I wouldn't touch with a barge-pole. I have Monaco EZ-Colour that came with my V750, and hated it. Personally I would go for an EyeOne system, i1Match is good software and can do monitors and scanners and printers. Have a look at the packages available.
I have no experience of the Spyder monitor stuff but I do use the Spyder3Print with good results; there is even a 729 patch "expert" target which can be enhanced with a further 238 patch extended gray target. The process is tedious but the results seem to be very good.
Reassure yourself - stroke an Ebony

DJ
Site Admin
Posts: 360
Joined: Thu Sep 07, 2006 1:48 pm Etc/GMT-1+01:00
Location: Norfolk

Re: basic scanner questions

Post by DJ » Wed Aug 26, 2009 12:57 am Etc/GMT-1+01:00

Joanna Carter wrote:Sorry DJ, but from real world experience, with real targets, it does make a difference. Also, see Tim Parkin's blog for his experimentations.
Ah! The old subjective "Real World" gambit :wink: Last time I checked, Norfolk was still part of the Real World ( arguable at times I'll grant you ), and I'll bet my targets are just as real as anyone elses!

Honestly, it really should make very little difference, if you're finding it does, I would be suspicious, maybe of the accuracy of your target's datafiles. There will be the slightest of variances, in the whitepoint if anywhere. I believe the whole separate target thing originated because of Kodachrome, which has such a different emulsion to the E6 transparency films. If I were scanning a whole bunch of Kodachrome, I might consider getting a Kodachrome target, but otherwise, just pick the widest gamut E6 tranny film target and it should be fine for the rest. Remember, it's the scanner's response you're measuring, not the film's! The film has already been measured with the spectro and that's what's in your datafile. The actual target data that was sent to the film and how the film recorded it, is at this point a complete irrelevance, you don't even have that data, you're only interested in what's actually on it, as decreed by the measurement data.

If, and I do mean IF the target was a very well thought out custom target designed specifically to explore the full gamut of a particular emulsion's colorspace, then I would imagine that would make for a noticeably smoother profile, and the closest to this is Don Hutcheson's HCT target. The standard IT8 targets are not, and were never even designed for scanner calibration in the first place!

Don't take my word for it, the industry experts say much the same. The CMS "bible" Real World Color Management, Second Edition - Bruce Fraser, Chris Murphy & Fred Bunting on page 150, and also Color Management for Photographers - Andrew Rodney on page 145.

http://www.amazon.co.uk/Real-World-Colo ... 904&sr=8-1
http://www.amazon.co.uk/Color-Managemen ... y_b_text_b

I've read Tim's blog about his profiling experiments, and was quite surprised to see such a discrepency between the profiles, I really don't think you should be seeing that. I downloaded and tried to inspect the gamut on Tim's ICC profiles, but they're not behaving as they should, I believe this is an artefact of Monaco EZ-Color's profiles. I also have Monaco EZ-Color and used it once, and the profile I created with it exhibits the same problems, the gamut map is not viewable or is all over the place. Worrying. Tim seems like a man after my own heart, he likes to tinker with these things and get them working right, I'll drop him a line and create him a couple of profiles with my software if he still has his data kicking around.
Joanna Carter wrote:Once again :wink: how ever you describe the means of recording the colours, the only thing that remains constant is the record of the colour percentages
Now you got it, percentages, not actual colours :D
Joanna Carter wrote:and this record is independent of the device from which, or to which it is input/output. Yes, input or working space profiles may become part of that file, as well, but it is possible to store a scanned file with no profile information and no corrections made to allow for the inadequacies of the input device. When I scan a piece of film, it arrives in Photoshop with no profile at all. Now, the screen profile is already at work, translating the pixels to accomodate the vagueries of the graphics card and monitor. And, if I so wished, I could send the unprofiled scan file to the printer and the printer profile would translate the yucky, uncorrected colours to the printer; thus ensuring that the yucky colours on screen are reproduced faithfully on paper. Once the file is in Photoshop, only then would I assign the appropriate profile for the film to the file, followed by converting the result to the working space profile, usually ProPhoto RGB.
Yep, pretty standard workflow, mine is much the same.
DJ wrote:Agreed, it is important to ensure that the colour management is consistent throughout the workflow. I would just disagree that the profiles necessarily translate the values from the input device directly to the output device
They don't! I was merely summarising the whole process :D Technically, you can get profiles which do this, they're called "Device Link" profiles, but you'll not likely encounter one. The ICC profiles themselves are merely conduits between spaces, they use (mostly) CiE LAB space as the "connection" space. An input profile will contain a map from the device's colour space to CiE LAB space. And output profile will contain a map from CiE LAB space to the device's colour space. Your working space will have CiE LAB inputs and outputs.
Joanna Carter wrote:As I mentioned, the scanner profile can be "optional", but the printer profile will still translate the unprofiled colours to the printer, just as the screen profile will faithfully display those colours to the screen.
Must be some new definition of "faithful" I'm not aware of :wink: Certainly there's no requirement to have an ICC profile embedded in an image file, but without an association with one at some point then the values contained within that image file have no point of reference, and any output will be anything but faithful.
Joanna Carter wrote:I have no experience of the Spyder monitor stuff but I do use the Spyder3Print with good results; there is even a 729 patch "expert" target which can be enhanced with a further 238 patch extended gray target. The process is tedious but the results seem to be very good.
I've had bad experiences with Colorvision and their kit and software, I'm sure they've improved in the many years since I used them, but I find their products to be inferior to the other vendors, and I just don't trust them. The Spyder3Print device is not a spectrophotometer but a spectrocolorimeter, which is not the same thing, I can't be sure of it's capabilities. The "standard" target as I think of it, and on most of the printer profiling software I've seen has 918 patches, so that Colorvision deem the 729 patch one the "expert" version, worries me slightly.

At the end of the day, you're not going to do any harm by having lots of targets for different emulsions, but excepting specific circumstances, there's very little benefit either. If you've the money and want them, go buy!

Joanna Carter
Founder
Posts: 1283
Joined: Wed Mar 29, 2006 7:26 pm Etc/GMT-1+01:00
Workshop Images: http://grandes-images.com/fr/Paysages/P ... _2009.html
Location: Plestin-les-Grèves, France
Contact:

Re: basic scanner questions

Post by Joanna Carter » Wed Aug 26, 2009 8:56 am Etc/GMT-1+01:00

DJ wrote:Ah! The old subjective "Real World" gambit :wink: Last time I checked, Norfolk was still part of the Real World ( arguable at times I'll grant you ), and I'll bet my targets are just as real as anyone elses!
It's a bit like the end quote I see on messages from a contributor on one of my software development groups: "I want to live in Theory, everything works there" :lol:
DJ wrote:Honestly, it really should make very little difference, if you're finding it does, I would be suspicious, maybe of the accuracy of your target's datafiles ... Remember, it's the scanner's response you're measuring, not the film's! The film has already been measured with the spectro and that's what's in your datafile. The actual target data that was sent to the film and how the film recorded it, is at this point a complete irrelevance, you don't even have that data, you're only interested in what's actually on it, as decreed by the measurement data.

If, and I do mean IF the target was a very well thought out custom target designed specifically to explore the full gamut of a particular emulsion's colorspace, then I would imagine that would make for a noticeably smoother profile, and the closest to this is Don Hutcheson's HCT target.
Well, I have no reason to doubt the effort behind Wolf Faust's targets, and they are hand-measured for the particular colour dyes, which, according to Tim and others, is relevant to how the light is transmitted through the film. Don't forget, in profiling, I don't want to eliminate the differences in tonality and saturation between, say, Astia and Velvia, I want to scan colours that are true to the way a particular emulsion renders them, not simply "true" to the "colour numbers"

I use Wolf Faust's targets, not because they are scientifically correct but because they give me superbly good scans that are true to the "feel" of the film I am scanning. What's more, there's a slight price difference between them an the Hutch targets; like $75 instead of $480; that's a lot of film!
DJ wrote:I downloaded and tried to inspect the gamut on Tim's ICC profiles, but they're not behaving as they should, I believe this is an artefact of Monaco EZ-Color's profiles. I also have Monaco EZ-Color and used it once, and the profile I created with it exhibits the same problems, the gamut map is not viewable or is all over the place. Worrying.
DJ, you'll have to get together with one of us and see what a difference the Faust targets make; take off your scientific eyes and simply accept that they work :wink: :roll:
DJ wrote:Tim seems like a man after my own heart, he likes to tinker with these things and get them working right, I'll drop him a line and create him a couple of profiles with my software if he still has his data kicking around.
So, what software are you using?
DJ wrote:Now you got it, percentages, not actual colours :D
Well, I knew what I meant 8)
DJ wrote:
Joanna Carter wrote:As I mentioned, the scanner profile can be "optional", but the printer profile will still translate the unprofiled colours to the printer, just as the screen profile will faithfully display those colours to the screen.
Must be some new definition of "faithful" I'm not aware of :wink: Certainly there's no requirement to have an ICC profile embedded in an image file, but without an association with one at some point then the values contained within that image file have no point of reference, and any output will be anything but faithful.
Precisely.
DJ wrote:I've had bad experiences with Colorvision and their kit and software, I'm sure they've improved in the many years since I used them, but I find their products to be inferior to the other vendors, and I just don't trust them. The Spyder3Print device is not a spectrophotometer but a spectrocolorimeter, which is not the same thing, I can't be sure of it's capabilities. The "standard" target as I think of it, and on most of the printer profiling software I've seen has 918 patches, so that Colorvision deem the 729 patch one the "expert" version, worries me slightly.
As with the Faust targets, for the level of sales that I am generating from my pictures, cost has to be a consideration; and the ColorVision kit does a very acceptable job for an acceptable price. Most people viewing my prints would not have a clue if I had spent pennies or pounds on profiling; they simply want a picture that looks right, at the right price. What I want is a means of easily giving me the results that people want to buy. :D
Reassure yourself - stroke an Ebony

Patrick Dixon
Forum Hero
Posts: 252
Joined: Thu Apr 09, 2009 4:20 pm Etc/GMT-1+01:00
Location: Yate

Re: basic scanner questions

Post by Patrick Dixon » Wed Aug 26, 2009 9:22 am Etc/GMT-1+01:00

DJ wrote: Remember, it's the scanner's response you're measuring, not the film's!
I think what you are measuring is the combination of the film and the scanner.

The light spectrum of the scanner combines with the spectral response of the film and the spectral response of the scanner's sensor. Combinations of peaks and troughs in each of these will interact and can produce unexpected results. My in-laws once had a carpet that looked a completely different colour under incandescent light and daylight - presumably because it had two spectral peaks each of which was activated by a different colour temperature light.

DJ
Site Admin
Posts: 360
Joined: Thu Sep 07, 2006 1:48 pm Etc/GMT-1+01:00
Location: Norfolk

Re: basic scanner questions

Post by DJ » Wed Aug 26, 2009 2:51 pm Etc/GMT-1+01:00

Joanna Carter wrote:It's a bit like the end quote I see on messages from a contributor on one of my software development groups: "I want to live in Theory, everything works there" :lol:
I like that! But then I'm slightly idealistic and stubborn, I keep hitting it till it does work :wink:
Joanna Carter wrote:Well, I have no reason to doubt the effort behind Wolf Faust's targets, and they are hand-measured for the particular colour dyes
They're just targets, each one individually measured. They're not formulated for individual dyes or anything, at best they may be created with an output profile, which would help to get the values on the emulsion as close to the intended value as the film will allow, but they're just standard IT8 produced on different films. What's on the target is not so relevant, the value is that they're accurately measured.

Don't get me wrong, I'm not having a dig at Wolf Faust, I think his targets are very good ( I have one! ) and he's not doing anything different from Kodak or Fuji with their targets, and he's a good deal cheaper than they are! I would buy his targets in preference to all others except Don Hutcheson's. If Wolf were to dig a little deeper and maybe design a new target specifically for profiling scanners, he could be sitting on a gold-mine, he has all the equipment to do it.
Joanna Carter wrote:which, according to Tim and others, is relevant to how the light is transmitted through the film
Tim talks of Metamerism, which is the phenomenon where a colour can appear different under differing light sources, however, we're not dealing with differing light sources ( unless you're planning on changing the scanner bulb after every scan! ), we're dealing with one light source, one scanner. Interestingly, in Tim's two profiles of his V and N targets, the white point is absolutely identical, the bases of those two films are the same.
Joanna Carter wrote:Don't forget, in profiling, I don't want to eliminate the differences in tonality and saturation between, say, Astia and Velvia, I want to scan colours that are true to the way a particular emulsion renders them, not simply "true" to the "colour numbers"
I think we may have hit upon the problem. If I'm reading that right, you're under the impression that the ICC profiling of your scanner is intended to change the colours or perhaps in some way homogenise the colours or casts in the source material? or perhaps munge them down into some common denominator of "true" colour? That's really not how it works. Remember, we're profiling the scanner, once we've worked out the casts and range of the scanner, it can see whatever is in front of it. Your scanner doesn't suddenly develop a magenta cast when you put Velvia over it, or a cyan cast when you put Astia over it, it doesn't know or care what you put in front of it. Think of it as simply removing the tinted spectacles of your scanner so it can see clearly.

Your scanner ICC profile really contains no information from your target, it's used to generate a map of the capabilities of the scanner and then discarded. It never had the information of what emulsion is used, it never had the original target colour values either, it has two pieces of information, what the scanner sees, and the spectrophotometer reading.

A good scanner profile will see what's in front of it, whatever is in front of it. If you put Velvia in front of it, it will see Velvia, Astia and it will see Astia, and all the subtle differences between them. Scanner ICC profiles will remove the casts and neutralise the scanner, not the source material.
Joanna Carter wrote:I use Wolf Faust's targets, not because they are scientifically correct but because they give me superbly good scans that are true to the "feel" of the film I am scanning.
Try not to think of targets as "correct", none of them are correct, they just are what they are, but as long as we have accurate measurements of them we can use them to profile. The quality of that profile may be affected by how well that particular target tests the gamut of the device, i.e. a very limited film may make a limited profile, but I believe that would manifest in clipping or posterisation, not an colour cast. This is why it's better to pick one that has as wide a gamut as you can.
Joanna Carter wrote:What's more, there's a slight price difference between them an the Hutch targets; like $75 instead of $480; that's a lot of film!
Can't argue with that! the prices of these things are horrendous.
Joanna Carter wrote:DJ, you'll have to get together with one of us and see what a difference the Faust targets make; take off your scientific eyes and simply accept that they work :wink: :roll:
Oh I don't doubt that they work, I have used them myself, I just don't believe that you need a separate profile for each emulsion, a good profile will allow you to see all emulsions.
DJ wrote:So, what software are you using?
I use ProfileMaker Pro 5 for all my profiling, in the colour management market it's the equivalent of Adobe Photoshop. I've tried many different packages over the years, from ColorVision PrintFix (dreadful and thankfully discontinued), freeware solutions, Gretag Eye One Photo / i1Match (good), Monaco EZ-Color. Some were awful, some good. The main reason I didn't like Monaco EZ Color was because it used proprietary encoded data formats rather than the standard open format ones, which meant the target could only be used with their software.
Joanna Carter wrote:As with the Faust targets, for the level of sales that I am generating from my pictures, cost has to be a consideration; and the ColorVision kit does a very acceptable job for an acceptable price. Most people viewing my prints would not have a clue if I had spent pennies or pounds on profiling; they simply want a picture that looks right, at the right price. What I want is a means of easily giving me the results that people want to buy. :D
Oh I'm not advocating everybody goes out and spends thousands on colour management equipment! Mostly this type of kit is the sort you use rarely to set something up, and then don't touch it again, it's a heavy investment. Compared with using nothing at all, the Colorvision thing will do a bloody marvelous job! I also believe it's the best product they've come up with to date, and probably makes pretty good profiles, I just wouldn't choose it myself.

Patrick Dixon
Forum Hero
Posts: 252
Joined: Thu Apr 09, 2009 4:20 pm Etc/GMT-1+01:00
Location: Yate

Re: basic scanner questions

Post by Patrick Dixon » Wed Aug 26, 2009 3:49 pm Etc/GMT-1+01:00

DJ wrote: Try not to think of targets as "correct", none of them are correct, they just are what they are, but as long as we have accurate measurements of them we can use them to profile.
Yes, but the 'light source' used to make those accurate measurements, is spectrally different to the light source in the scanner, and different films may respond to the scanner light source differently.

Maybe you are overlooking that a scanner has both a light source and a detection device, and both will introduce their own character.

DJ
Site Admin
Posts: 360
Joined: Thu Sep 07, 2006 1:48 pm Etc/GMT-1+01:00
Location: Norfolk

Re: basic scanner questions

Post by DJ » Wed Aug 26, 2009 5:14 pm Etc/GMT-1+01:00

Patrick Dixon wrote:Yes, but the 'light source' used to make those accurate measurements, is spectrally different to the light source in the scanner
The "light source" for the measurements is not just any light source, it's a spectrophotometer. It is a device for measuring photons, it has a calibrated detector and light source, and measures light in the full visible spectrum and often beyond, in the case of the most popular unit, the Eye One Pro, from 380-730 nanometers in 10nm steps. With a full spectral reading, you can predict how the medium will react for any light source.
Patrick Dixon wrote:and different films may respond to the scanner light source differently.
They will! and we want them to, otherwise they'd all look the same, and there'd be no point in buying one film over another! The only thing we want to behave the same way every time is the scanner...
Patrick Dixon wrote:Maybe you are overlooking that a scanner has both a light source and a detection device, and both will introduce their own character.
Not at all, it is exactly that character, of both light source and detector, that we are trying to profile, just not what is in-between them... :wink: If we can map the character of the scanner's light source and detector, we can put what we like between them and it will see it correctly.

User avatar
Thingy
Forum Hero
Posts: 578
Joined: Wed Aug 20, 2008 11:13 pm Etc/GMT-1+01:00
Location: Orpington, Kent

Re: basic scanner questions

Post by Thingy » Tue Sep 01, 2009 12:03 pm Etc/GMT-1+01:00

EEK! :? :lol:
Love is an Ebony mounted with a Cooke PS945.......

timparkin
Forum Hero
Posts: 472
Joined: Tue Aug 21, 2007 1:40 pm Etc/GMT-1+01:00
Location: Leeds, West Yorkshire
Contact:

Re: basic scanner questions

Post by timparkin » Mon Dec 07, 2009 6:56 pm Etc/GMT-1+01:00

I don't think I saw all of these posts go by at the time but I'll try and post what I ended up finding out..

to summarise. the colour orange on a film may be made up of a single band pass filter at exactly the frequence of orange light or it may be made up of two band pass filters at red and yellow. The result is the same when the source is a perfect white light.

If the source just happens to have a spike at the exact red frequency and a dip at the exact yellow frequence, then any oranges will end up with a red cast.

Hence, depending on what dyes are used in a film stock, a non perfect white light source can create colours that are different than a perfect white light source.

I *think* this is the reason why we see a magenta boost when we scan a velvia slide using a profile that wasn't made using a velvia target (even though with a perfect white light source the velvia and non-velvia targets will measure as identical).

Hope that helps..

Tim
Waiting for the developing bill - 2 hours (and it's so small now!)

User avatar
Thingy
Forum Hero
Posts: 578
Joined: Wed Aug 20, 2008 11:13 pm Etc/GMT-1+01:00
Location: Orpington, Kent

Re: basic scanner questions

Post by Thingy » Thu Dec 10, 2009 4:42 pm Etc/GMT-1+01:00

Well, I have finally finished scanning in my photos from the Malton meeting (some dozen images in Acros) and my trip to Norway in July (35 images variously taken using Acros, Velvias 50 & 100 and Pro 160S). The Ones of Spitsbergen (Svalbald) looks superb.... I wanted to scan in the 5x4 images so they could be printed at roughly A3 size, so scanned them in a TIFs with 24 bit colour or 16 bit grey gamut, resolution at 600ppi and size at 398% plus a set of lower resolution Jpegs to post on this site later with a 300ppi resolution at 200% size. As I won't have a printer for well over a year, I hope they are adequate? :mrgreen:

My next request is, how do I apply the Wolf Faust profiles? I have scanned in both the one for Velvia 50 & Velvia 100 and saved the file. Can I apply these in Lightroom 2 or using the software supplied with the Epson V750..... :idea: ...or do I need to purchase Photoshop CS4? :'(
Love is an Ebony mounted with a Cooke PS945.......

Post Reply