Matching Scans And Reality

[mke]

I have scanned many of my ink sample pages - but the scans are all off a little bit. With changing contrasts, I get nearer - but still - the scans are a little bit off.

How to do color matching correctly?

 

Any advice?

[weltyj]

It may be that the scan is just fine, but your monitor/screen is not correctly displaying colors. If you search for "monitor color calibration" you will find some online sites to try to help you adjust your display via the underlying operating system.

 

Try looking at the scan image on an number of devices and see how the individual displays represent the colors somewhat differently. It can be frightening.

 

If you are simply trying to have all the scans conveniently on a tablet or cell phone screen, you may want to adjust the scans to make them look correct on a specific screen. That will require better image editing software than just a simple image viewing type of application. For windows/linux I can recommend GIMP. For anything else I don't know. You are probably going to want to edit the separate red, green and blue values of each image separately.

 

Should you go down this whole route, be sure to make a backup of your original scanned images!

 

Hope that helps.

[A Smug Dill]

http://www.fountainpennetwork.com/forum/index.php/topic/266271-presenting-your-ink-with-realistic-colours-counter-the-differences-between-paper-and-monitor/

 

https://www.fountainpennetwork.com/forum/topic/339106-inky-question-for-ef-nib-users/?p=4116791

 

https://www.fountainpennetwork.com/forum/topic/339106-inky-question-for-ef-nib-users/?p=4116898

Edited September 23, 2020 by A Smug Dill

[Mercian]

My advice would be to NOT do ANY ‘colour correction’ on your photos and/or scans.

 

Why?
Because everyone viewing this website has different OS & browser software, and is using different screen hardware.

If anyone manipulates an image they have made in order to get it to appear ‘true’ on their own equipment, that manipulation will only make it harder fo other viewers to ‘correct’ the image so that it appears ‘true’ on the viewers’ equipment.

So, how is anyone supposed to make images in such a manner that Other People who are looking at them via the internet can get the image to look ‘true’?

 

I suggest the following two things:

1) always use the same ‘light’ when making images.
For photographs one could e.g. always close all the curtains in one’s room and enable the flash on one’s camera.

When making scans one could make sure to always use the scanner’s default settings (for ‘colour photograph’ images).
If you can specify those settings when uploading one’s image (e.g. “I scanned this using an Acme 5000 scanner that was set to scan a ‘colour photograph’ at 600dpi”, or “Photo taken with flash on an iPhone 7”) it will assist those of your viewers who have pertinent technical knowledge in adjusting their own equipment to match the profiles of the equipment that the image-maker used.
I don’t suggest that anyone takes photographs in daylight, because its quality and ‘colour temperature’ will vary according to your latitude, the season, the weather conditions, and the size of the windows in your room and your distance to them.

2) include in the scan/photograph a ‘standard’ ‘target’ that viewers can use as a reference in order to calibrate the appearance of the image on their own equipment.
Some people who are reading this have got commercially-produced ‘calibration targets’, and these are (obviously) superb tools to include for this purpose.

For those of us who do not have access to such equipment, I suggest that we attempt to include items whose colour is universally recognisable throughout the world.

 

I would suggest the soft plastic labels that are wrapped around 500ml bottles of Coca-Cola, 7-Up, and Pepsi Cola.

They are (I think) lightfast, and their colours are the same in every market. I also think that they are available all over the world.

Why these three products in particular? Because their labels are, respectively, shades of Red, Green, and Blue.

Ok, they are not ‘true’ Red, Green, and Blue as defined by screen manufacturers or by physicists, but they will at least provide ‘standard’ shades of Red, Blue, and Green that nearly everyone ought to be able to recognise, and be able to adjust the settings of their own screen until these three items (& therefore every other thing in the image) appear as-close-as-possible to their true colours.

Caveat lector:

I am not a professional photographer, artist, screen manufacturer, or Physicist, and nor am I a manufacturer, programmer, or even a vendor of I.T. software or hardware. I’m just a guy who is making a partially-educated guess. My idea may therefore be utterly stupid and entirely worthless!

[BaronWulfraed]

I have scanned many of my ink sample pages - but the scans are all off a little bit. With changing contrasts, I get nearer - but still - the scans are a little bit off.

How to do color matching correctly?

 

Any advice?

 

"Changing contrasts" is probably the last thing you want to do.

 

If you are intent on accurate reproduction, you need to calibrate both the monitor and the scanner so that /they/ correct for imbalances in the lighting.

 

Unfortunately, Monaco EZColor was discontinued decades ago. It provided functions for calibrating printer, scanner, and (rudimentary -- unless one purchased the $$$ hardware unit) monitor. One would print a special target file (note: one had to develop profiles for EACH paper used on the printer), attach a provided film target below it on the paper, then scan the print/film together. The software would first calibrate the scanner using the film image (since it "knew" what each color square was supposed to be), and then with that correction table it would develop a profile the ink/paper printed version.

 

Monitors were usually a case of first setting the white point. Many monitors ship with an unset (~9500degK) white point (which is a rather bluish and bright setting). Photograde monitors are commonly set to something like Daylight (around 5500degK), "Cloudy" (around 6500degK -- yes, cloudy days have more blue), or Tungsten (~3400degK, red-orange). Then one has to adjust the gamma curve -- Software presents red, green, blue squares in which part of the square is middle-shade (Monaco used a large M), and the rest of the square is alternating lines of full color and black. While squinting at the monitor from a distance, one adjusted the drive so that the alternating color/black blends into the mid-tone.

 

 

Having harangued you'all with that. Instead of CONTRAST adjust you want to work with (In Photoshop: LEVELS layer). Use the middle eyedropper (grey -- actually for white balance), locate a portion of the scan that is supposed to be neutral (any true grey other than pure white and pure black) and click on it. That will shift the color balance to remove any tint bias from the scan. The next two steps would be to use the white eyedropper (highlight) and click on something that is supposed to be full white, and the black eyedropper (shadow) on the darkest shadow. Those two will adjust the spread of brightness levels to the full span (hence, will tweak contrast). The last steps would be to move the gamma slider -- which will shift the mid-range brightness, and then tweak the output levels (at least one guide suggests setting output to 5 (blacks) and 250 (whites) -- which would mean you do not have full black and whites in the output/saved image; apparently they are considered "ugly" in prints, 250 means the whites get a very light speckling of dark, ...)

 

[grmbrk]

Jetpens.com scans are very close to what I get on paper.

[sansenri]

My advice would be to NOT do ANY ‘colour correction’ on your photos and/or scans.

 

Why?

Because everyone viewing this website has different OS & browser software, and is using different screen hardware.

If anyone manipulates an image they have made in order to get it to appear ‘true’ on their own equipment, that manipulation will only make it harder fo other viewers to ‘correct’ the image so that it appears ‘true’ on the viewers’ equipment.

 

So, how is anyone supposed to make images in such a manner that Other People who are looking at them via the internet can get the image to look ‘true’?

 

I suggest the following two things:

1) always use the same ‘light’ when making images.

For photographs one could e.g. always close all the curtains in one’s room and enable the flash on one’s camera.

When making scans one could make sure to always use the scanner’s default settings (for ‘colour photograph’ images).

If you can specify those settings when uploading one’s image (e.g. “I scanned this using an Acme 5000 scanner that was set to scan a ‘colour photograph’ at 600dpi”, or “Photo taken with flash on an iPhone 7”) it will assist those of your viewers who have pertinent technical knowledge in adjusting their own equipment to match the profiles of the equipment that the image-maker used.

I don’t suggest that anyone takes photographs in daylight, because its quality and ‘colour temperature’ will vary according to your latitude, the season, the weather conditions, and the size of the windows in your room and your distance to them.

 

2) include in the scan/photograph a ‘standard’ ‘target’ that viewers can use as a reference in order to calibrate the appearance of the image on their own equipment.

Some people who are reading this have got commercially-produced ‘calibration targets’, and these are (obviously) superb tools to include for this purpose.

For those of us who do not have access to such equipment, I suggest that we attempt to include items whose colour is universally recognisable throughout the world.

 

I would suggest the soft plastic labels that are wrapped around 500ml bottles of Coca-Cola, 7-Up, and Pepsi Cola.

They are (I think) lightfast, and their colours are the same in every market. I also think that they are available all over the world.

Why these three products in particular? Because their labels are, respectively, shades of Red, Green, and Blue.

Ok, they are not ‘true’ Red, Green, and Blue as defined by screen manufacturers or by physicists, but they will at least provide ‘standard’ shades of Red, Blue, and Green that nearly everyone ought to be able to recognise, and be able to adjust the settings of their own screen until these three items (& therefore every other thing in the image) appear as-close-as-possible to their true colours.

 

Caveat lector:

I am not a professional photographer, artist, screen manufacturer, or Physicist, and nor am I a manufacturer, programmer, or even a vendor of I.T. software or hardware. I’m just a guy who is making a partially-educated guess. My idea may therefore be utterly stupid and entirely worthless!

I agree with this.

It's mostly a lost battle.

There are other factors that play against the possibility of correctly showing colours on a web site like FPN (unless we all buy the same hardware, software, and calibration systems...) and that is the fact that every hardware and every software we use generate different results.

And electronic equipment does not have (unless you use extremely expensive equipment, and even then) the same capacity as our eyes to distinguish colour shades.

It's typical, in a comparison of blue inks on screen they all look alike, when in real life they are dramatically different...

I much prefer to rely on the reviewer's comments and to his description of the colour which is often more accurate than the photo...

Still, it's nice to see those photos, they give a guide, so don't be put off, we'd be happy to see yours, and hear your descriptions!

[Chouffleur]

Jetpens.com scans are very close to what I get on paper.

 

This is useful information. Often I think a an ink looks attractive based upon an FPN review only to discover that the color I liked isn't what comes out of the bottle when it arrives.

 

It would be nice to know how you all rank the various purveyors of ink for their color accuracy (Jetpens, Goulet, Vanness, etc.)

 

Once I establish which one works the best for me I could - Read the FPN review - Decide if I wanted to buy or sample the ink - Double-check with the trusted purveyor image - *then* buy the ink.

Edited September 23, 2020 by Chouffleur

[weltyj]

 

This is useful information. Often I think a an ink looks attractive based upon an FPN review only to discover that the color I liked isn't what comes out of the bottle when it arrives.

 

It would be nice to know how you all rank the various purveyors of ink for their color accuracy (Jetpens, Goulet, Vanness, etc.)

 

Once I establish which one works the best for me I could - Read the FPN review - Decide if I wanted to buy or sample the ink - Double-check with the trusted purveyor image - *then* buy the ink.

This is not really possible, @grmbrk can only tell you the jetpens looks ok on @grmbk's device. Your device, if different in hardare, OS or display software will have a different color representation of *exactly* the same images as @grmbrk is talking about. If you had exactly the same device, operating system and display software as @grmbrk, then there would be a good chance you'd see the same color as @grmbrk.

[Chouffleur]

This is not really possible, @grmbrk can only tell you the jetpens looks ok on @grmbk's device. Your device, if different in hardare, OS or display software will have a different color representation of *exactly* the same images as @grmbrk is talking about. If you had exactly the same device, operating system and display software as @grmbrk, then there would be a good chance you'd see the same color as @grmbrk.

But even that knowledge - Jetpens + grmbrk = good is useful (as long as grmbrk notifies me when he/she/it buys a new monitor ;-))

[A Smug Dill]

My advice would be to NOT do ANY ‘colour correction’ on your photos and/or scans.

 

Why?

Because everyone viewing this website has different OS & browser software, and is using different screen hardware.

If anyone manipulates an image they have made in order to get it to appear ‘true’ on their own equipment, that manipulation will only make it harder fo other viewers to ‘correct’ the image so that it appears ‘true’ on the viewers’ equipment.

 

So, how is anyone supposed to make images in such a manner that Other People who are looking at them via the internet can get the image to look ‘true’?

 

Don't you think that is being somewhat presumptuous?

 

For all we know, @mke wants to digitise his ink sample pages for his own reference, so that he can review and/or search electronically for information he has compiled, without retrieving or even keeping the physical artefacts, when selecting an ink to fill one of his pens next. The question was framed from his point of view, so whatever post-processing that will make the digitised information most useful to him for future reference is supposedly the point, and not whether you or anyone else can accurately interpret the data in images that he may (or may not) choose to share and set your expectations accordingly.

 

If there are two (or more) solutions, and one of them will work better for @mke's purposes for his own information review or 'consumption' in the future, and the other(s) may be better for prospective readers who don't have physical access to the particular inks or @mke's sample sheets, don't you think that the 'right' thing to do here in answer to his question is to suggest what would help him the most, instead of 'advising' him not to go for such a solution because you want others in your position to benefit more?

[mke]

I really need to study the answers you provided - before I can answer.

I will provide, off course, my scans - but as I said, they don't match (exactly) the reality.

 

I will also try to take photos instead of scans.

 

I did my scans at 600 dpi with a Xerox DocuCentre.

[XYZZY]

My Dad used to do insane things with contrast and brightness, a habit he formed when looking at Fortran on a terminal, and later spreadsheets on a monochrome monitor. His settings actually yielded good results for viewing of monochrome information, but he was frustrated to no end when the photos he included with his genealogy stuff (one of his hobbies) never looked good on any display other than the computer in his office. I was never able to successfully help him, as that meant his spreadsheets didn't look right to him. Sigh.

 

Here's a quick result for color management books on Amazon, I'm sure there are many others:

 

Real World Color Management, by Bruce Fraser (don't know if it's current, but I used this Back In The Day)

Color Management & Quality Output: Tom Ashe

Understanding Color Management: Abhay Sharma

 

And most books on Adobe Lightroom or Photoshop will have sections devoted to color management even though the book isn't devoted to color management.

 

Color management is its own rabbit-hole of problems and solutions. For photo-to-print the problem is constrained because you have control over final output (the print). If your goal is web display for other people then your hope is to get accurate color into a JPG with proper color profile information, and then hope that the person viewing the images is using an operating system and browser that supports color profiles, and hope that they haven't done bad things to their display settings.

Edited September 24, 2020 by XYZZY

[Mercian]

Don't you think that is being somewhat presumptuous?

You are quite correct!

I presumed that mke’s intention was to make scans that he could share with us here, e.g. in an ink review - i.e. I read in to the OP something that was not actually said by mke

​Indeed, the context of the OP implies nothing of the sort, and indeed suggests that what you go on to add is in fact a far more likely interpretation of his intention.

Thank you for the reminder that I need to pay attention to what people actually write, rather than leaping-to (& then pursuing) my own subconscious/internal ideas

 

For all we know, @mke wants to digitise his ink sample pages for his own reference, so that he can review and/or search electronically for information he has compiled, without retrieving or even keeping the physical artefacts, when selecting an ink to fill one of his pens next. The question was framed from his point of view, so whatever post-processing that will make the digitised information most useful to him for future reference is supposedly the point, and not whether you or anyone else can accurately interpret the data in images that he may (or may not) choose to share and set your expectations accordingly.

 

If there are two (or more) solutions, and one of them will work better for @mke's purposes for his own information review or 'consumption' in the future, and the other(s) may be better for prospective readers who don't have physical access to the particular inks or @mke's sample sheets, don't you think that the 'right' thing to do here in answer to his question is to suggest what would help him the most, instead of 'advising' him not to go for such a solution because you want others in your position to benefit more?

The problem with trying to answer the question of how to make/manipulate digital images so that they will appear as much as possible like the object(s) being imaged is this:

it is NOT possible to make an image that is a 100% accurate representation of any object.

 

An image that is displayed by a screen is produced by the emission of light by a display screen, which light then enters one’s eye.

This is not the same physical process as ambient light bouncing off an object and then entering one’s eye. There will always be objectively-different qualities in the light that enters one’s eye, and therefore subjective differences in the qualia that get stimulated in one’s brain by these two different processes.

Looking at a printed image - whether on photographic film, photo-paper, or any other medium - also entails different physical processes than looking at an actual object. Metal, plastic, and wood (& flesh, fur, bone, vegetation, paper, clothing, etc) all reflect light differently to the way that photographic paper or projector screens do, and this is also different to how visual display units (CRTs, LCDs, or LEDs) produce light to form an image.

In short, an image of an object is NOT the object itself. And your brain can tell that it’s different.

 

With all that (pseudo-philosophical) abstract stuff out of the way, the initial question in the OP does still remain valid; what is the best way to try to make images for future reference?

 

If one is trying to create a reference library of images of ‘freshly-written’ ink (instead of keeping a collection of reference pieces; which may fade over time) I still think that one must try to use a method whose vulnerability to the tendency of technical specifications of (& standards employed by) viewing equipment to change over time is as low as is ‘reasonably’ possible.

Unless, that is, one already knows for certain that one is never again going to use any equipment or software to view the image(s) that one is making other than the equipment & software that one is using to create the images.

In my opinion, the least-unreliable way to do this is to include within one’s images items that can act as recognisable ‘standards’ for future comparative reference.

Those could be calibration targets, they could be products like the ones I mentioned before, or (perhaps best of all for use in a personal reference library of writing in different inks) they could be passages that one has written in a ‘standard’ reference ink alongside the text written in the ink whose appearance one wishes to record. Ideally with the same pen(s) that one used to write with the ink whose appearance one is imaging.

One would thus be provided with a way to judge the appearance of the sampled ink in comparison against a familiar reference.

The technical ‘shortcomings’ of (& any variations in) the image-capture and image-display processes will thus be at least partially negated.

One could perhaps use one’s ‘standard’ comparator ink to write down one’s naked-eye impressions of how the appearance of the ink whose appearance one has made the image to record varies against one’s naked-eye impression of the appearance of the ‘standard’ comparator ink. This would further negate any technical challenges in the processes of image creation/reproduction.

 

For example, if sampling blue inks I might personally use Waterman ‘Serenity Blue’ (the artist formerly known as ‘Florida Blue’) as a comparator and, if sampling black inks, Noodler’s Black - but when choosing one’s ‘standard’ comparator ink (or inks) the individual image maker ought to pick whatever ink(s) s/he is already very familiar with, or uses most-often.

 

In order to serve its purpose, the ‘standard’ comparator(s) must be one(s) that are well-known to the intended viewer(s) of the image.

 

TL;DR Summary:

When making images, try to include familiar comparators alongside the ‘target’ of one’s images.

Doing this provides a way to compensate for many of the scientific limitations (& the variations in image capture/display equipment) with which the process is fraught.

[weltyj]

 

TL;DR Summary:

When making images, try to include familiar comparators alongside the ‘target’ of one’s images.

Doing this provides a way to compensate for many of the scientific limitations (& the variations in image capture/display equipment) with which the process is fraught.

Many good points made Mercian. If I may add, if the familiar comparison item can be as "grey" as possible that would be best. If we were being really picky, a photographer's grey card is ideal. To do it best comparison item needs to be obtainable by the final viewer of the image, whether this is you or some other person.

 

Knowing the exact type of paper the ink swatch is made on can also be very helpful, though it's not gray, it is nearly white and should be very standardized -- paper manufacturers go to some length to guarantee the final whiteness of the paper, some even adding a touch of blue.

[mke]

I took photos from one sample today and found that they are much closer to reality than the scans.

 

Reality ~ photos taken with Pixel 4a > photos taken with Nexus 5x >> scan

 

So, it is the scanner which doesn't perform well.

Edited September 25, 2020 by mke

[BaronWulfraed]

I took photos from one sample today and found that they are much closer to reality than the scans.

 

Reality ~ photos taken with Pixel 4a > photos taken with Nexus 5x >> scan

 

So, it is the scanner which doesn't perform well.

Which may just mean you need to create a color profile for the scanner. Though doing that may not be easy without $$$ (the discontinued Monaco EZColor could do it, but even that was a $100-200 package). The scanning software should then apply the color profile to normalize the scanned images -- if it doesn't, you'll need a photoeditor that lets you specify the source profile, and an output/display profile.

http://www.ddisoftware.com/prism/help/steps.htm

 

Something else to consider is if all those devices are using sRGB or Adobe RGB (viewing Adobe RGB images on an sRGB display, without applying a transform, will result in somewhat muted colors).

[mke]

Thank you very much for all your recommendations. Lots of things to learn now. Until then, I just use pictures.

[Mercian]

Which may just mean you need to create a color profile for the scanner. Though doing that may not be easy without $$$ (the discontinued Monaco EZColor could do it, but even that was a $100-200 package).

I think that one can do this in GIMP.

GIMP is open-source and free to download.

 

But one does of course still need to invest one’s time in learning how to use it!

[BaronWulfraed]

I think that one can do this in GIMP.

GIMP is open-source and free to download.

 

But one does of course still need to invest one’s time in learning how to use it!

One would still have to obtain a calibrated IT8 scanner TARGET, even if GIMP can generate a profile from one.

https://www.filmscanner.info/en/Scannerkalibrierung.html