Harry Mendryk was moderator of the Yahoo Group ‘Digitizing Comics‘. The Group no longer exists, as Yahoo closed all their Groups in 2019.
Group Description
This was a group for discussions about using scanners and computers to save and restore comic book art. With the continuing deterioration of Gold and Silver Age comics, and the high cost of those comics, some are turning them into digital format. This list allowed amateurs to discuss what they were doing, exchange scanning and restoration techniques, request and receive advice, and develop a community of like minded individuals.
This deals with the following scanning topics:
- Golden Age printing methods
- Scanning Resolution (400 dpi vs 600 dpi)
- Digital Bleaching to extract line art
(a) Harry Mendryk’s method
(b) Kris Brownlow’s method
(c) David’s method - Digital Colour Correction
(a) Harry Mendryk’s method
(b) Rand’s method - Colour Correction : Conversion to CMYK alters colour
- Colour Correction : Yellow & Magenta – Edit as CMYK
- Colour Correction : Avoid the Red halo
- Colour Correction : Greys
- Colour Correction : Colour Noise
- Colour Correction : Limit Colour to 8 bit
- Resizing : Moire Patterns
- High Resolution scanning : Advantages
- LAB Color Mode
- Modern Reprints : Colour Techniques
- Note on other Methods : Destructive & Non-destructive
Background : Golden Age Printing
Darci (2007/09/04) [#147]:
Bob Rozakis said (reported from a magazine interview) (discussing professional comics printing at DC Comics) –
“Once all the art and colouring was done, the pages were sent to Chemical Color Plate in Bridgeport, CT, where the colour separations were done by painting acetates for each of the 25%, 50% and 100% screens of red, yellow, and blue. This changed with the advent of computerised colouring and separations.”
Q: You’ve doubtless seen the piecemeal auctioning of the fabled “Jack Adler Collection”. I have an approval cover (“Adventure Comics” #374) I received as a gift. How did he get hold of those?
A: From what I know, Jack Adler took the proofs home with his original colour guides, and now they’re being sold off. The proof was created at Chemical, using the separations they’d generated. If it was okayed, the film negatives were shipped out to Spartan Printing in Sparta, Illinois, for printing.”
Background : Chemical Bleaching
Harry Mendryk (2006/01/18) [#2]:
When I decided to try restoring line art, I was already comfortable working in Photoshop. I bought a HP scanner, and came up with a technique to “digitally bleach” colour scans.
But there are problems, due to the poor Golden Age printing techniques, yellowing of the comic paper with age, the limitations of the scanner, and the limitations of the technique itself.
Over the years I have become adept at squeezing the most out of this method. But there will always be problems, due to things like the inability to distinguish a black pixel made from the overlapping of CMY inks from a black pixel on the K plate.
Or the use of C ink under black in the old printing to improve its look. When you remove the C, the black channel suddenly has lots of little holes in it.
After many years of experimentation, research and thought, I have come to believe that it simply is not possible to digitally bleach a comic as well as can be done chemically.
Although digitially bleaching saves a lot of time compared to using Photoshop tools on a unbleached scan, it still requires a lot of effort to make a really nice line art restoration.
I spent years digitally bleaching the line arts for all the Simon and Kirby covers (something like 386 covers). I was quite pleased with the results. During that project, I showed what I was doing to Joe Simon. This led to frequent visits to Joe’s place. I learned a lot of the techniques Joe has used over the years, and still uses.
He showed me how to chemically bleach a comic. From what I understand,
he had also shown this to Greg Theakson. Greg apparently added his own processes to improve the results. Bleaching by Joe’s method pretty much removes the magenta (red) and yellow inks, but only partially affects the cyan (blue). When I tried it, I did not do as well with the blue.
But one time Greg showed me bleached pages he made for DC’s “Spirit” archives. I can attest that he does something different: his process truly left only the black ink.
Of course, with the poor techniques of the original comic book printing, even the bleached pages still needed touching up.
When I finished the S&K cover project, I wanted to do something with the actual S&K stories. But knowing how many S&K pages that was, and how much time was required to fix up digitially bleached pages, I knew that there was no way I was going to do that.
But affordable photo printers were now available, so I decided to work on colour restoration. To that end I have developed Photoshop methods to remove yellowing that the pages have undergone and to improve the often poor inking quality of the original printing.
Harry Mendryk (2006/02/01) [#89]:
There is no perfect bleaching process. Chemical Bleaching, when not faced with the horrible clay paper, produces the best results. But as an amateur it is not affordable for me, even with low grade comics.
I am not one of those who criticise chemical bleaching because of the loss of the comic. One comic is destroyed, but when the restoration is published more copies are created.
And how long will the original comics last? The paper is low grade and very acidic. I am amazed they have lasted as long as they have. I am sure in 100 years time all the Golden Age comics will be dust. Do libraries still keep original old newspapers anymore? Most have switched to microfilm.
There are chemical processes to remove the acid from newsprint. But the cheaper ones do more harm than good. The only process that libraries are willing to do is place the material in a sealed chamber with gas. But that is probably too expensive a process for comics?
Scanning Resolution (300 dpi vs 600 dpi)
Tom Kraft (2006/01/23) [#15]:
These settings were specified on the Kirby list for scanning original art (scanner settings):
– 100%
– 300 dpi
– RBG colour
– Scan front and back.
– No unsharp masking or auto adjust settings.
– Include space in between edges of paper and scanned image (don’t crop the scan to the edge of the paper, let us see the actual paper edge and some non-paper space).
– Save as JPG at 99% or “maximum”.
Does this group feel these specs are best for archiving a record of the original art? At 300 dpi you should be able to print the file and get very close to the same quality as the original.
Should it be 600 dpi (although the file size would be too large to e-mail)?
Randolph Hoppe (2006/01/23) [#16]:
600dpi would be better in the long term.
Harry Mendryk (2006/01/23) [#17]:
There is no completely correct answer to this question. It boils down to finding the best compromise for the intended use.
Let’s consider 300 vs 600 dpi based on e-mail capability. When Jack started penciling, the industry standard was to work twice-up [double the size of the printed page]. However sometime later (after Silver Age?) the industry switched to 1.5 times up.
Obviously, the paper size would affect the image size. Twice-up scanned images would be about 81 MB at 300 dpi, and 325 MB at 600 dpi. 1.5-up scanned images would be about 48 MB for 300 and 192 MB for 600. However, these are not the actual JPEG file sizes. With JPEG set to Maximum, there is still data compression, only it is a lossless type of compression (that is, when the file is de-compressed you get an image identical to the original). The amount of actual compression depends on the image. It is not unusual to see JPEG files compressed to 1/4 of the original size: with twice-up, files are 20 MB at 300 dpi; and 80 MB at 600 dpi. With 1.5-up, files will be 12 MB at 300 dpi and 48 MB at 600 dpi. Better results would frequently occur, but even 300 dpi images are too large to e-mail, so e-mail capacity cannot be used to decide scanning resolution.
So now let’s turn to printing the image. Here, much depends on how the image is to be printed. Let’s assume the image will be printed life size, but with the quality found in the better magazines. Those types of magazine use 150 lines/inch printing. LPI is not the same thing as DPI. When I started in computer graphics, I was told the rule is that DPI should be twice the LPI. Nowadays I hear that the rule should be 1.5. Using the x2 rule, magazine quality printing would require 300 dpi image resolution. If you use the x1.5 rule you need even less resolution. This suggests that 600 dpi would be overkill. Personally I really do not see the need to print original art better then a magazine’s quality.
One other possibility comes to mind. What if you wanted to print the original art in bitmap format. That is, convert the image to just black and white; no grey tones. This is effectively what was done originally, in making a stat from the original art, to be used to make the actual comic book. The experience I found when making the Simon & Kirby covers, is that with bitmap at 300 dpi I could barely see the little digital steps, at 600 dpi I could not. But keep in mind that my book was of covers at comic book size. But original art is larger than the comic book, and would be viewed from further away. I suspect then the small steps at 300 dpi would be unnoticeable.
So my suggestion is to remain with 300 dpi. The benefits for 600 dpi do not seem to be worth the larger file size.
Randolph Hoppe (2006/01/23) [#29]:
After a little web research that took me to some Museum/Archive websites, I’m as keen on 600 dpi as I was in my last post.
But I was wrong with my 99% jpg recommendation. Any lossy compression is to avoided when building a Museum-quality digital archive. For web-posting and e-mailing, a jpg would be preferable; but the non-lossy compression available in a TIFF file is preferable for a digital archive.
So I’d like to see:
– 100%
– 600 dpi
– RGB colour (24 bit)
– Scan front and back.
– No unsharp masking or auto adjust settings.
– Include space in between edges of paper and scanned image (don’t crop the scan to the edge of the paper, let us see the actual paper edge and some non-paper space).
– Save as TIFF with LZW compression.
– If you want to stitch the pieces together, go ahead, but send the pieces, as well as the result of your stitching, for safety’s sake.
Greg T [Greg Theakston] (2006/01/24) [#35]:
400 dpi is the industry standard. I used 600 dpi at Pure Imagination for a long time, but I found that 400 dpi works just fine, with fewer MBs eaten-up.
Nobody has mentioned the Median filter (Filter > Noise > Median), in greyscale, for use with line art. I find it indispensable. A fast cure for poorly printed lettering, and large areas of black which are breaking up.
Used with the marquee tool in Photoshop, the Median filter cuts my work time by at least 25%.
Consequently, I retouch most of my pages in greyscale, so that I can use the Median filter.
Harry Mendryk (2006/01/24) [#36]:
It’s true that whenever I supply Marvel with a file they request 400 dpi. And I do not believe there are many people, if any, that can actually see a pixel at that size.
For my restoration work, however, I do not find that 400 dpi cuts it. The lowest resolution that works for me is 600 dpi. Here it is not the case of the eye seeing, but of the ability of Photoshop tools to distinguish the comic book screen dots from the paper background. Having done my work at 600 dpi, and with CD disk writers available, I find no need to re-size it. Besides, re-sizing can create Moiré. My printer handles 600 dpi nicely.
When doing colour restorations, I do not use the Median filter. But when doing line art restoration I frequently do. A lot depends on the quality of the image. If it has a lot of noise (small black or white dots) I will use the Median filter with a setting that is a compromise between cleaning up the noise and the loss of details in the real line art. I then use the Pencil and Eraser tools to fix those dots that were too large for the Median filter to remove, and for re-sharpening those areas of the Line Art that were lost.
Marquee tools are useful to restrict my adjustments to a particular area. I also use the Magic Wand tool for the same purpose.
Matthew Moring [m.moring@comcast.net] (2006/01/24) [#38]:
400 dpi is what Marvel wants. However it certainly is not the industry standard. Every other company doing Golden Age reprints which I’ve done work for has been using 600 dpi for some time.
In this age of huge hard drives, the difference in file sizes is of minimal significance.
Harry Mendryk (2006/02/02) [#95]:
Low resolution not only makes digital bleaching more difficult, but makes the manual editing a problem. I prefer to work at 600 dpi.
Darci [darci386] (2007/03/08) [#143]:
Golden Age comics were probably printed at between 65 to 85 LPI (lines per inch). The general formula for dpi is 1.5 x to 2.0 x the LPI. As such, 150 dpi should be plenty for reproduction, unless you are going to scale up.
Someone mentioned that modern comics use higher LPI settings. However, I thought you might be more interested in Golden Age comics.
Harry Mendryk (2007/03/08) [#144]:
There are two problems with scanning Silver and Golden Age comics at 150 dpi. The first is, it is easy to encounter Moire problems. The second is the line art of comics. At 150 dpi the line art will develop easily seen pixel steps. The formula the fellow gives is an old one, developed when disk space was expensive. I would advise you to continue scanning at 300 or 600 dpi.
The eye begins to detect image deterioration below 300 dpi. High quality images require 400 dpi. 600 dpi is used for convenience, simply because most scanners can do that.
Further, the restoration techniques I’ve described work best when the scan resolution is much higher than the screening resolution used in the original printing process.
Harry Mendryk (2007/03/08) [#145]:
No sooner had I sent my last response, I remembered what the quoted formula was originally used for. It was meant only to calculate the scanning resolution required when scanning an unscreened image, such as a photograph, that will be screened for printing.
It does not apply when scanning printed images that are already screened.
NB: Screening is the technique used in printing to simulate tints or continuous-tone images (such as photographs) using dots. Almost all printing technologies – such as offset, gravure or inkjet printing – simulate shades of colours using dots. See the technical note, next.
AM Screening [Half Tone]
AM screening (Amplitude Modulation) uses a fixed linear dot pattern, of various sized dots, to emulate the tonal range in photographic images.
Standard AM line screens vary in resolution depending on the reproduction process and equipment quality. In commercial offset printing, these line screens are typically 100, 133, 150, 175 and 200 dots per linear inch.
The larger the dot, the darker the image area; and the smaller the dot, the lighter the image area.
Colour images use a separate AM screen for each of the primary printing colours: Cyan, Magenta, Yellow and Black (CMYK). These screens are printed on top of one another, which gives the range of colour we see on paper. The colour we see in a printed image is an illusion, caused because our eyes can only discern so much detail at a given distance. If we use high magnification to enlarge an area of a printed photo, the image becomes unrecognisable.
Problems With AM Screening –
a. Limited Minimum Dot Size: In printing, we are limited to a minimum dot size for ink to adhere and transfer back to the sheet of paper printed on. We’re also limited at the other end of the tonal spectrum, because we can only go so large with the dot before the printed area becomes a solid. This results in an inherent flaw in the process called posterisation, and we have to adjust the photographic image before printing to reduce the problems it creates on the press. When we make these adjustments, we are actually degrading the quality of the original image slightly; so we lose detail, colour, and contrast.
b. Size of the Dots: AM screening uses a fixed dot pattern, and the tonal range is achieved by varying the size of the dot within that fixed pattern. Printing presses can only print so small a dot, so are limited to a printing range between the smallest dot possible and the largest dot possible in achieving a tonal range. Thus the peak resolution in an AM Screen is set by the largest (coarsest) dot, not by the smallest one. For a 175 line screen, the smallest possible dot is approximately 10 microns, and the largest dot is approximately 200 microns.
c. Visible Patterns in the Image: Sometimes such patterns conflict with the actual subject matter of the photograph, so amplify the negative visual effects of the printing process. The human mind recognizes patterns easily, so anytime we incorporate a fixed pattern into the process we naturally detect that pattern. Colour images are built on a series of screens, printed over the top of one another, and these screens are turned at specific angles to reduce the inherent negative effects. The flaws are still present, such as moire patterns and rosette patterns, but it is possible to reduce their more obvious effects.
For all these reasons, customers demand FM screening instead, especially in the clothing industry, where subject matter is all about patterns, as there can be a plethora of adverse pattern conflicts from using AM screening.
NB: A detailed note then follows, regarding the benefits of FM screening (omitted, because comics printed in the period 1940-1980 didn’t use FM screening, as it has the drawback of being very expensive).
[Source: http://thedivision.co.uk/everything-need-know-print-screening/]
Digital Bleaching (Generate Line Art)
Bleaching is a chemical process applied to printed comics pages to remove the cyan (i.e. blue), magenta (i.e. red), and yellow inks, to leave only the black line art. Digital bleaching is a computer process which simulates chemical bleaching for digital images.
Harry Mendryk (2006/01/19) [#8]:
When I correct a Simon & Kirby cover, to colour-correct the cover I first go through a digital bleaching process, so that I have line art which exactly matches the colour plates used.
I work in Photoshop 5 and (for some features) Photoshop 7.
Step 1 (CMYK colour setup) –
Harry Mendryk (2006/01/31) [#65]:
Digital Bleaching is not as effective as a Chemical Bleach. After you Digitally Bleach an image you will have to spend a lot of time editing the image to get it correct.
But if you are willing to spend that time, you can get really nice results without destroying the original comic.
There are even cases where Chemical Bleaching will not work. This is so with the Joe Simon cover for “Silver Streak Comic” #2. That cover was not printed with a Black plate: instead, the fourth plate was for a special silver ink used in the title.
The black on the cover is actually caused by overlapping Cyan, Magenta and Yellow. If you were to Chemically Bleach this cover all the line art would disappear.
Like my colour restoration technique, my Digital Bleaching technique requires the correct CMYK colour setup. In Photoshop 7, go to menu item: Edit > Color Settings. In ‘Working Spaces’ click on ‘CMYK’, then select (from the list of options): “Custom CMYK”.
When the CMYK dialog appears, in “Separation Options” select ‘GCR’, in “Black Generation” select ‘Maximum’, then click “OK” twice.
Step 2 (Level Adjustment) –
Harry Mendryk (2006/01/31) [#66]:
Digital Bleaching begins in the same way as Colour Correction, using the Level tool (Image > Adjustments > Levels). The purpose of this is to adjust for strong Black, and for the paper to become near White.
The dialog box settings I used for each colour channel are:
Adjustment Channel Input Levels Output Levels
Scan levels RGB 0 1.00 255 0 255
R Adjust Red 60 1.00 239 0 255
G Adjust Green 76 1.00 206 0 255
B Adjust Blue 52 1.00 152 0 255
C Adjust Cyan 31 0.64 209 0 255
M Adjust Magenta 32 0.84 235 0 255
Y Adjust Yellow 2 0.53 185 0 255
K Adjust blacK 7 0.77 217 0 255
Step 3 (Level Adjustment, per channel) –
Harry Mendryk (2006/01/31) [#67]:
Firstly, convert the image to CMYK mode: Image > Mode > CMYK color
NB: An essential step for an image printed originally using CMYK (such as American 4-colour comics), this conversion should be omitted for ordinary photographs or other images which were created using RGB.
Secondly, open the Level tool (Image > Adjustments > Levels). The adjustments for Digital Bleaching are somewhat different to those for Colour Correction: I’m not concerned with making the image look correct, colour-wise. I adjust each channel so that the left input level is at the point where the histogram starts to climb, this provides a deep colour to that channel. I then adjust the right input level to past the right peak, this converts light tones to white. Some of these light tones may be under the black line art, and if so need to be removed.
Thirdly, examine each of the image’s colour channels separately (Cyan, Magenta and Yellow), by selecting them one at a time in the Channel Window. I have set up Photoshop to display single colour channels as greys, not as colours. Notice that over most of the image there does not appear to be any Cyan (blue) where there would be line art. An exception is in the steering wheel. If my ‘A’ channel looked like it had colour in the line art area, I could go back to the Level tool and push the right input level more towards the left to remove it. But sometimes getting ride of colour in the line art degrades the colour outside of the line art too much. In the Cyan of my example, that is the case: to get rid of the Cyan from the steering wheel line art, I pretty much loose Cyan everywhere. So I decided not to adjust the Cyan channel any further In fact I did no further level adjustments to Magenta or Yellow either. And Black is exempt from these secondary adjustments, all of the time.
Step 3(a) (Color Dodge) –
Harry Mendryk (2006/02/01) [#84]:
I found a use for the Color Dodge “Apply”: a new step, between my original steps 3 and 4. Selecting the Black channel, I run: Image > Apply Image. Instead of choosing a colour channel (as in Step 4), I choose the Black channel and Color Dodge, but do not tick ‘Invert’.
Doing this seems to have some bleaching effect. I tried my usual practical and theoretical tests. This time the theoretical tests indicate similar, but not identical, results compared to my original Digital Bleaching sequence. But the practical do show some positive results: some line art disappeared using my original steps, but did not when the new step was added.
From this, I can’t say definitively that this new step should be added to my Digital Bleaching. But I do plan to try it when I generate line art from a scan.
Step 4 (Apply Image) –
Harry Mendryk (2006/01/31) [#68]:
Having adjusted the levels for the channels, I now select the Black channel by clicking it in the Channel window. Viewing should also be for Black only.
Now I use the menu item Image > Apply Image on the Black Channel (it was selected above), using each of the colour channels (Cyan, Magenta and Yellow) in turn. I am going to use the Screen with invert.
Attached is an example of the settings I used for Cyan. Before I accept them with “Apply”, I click the preview on and off to see the effect on the Black channel. If the preview is visibly better than the non-preview I will accept that particular “Apply”. By “better”, I mean that some of the tones outside of the line art disappear or diminish. I also do not want to see the line art deteriorate much. In this particular example, the “Apply” of Cyan to Screen the Black clearly helps, so I accepted it.
I repeat the “Apply” operation for Magenta, and after that for Yellow. In each case I use the preview to ensure the change gives a benefit. In my example, the Black channel improved with the change to Magenta and to Yellow.
After all the “Applies”, bleaching progress has been made.
David [betroot] (2006/01/31) [#79]:
At the “Apply Image” stage, I accidentally chose “Color Dodge Mode” and it seemed to have a stronger bleaching effect.
Harry Mendryk (2006/02/01) [#84] (see also Step 3a):
NB: In summary, Harry will not modify his step 4, but will add a new step (named by me as Step 3a), between his original steps 3 and 4.
There does seem to be a stronger bleaching effect using Color Dodge instead of Screen in the Apply step (step 4). However I also observe that the “bleaching effect” was stronger in light tones of the Black channel than in the darker tones. Remember that the Apply step is followed by a Threshold adjustment (step 5). What is important is the combined effect of the two steps.
Frankly, I do not understand Photoshop’s description of what Color Dodge does, although I do understand what the Screen is doing.
So I decided to investigate further, using practical and theoretical examples. For the practical test I used the “Journey Into Mystery” cover David posted and also my “Young Romance” high-resolution panel.
For theoretical testing, I created new grayscale images in Photoshop with two channels. I used the Gradient Tool horizontally in one channel, and vertically in the other. I could then use this image file to run the Apply and Threshold, selecting one channel and applying the other.
In the end I did not see much difference in my practical examples between using Color Dodge and using Screen. However, I used only two samples. Perhaps other comic scans would show a difference.
But the theoretical examples showed very different results. Here the use of Screen did exactly what I wanted, but results from the use of Color Dodge were not satisfactory. My conclusions from these tests is that personally I will continue using “Apply Screen”, as outlined in my Step 4.
However, I did find another use for the Color Dodge “Apply”. This would be a new step between my original steps 3 and 4. Selecting the Black channel, I ran the Image > Apply Image. But instead of choosing a colour channel, like I did in Step 4, I chose the Black channel and Color Dodge, but did not tick ‘Invert’. Doing this seemed to provide some “bleaching effect”. I tried the same practical and theoretical tests as before. This time the theoretical tests seemed to indicate similar, but not identical, results between using my original Digital Bleaching sequence compared with adding the new Color Dodge step. But the practical did show some positive results. Some line art disappeared using my original step sequence, but did not when the new step was added. From this, I can’t say definitively that this new step should be added to my Digital Bleaching. But I do plan to try it next time I generate line art from a scan.
Step 5 (Median filter in Greyscale) –
Harry Mendryk (2006/01/31) [#69]:
At this point I no longer need the colour channels. There are lots of ways to get rid of them. What I usually do is use menu item Select > All. Then, in the Channels window, I select each colour channel (Cyan, Magenta, and Yellow) and press the delete key. With all the colours gone, and only Black remaining, I use Image > Mode > Grayscale to convert the image from CMYK.
Next, click the little triangle in the Channels window and select the “Duplicate Channel” option. I do this because I am going to perform some operations that may cause the line art to loose details. I will not perform these operations on the duplicate copy, so it will be a reference when I manually edit. I select the Black channel in the Channels window.
Next I use the Median filter (menu item: Filter > Noise > Median). I usually set this to the lowest Radius, that is 1. The Median will help reduce the noise that can occur in the image. Unfortunately it also affects the line art itself. The larger the Radius used the less noise but the more detail of the line art is lost. But I know I am going to have to manually edit some of the noise out, as well as manually restore some of the line art detail. Like I said, I generally use a Radius of 1; others might choose a higher value.
After the Median filter, more bleaching progress has been made.
Step 6 (Threshold adjustment) –
Harry Mendryk (2006/01/31) [#70]:
Before proceeding, I usually pick some appropriate section and magnify it to 100% or 200%. Still leaving the Black channel selected in the Channel window, I click the view for the duplicate copy.
This sets me up for the next step, which is the use of the Image > Adjust > Threshold tool. The Threshold tool turns the image to just pure Black and White. I can select where the Threshold point should be. Anything below the Threshold level will turn Black, everything above will be White.
By setting things as I did above, I can judge what might be a good Threshold setting. Moving the setting to the left will remove some of the unwanted non-line art. But it will also remove some of the wanted line art. Moving the adjustment to the right will have the opposite effect.
In the magnified view, areas with white are areas that will be pure white in the image. Areas of light red are those removed from the image: often we don’t want them, but there might be line art that we would like to have. Areas that show up as dark red are those that are not removed from the image but we wish they were. Areas with a mid-red tone are the line art that will be included as desired in the image. I can tell you right now, you are not likely to be able to find a perfect point. You will always lose some line art and get some non-line art. You have to pick a good compromise point.
Attached is a copy of what the magnified image looks like while I am making
the Threshold adjustment. Once I achieve an acceptable adjustment, I click
the “OK” button. Also attached is what the image looks like at this point.
Step 7 (second Median filtering) –
Harry Mendryk (2006/01/31) [#71]:
This step is optional. I look at a magnified view (200%) of the Black channel. If it is noisy, I apply the Median filter again. It is the same sort of compromise as before. The higher the radius the less noise, but the less detail in the line art. In this particular case I decided to use the Median filter again, with Radius of 1.
NB: For the Median filter (Filter > Noise > Median), applied in Greyscale, see messages #35 and #69 (above).
Now I have finished the easy part. From now on, I have to manually edit the black Line Art image, using the Pencil and Eraser tools. I also have a Black Copy channel, that hasn’t been filtered or threshold adjusted, to use as a reference to aid my editing. To completely clean up the image takes some further effort.
Matt Moring [m.moring@comcast.net] (2006/02/01) [#91]:
Right now I’m trying to finish off a story for an upcoming DC Archive book.
To do this right, you need to be an artist yourself. There’s no copy machine that will take a colour page and spit out a finished page of black & white line art. There’s a lot of effort that goes into doing a page right.
Conclusions
Harry Mendryk (2006/01/31) [#74]:
I use Digital Bleaching to restore the line art from scans of Simon & Kirby covers. Ideally it would remove all colours and leave just the line art. But it is not 100% effective.
Depending on a particular scanner, and the settings used in RGB Level Tool adjustment, it may not remove Purples (Cyan + Magenta) very well. But even under the best of circumstances it cannot remove grey tones, such as found in colours such as Brown.
But it doesn’t destroy the original comic, which for an amateur restorer like me is a paramount concern.
Chemical Bleaching is destructive, but you get very clean line art. The only retouching that would be required would to fix up creases and original printing errors.
Harry Mendryk (2006/01/31) [#83]:
A cover will be a problem if it contains a lot of purple and brown.
Harry Mendryk (2006/02/01) [#85]:
Some observations on using Digital Bleaching on the low resolution “Journey Into Mystery” cover David provided.
Because of the low resolution of that scan, the line art is often just one pixel wide. Using the Median filter, as I describe in Steps 5 and 7, has the effect of wiping out a lot of line art: so I suggest you skip using the Median filter when Digitally Bleaching low resolution images.
You will end up with a lot of noise to clean up. But that will be better than all the line art that would need to be edited back in.
The other observation concerns Apply Screen (step 4). Certainly Apply Screen should be used as I described using the Cyan channel. But when I did the same thing using Magenta or Yellow, the effect was to drop some of the line art without any other benefits. So for the “Journey Into Mystery” cover, I would suggest to do Apply Screen using only the Cyan.
This has nothing to do with the low resolution of the scan. Actually the “Journey Into Mystery” cover is more typical with respect to step 4 than my high resolution example.
My attempt on this cover did a nice job cleaning up the background “grey”, but, as I expected, there has to be a lot of manual editing of those areas which were originally purple or brown.
Little Bumps in CMYK histograms
NB: These notes relate to Step 3 in Harry’s above procedure for Digital Bleaching of the image.
Dario [vulcaniano99] (2006/03/22) [#118]:
I see some “little bumps” in the histograms, in CMYK colour mode, in scans of my old comics. I was wondering what are they. Is it an artifact of the ageing of the inks on paper, or are they a feature of the colour printing, even for new comics?
Harry Mendryk (2006/03/22) [#119]:
For the most part it is due to the aging process. Aging adds black to areas where there is no black ink. This is due to dirt and grime on the page, changing of ink with age, and the yellowing of the paper. With the increase in the K channel, the level for the other channels typically decreases. But the effect is not uniform. Areas with ink from one channel will change the least, areas with colour overlaying one another will change more. Hence the extra bumps.
I would like to think that this effect would not happen as much with recent comics. But I have done little work with the more recent stuff. One problem though is recent comics have a screen density much higher than that in the Golden or Silver Age. Scanning at 300 to 600 dpi works very well with those comics. But for modern comics it is nowhere near enough. I am not saying you couldn’t do it, just that I suspect the extra bumps will be there.
Mix Channels : Apply Image
NB: These notes relate to Step 4 in Harry’s above procedure for Digital Bleaching of the image.
David [betroot] (2006/01/24) [#42]:
Apply Image: In the past I would’ve put the Cyan channel in a new document, and used the Black channel set to Screen mode as the Layer mode. Apply Image saves the bother of creating a new document.
You can mix channels with any layer Mode, using Apply Image.
“Calculations” allows you to generate a new channel by “mixing any 2 channels” – useful for extracting masks for photo retouching (like say a model’s hair in a blue sky and you want to composite against a different background – calculations would help in creating a silhouette mask).
Harry Mendryk (2006/01/31) [#60]:
The purpose of the Apply step is to remove some Black from areas of Cyan. I’ve attached an image of the Apply dialog to make the settings I used clear.
This apply is done on the Black channel only. But when I worked I clicked the CMYK view. I also attach an image of the Channel Window to show this. I select the CMYK view because I want to see the effect of the Apply in order to adjust its strength (opacity).
If you try this Apply you can tick and untick ‘Preview’, to see the effect of this operation.
Although the effect of the Apply step is not dramatic for this particular image, it is an operation I use in the Digital Bleaching process.
David [betroot] (2006/01/31) [#72]:
I hadn’t used Apply Image before. It puzzled me, then I realized it was similar to what I’ve done in the past: copy the K channel and paste (so it’s a Layer), and set the Mode to ‘Screen’.
The Apply Image step does it for me. I don’t have to make the Layer, etc.
Then I realized it’s like Calculations – where you can do a similar change to 2 separate channels (screen, Multiply, etc) to generate a new channel – this is useful for extracting masks in photography.
Example: A girl with windblown hair, and you want to make a silhouette mask including the hair strands.
Digital Bleaching (Kris Brownlow’s method)
Kris Brownlow (2006/02/01) [#92]:
I tried to “bleach” a colour scan on my Epson HP scanner, to see if it could be done. The scanner software does not have a traditional “black and white” function, so I used the “old photo” function.
The process:
1. Once an image is scanned, go to EFFECTS and select “Old Photo”.
2. Go to ENHANCE and move “Highlight” and “Midtone” to 89.
Harry Mendryk (2006/02/02) [#94]:
I am impressed with the results you got using just the Epson scanner adjustment. It was hard to judge from your image, because of how faint it was, so in Photoshop I converted it from RGB to greyscale, and then used the Level tool to adjust the lower end.
Most of the colour has indeed been bleached.
I then opened the original scan file, created a new channel, and copied my adjusted version of your file into it. This allows me to better compare the two by either switching views or viewing both at the same time (the bleached image channel acts as a red mask).
Initially I observed that the blacks in the Epson bleached image are pretty noisy. But when I examined the combined file, I found that the noise was in the original comic printing. This is not surprising, as the comics were printed with a rather crude printing method on rather poor paper. No matter what bleaching technique is used, retouching of some kind is required to correct for this.
I then magnified the face of the foreground woman. I found the Epson bleached line art a little narrower than the original line art. I also found some of the finer line art had disappeared in the Epson version.
So the Epson bleaching is not perfect, but none of the digital bleaching processes are. Perhaps with a little tweaking of the settings in the Epson bleaching, you could achieve better results.
But using Gaussian Blur and Threshold in Photoshop would easily get the line width back to what it should be. And following that, one could manually edit back in any details that had been lost.
All that matters to me is that the results are accurate. I may not understand exactly what your Epson software is doing, but it seems to me it could be a viable tool.
Digital Bleaching (David’s method)
David [betroot] (2006/02/01) [#87] (use “Color Dodge” to make Line Art):
“Color dodge” is a technique to do with obtaining Line Art (better than the “Find Edges” filter), a method to turn a scan into Line Art –
1. Duplicate the background layer and Invert it (making a second layer).
2. Invert (negative) the duplicated layer (looks like a negative photo).
3. Set mode of the layer to ‘Color Dodge’ (the image will appear to disappear!)
4. Gaussian Blur the duplicated layer (the one which has been inverted, and set to Color Dodge) with a very small setting, like 0.8 (you will see in the preview the effect and can adjust it).
5. Flatten the Layers.
6. If the line work is light, you can duplicate and set the new layer to ‘Multiply’ (and you can duplicate the Multiply layer more times as required).
Optionally, you can use Threshold if you want black-and-white Lines.
You can save the steps as an Action in Photoshop.
Colour Correction : Harry Mendryk’s method
Harry Mendryk (2006/01/24) [#32] (CMYK Settings):
NB: This was originally posted as part of Harry’s discussion about High Resolution scanning (see also #33 and #34). But it’s of general applicability, so needs to be here.
Before doing the actual colour correction, make sure that CMYK conversion is set properly. In Photoshop 5 the setting dialog can be found using: File > Colour Settings > CMYK Setup
In Photoshop 7 getting the dialog is a little more involved. First bring up menu item Edit/Colour Settings. At the end of the CMYK field is a checkmark, hitting it causes a list of options to be displayed. Choose “Custom CMYK”.
Once you get the CMYK Setup dialog in the Seperation Options, select GCR and in the Black Generation select Maximum. Hit OK (twice in Photoshop 7).
1. Level Tool –
Red Input Levels : 32, 1.00, 255
Green Input Levels: 32, 1.00, 222
Blue Input Levels : 0, 1.00, 182
Note that I am actually moving the left and right triangles to where the particular histogram curves up. The left side makes Black, the right side makes the White. The exact point of a setting is based on the histogram, not the image.
2. Select from menu Image/Mode/CMYK colours
3. Level Tool –
Cyan Input Levels : 19, 0.88, 233
Magenta Input Levels: 14, 0.66, 204
Yellow Input Levels : 21, 0.65, 205
Black Input Levels : 5, 0.54, 185
Note that here I adjust using the left triangle first, remember the position of the center triangle, adjust the right triangle, and finally reposition the middle triangle back to where I remembered it. Although the histogram gives clues to what might be good settings, ultimately it is the preview of the image itself that is most important. And there may not be a perfect setting. With this example the Cyan adjustments are a compromise of getting the Cyan out of the whites and not loosing the light green background. I suspect this adjustment might have been easier at 600 dpi.
4. Select Black channel but click CMYK for viewing, I am going to remove some of the black undertones to the Cyan. The actual change must only be done to the Black Channel, but I want to view what is happening to the colour version of the image.
5. Image/Apply Image; Channel: Cyan; check Invert;
Blending: Screen; Opacity: 50%.
The 50% is my personal judgement. I like to leave some undertone Black in the Cyan.
6. Colour conversion completed. I would then switch to manual editing. In particular I would want to use the Eraser tool, selecting just the Cyan channel. Selecting just the Cyan makes Erasing the Cyan out of the word balloon area easy, because Black is not affected.
I might further want to fix up the Green sidewalk in the background, remove a little more of the Black from it, and boost the Yellow a bit. I did not do that here, because one respect my example panel is no better than David’s eBay examples. In order to make the file smaller for email purposes, I set JPEG compression to 3. This leaves rather severe patterns in the colour channels. Normally I use JPEG compression of 7. That level avoids that sort of pattern.
7. Select from Image > Mode > RGB colours. Put the image back to RGB. Although JPEG is happy with CMYK, most browsers and many printers are not.
Harry Mendryk (2006/01/19) [#7]:
This is my method for colour correcting comics scans. I work in Photoshop 5 and (for some features) Photoshop 7.
I have added photos of some of the steps and tool dialogs used in a folder, “HM’s color adj”, in the Photos section. If you do not find what I’m saying clear, check them out:
https://groups.yahoo.com/neo/groups/digitizing_comics/photos/albums
Step 1: Scan –
I scan at 600 dpi with auto colour correction or auto levels features turned off. My last scanner had really poor gain, so I had to set the manual level adjusments to get more out of it. But I was careful not to push the levels too far up. Once the gain is too high you loose data that I need for my process. If the gain is low, my method will fix that up with no apparent loss in quality of the final result. You have to use really low gain to harm the corrected scan. My present scanner does a nice job without any manual adjustments.
You can use either the Level tool or the Histogram tool to judge if the scan has all the data needed. The histogram I want to see in a scan is the one for the combined RGB. Going from right to left (dark to light) the curve I want to see is flat at the start, starts to turn up, can go through any number of peaks, but eventually turns back down, and is flat by the time it gets to the right side (255). I do not care about the actual values. What I care about is the flat start and end to the curve. If that is missing and either 0 or 255 has a curve off the bottom axis, then data has been lost. In the photo section I have a photo of an original scan, and the RGB histogram (using the Level tool) for that scan, before I have done any adjustments.
Step 2: Level tool –
I do my colour correction in two steps. Scanners use RGB sensors. So I do my initial rough adjustment in RGB. I do not touch the combined RGB level, all adjustment is done on the individual Red, Green and Blue channels. I adjust each Red, Green and Blue channel separately. But how I adjust them is the same. In each case I adjust the low input level by dragging the little black triangle from 0 up to the point in the curve where that channel starts to ascend. I also adjust the high level, by dragging the little white triangle from 255 to the point on the curve where again it goes up. With my previous scanner I would also have to set the middle input value. I had figured what seemed to be a good setting for it and would effectively just load that number into the box. My present scanner does not need that sort of fix, so the middle box always remains 1.00. The photo sections have the settings for each of the channels that I used on the scan. After the Red, Green and Blue channels have been set, and the OK button clicked, the image should have richer blacks and the paper should be closer to white. The photos section has an image of what the page looks like after the adjustment. In this particular case, the original scan was very yellow, but the yellow was pretty even across the page. After the adjustment, the white in the word balloons looks pretty good, but the paper edges are rather splotchy. Other pages may have browning toward the edges and the end result may be a pretty good white on the interior, but bands of yellow or brown along some edges. This adjustment increases the contrast of the original scan, so these effects can look worse than on the original. I can’t afford pedigree comics with pure white pages, so this is something I just live with. Although I have seen better, I am please with the results so far for this particular scan.
Step 3: Convert to CMYK –
Comics are printed using CMYK inks, so it makes sense to me to do the final colour adjustments in CMYK mode. The only important thing is that the Photo CMYK setup be set to GCR Seperation Type, with Maximum Black Generation. I leave it that way all the time.
Step 4: Level tool –
I now return to the Level, but now the image is in CMYK mode. Here I adjust using not just the histogram, but also watching the image, and using the Info tool window. I do not touch the combined CMYK channel, but work on the individual Cyan, Magenta, Yellow and Black channels. For any channel I generally start with the Low Input Level, dragging the black triangle from its right corner. I do not try to push the C, M or Y channels to achieve the strongest possible colour. Rather I prefer to leave it somewhere where near where the actual data starts. Precisely where is a judgement call based on the preview of the image. If you want to push a channel to the max the histogram will sometimes show where that is as the first small peak going from right to left. Unfortunately, depending on the image, that peak may not show up. In my example I can see the peak for the Cyan and Yellow channels, but I can’t make it out for the Magenta. In any case my advice is to move the cursor over an image area that appears to have a solid example of the colour of the channel of interest and look at the Info window. If the Info window shows that particular channel has reached 100%, you have reach or exceeded the peak. Go past the peak, and final results are bound to be pretty bad. In fact when I adjust the magenta that way to find the peak, the girl’s dress has a nice red, but the flesh tones are terrible. My tastes are to find some good point near where the data starts. Then I note the position of the Middle Input Level triangle. I use landmarks on the curve to judge that position. If the curve fails to provide a good landmark, I fall back to landmarks in the dialog box itself. I do not more the middle triangle yet, just remember its location. I then move the High Level Input (white triangle) from right corner to the left. Now I am trying to remove those tones that make the paper off white. Some cases the tones maybe under some other colour. The histogram will invariably show some peaking on the the right side. The peaking may have an abrupt beginning. That often is the a clue to how far I have to push the white triangle to remove unwanted tone. But even if there is an abrupt edge, and especially if there is none, I keep and eye on what is going on in the image. And I stop periodically to use the Cursor and Info window to judge the effect of the area having the unwanted tone (usually the white). But like I said I watch the image. It’s no good to remove a coloured edge due to paper browning if you remove all the yellow inks also. I use my judgement on how far to push the High Levels before it damages the final results. In this example in the Cyan I moved it past the abrupt edge. The Magenta had a secondary peak right on the edge; I moved the white triangle past the abrupt edge but not into the secondary peak. The yellow channel had a gradual climb as it approached the right. For yellow I moved the adjustment to about where this curve starts. This was a pretty extreme adjustment and in the end did not get rid of all the yellow, but judging from theimage it was as far as I was confortable with. Up to now I have not mentioned the Black channel. Actually it is adjusted pretty much like the others. However it almost always shows a peak on the left. I prefer to push the black further then I do the other colours. I usually put the Low setting at the where the left peak starts. After setting both the Low and High Level adjustments I then move the Middle adjustment to the landmark I remembered from before. Moving the other adjustments moves the Middle one automatically. But I want it back to the location I remembered from before (based where it is when set to 1.00, and the Low setting where I want, but the High Level still at 255). The final adjustment is not as cut and dry as my initial RGB adjustment: once I get the channels where I think they should be, I do not hit the OK button right away, but look at the image and decide if I might want to try other adjustments. But any channel I do adjust I redetermine where the Middle level triangle should be. Once again the photo section has pictures of the settings I used for each of the channels as well as a copy of the image the adjustment was actually made.
The final results of colour correction in this case are somewhat mixed. One problem is this was obviously originally a poorly printed page. With my minimally adjusted Magenta channel the dress is not a great red, and yet the flesh tones are a bit strong. When restoring interior pages, I personally do not try to make them perfect. And if you keep the imperfections, you are going to make some compromises. What those compromises are will vary based on personal tastes. Another problem with this particular final correction is there still is a bit of yellow in the white above the printed image. From experience I know that yellow will be even more noticable when printed on white paper. In this case I will do some editing with the Eraser tool on the Yellow channel before I switch the mode from CMYK back to RGB. I find it easier to edit the image in CMYK.
When I do interior restorations, I typically scan all the pages I want at one time. I then use the Automation feature to do all the file opens, RGB to CMYK conversions, and final CMYK to RGB conversions, leaving me to do the individual Level adjustments manually.
Harry Mendryk (2006/01/22) [#14]:
I mentioned in an earlier post that when I do colour correction I make use of Line Art that I had made previously. The Line Art is generated using digital bleaching of the same scan files, and therefore match them perfectly.
I just finished restoring the Young Romance #7 cover. I decided to use de-screening on it. Rand’s “bulletin board” procedure indicates resizing in the middle of the de-screening steps. I can offer no firm reason that this sequence should be followed. But I can say that resizing should never preceed the blurring step. And in general it is best to do filtering at the final resolution. On that basis, the bulletin board makes sense.
But I decided against following that sequence. The reason is that although I know the resize that will need to be done for my book, I may have other uses for the cover and I do not know what resize they will require. If the image is properly de-screened at the original scan resolution, it can then be reduced to any size without the fear of Moire occuring.
I also did not follow Rand’s bulletin board in respect to blurring the channels individually. Here my reason is different. I have only recently started experimenting with de-screening. I would first like to get more experience with my present technique. That way when I do try individual channel de-screening, I will be better able to evaluate the results.
I have added to the photo sections, the original scan, the original line art, the cover after colour correction and touch-up, and the cover after de-screening and touch-up. Below I outline the steps I take.
Step 1: Colour correction
Perform colour correction as described in a previous post. I leave the cover in CMYK mode.
Step 2: Add Line Art Layer
I create a new channel and paste into it the Line Art that I created when working on the Simon & Kirby covers. This Line Art was made using this particular colour scan, so it perfectly lines up. I will always keep this as the top layer. And none of the editing, colour correction, or descreening steps defined below are done on the Line Art layer.
Step 3: Create special selection channels from Line Art
The Line Art from the book includes black letters which on the colour cover are Yellow (upper right) and White (in box on the right side). I use Duplicate Channel on the Line Art to make three copies. I edit these copies so one is just the Yellow letters, another channel is just the White letters, and the third copy is that part of the Line Art that truly will be Black. I do this so when I am editing I can switch from one selection to another quickly, as needed.
Step 4: Create Line Art Layer
I create a new empty layer. I use Load Selection to load the invertion of the Black part of Line Art. I fill the selection with 100% Black. This layer remains the top layer so that when I am done, the Line Art is fully Black, which is how I like it.
Step 5: Use Erase tool to remove unwanted undertones
Now I use the eraser tool on each channel to remove colour undertones. These undertones are artifacts left over from the scanning and colour correction processes They will have a tendancy to give the correct colours a muddy look. I do not remove the black undertones from under the cyan areas. They will be handled later in a separate step. Each channel has its own screen angle. I use this to help recognize the undertones that I want to remove. Using the previously made selections help me to do things like remove the Cyan undertone to the Yellow Letters without affected the Green background. During this editing step I also remove registration problems, ink smudges, and uncorrected paper browning.
Step 6: Use Screen in Apply Tool to reduce Cyan’s Black undertone
Now for the black undertone to cyan. My experience is that Cyan seems to have stronger black undertones then Yellow or Magenta. I have several explanations for why that is. But the important thing is that I personally do not like these comics restored with all of the Cyan’s Black undertone removed. So I handle them differently. I start by making a duplicate of the Cyan channel. I use the Level Tool and bring the left triangle to the start of Cyan’s left peak and hit the OK button. I then select the Black channel but display all the CMYK channels. I then open the Apply Tool from the Image Menu. I set the layer to Background, the channel to Cyan, turn on Invert, and set blending to Screen. Doing this allows me to selectively mask out only the Black under the Cyan. I try various values of Opacity for a value that looks correct to me judging by the preview. In this case I accepted 50% and hit the OK.
Step 7: Use Curve Tool to adjust hair and hat colours
I was pretty happy with the cover’s coluor at this point except for the woman’s hair and the man’s hat. I am always a little uncertain about my CRT’s calibration so I print just those sections.
The printout looks better, but on the print the hair and hat are more grey compared to the comic’s dark brown. So I duplicate the Line Art channel twice and edit to make one a selection of the hair and the other a selection of the hat. I then adjust with the Curves tool on each section to get the brown I want, and I save a copy of the curve before I accept the changes. I proof my changes, and if still not happy I go back in the history pallet to try again. But I still have my selection channels and stored curves from which I can tweak. After a few iterations I am finally happy.
Step 8: Create De-Screen and Trap Layers
I duplicate the Background Level (the one I’ve been working on) to make a De-Screen Layer. (I do not want to de-screen the Background Level directly in case I want to go back and change it some day). I then duplicate the Image because I am going to perform an operation that can only be done on a flattened image, I call this copy the Trap Image.
I flatten the Trap Image (remember the Black Line Art is the topmost layer). I use the Trap tool set to maximum (10). I copy the Trap Image and paste it back on the original image and name it the Trap Layer (I discard the Trap Image). I make sure that the Trap Layer is below the Line Art Layer and above the De-Screen Layer.
Step 9: Limit Trap Layer to colours that were trapped under Line Art
I clear the entire contents of the Black channel for the Trap Layer. I use Load Selection with the Black Line Art channel. With delete, all that remains is the trapping under the Line Art. This will help to reduce the halo affect from the de-screening procedure later.
Step 10: Merge the Trap Layer into the De-Screen Layer
I then select the De-Screen Layer. I duplicate the Line Art channel to make a De-Screen channel. The De-Screen channel already has the Line art for the Yellow and White lettering, but I also edit it to include the whites inside the title characters. On the CMYK of the De-Screen layer I select Load the De-Screen channel with invert set on and then Fill with 100% white. The Trap Layer is merged down into the De-Screen Layer. The De-Screen Layer now has the CMY colours trapped under the Line Art, but the Line Art itself removed.
Step 11: Use Gaussian Blur on the De-Screen Layer
On the De-Screen layer I use the Gaussian Blur tool with the Radius set to 4.0 pixels.
Step 12: Use Unsharp Mask tool on the De-Screen Layer
On the De-Screen Layer I used the Unsharp Mask tool. I set the Radius to 4.0 pixels (to match what I used during the Gaussian Blur) and initially set the Threshold to 0. I then adjusted the amount until I thought it was sharped enough (my case was 130%). I scrolled the image to a part of the image that had flesh tones, then adjusted the Threshold up to where the tones looked good (15).
Step 13: Delete from De-Screen Layer those parts not to be de-screened
I Select Load the De-Screen channel with invert and then delete the parts from the De-Screen layer that do not require de-screening.
Step 14: Correct remaining defects with Paintbrush
Using the Paintbrush tool, I then corrected the defects caused by de-screen as well as those on the original comic.
Harry Mendryk (2006/08/07) [#121]:
I have posted my technique for colour correction. I still use my method. You can see some of my results on my Simon and Kirby blog:
http://kirbymuseum.org/blogs/simonandkirby/
Colour Correction : Rand’s method
Randolph Hoppe (2006/01/18) [#3]:
Rand’s “bulletin board” method (so-called because Rand pinned a note of these steps to the bulletin board behind his monitor):
1. Rotate
2. Crop
3. Convert to CMYK
4. Auto Levels
5. Curves to Blacken
6. Gaussian Blur each channel separately
7. Re-size
8. Curves on K channel only
9. Sharpen all
Harry Mendryk (2006/01/19) [#4]:
> 1. Rotate
> 2. Crop
Harry Mendryk (2006/01/19) [#4]:
Partly self explanatory. But lately I have been using the crop tool to rotate also. I used to try to be very precise and get one edge of the comic perfectly vertical. But, frankly, I have found that when they inked the original panel layout they were pretty sloppy. When you get one edge perfect the others might look pretty bad. Rotating using the crop tool allows me to better visualize, and to find a compromise for all edges.
Randolph replied (2006/01/19) [#11]:
I start with the measure tool and “image|rotate canvas|arbitrary” to get something either horizontal or vertical. Then compromise.
3. Convert to CMYK
Harry Mendryk (2006/01/19) [#4]:
Again self explanatory. Scanners usually have RGB sensors so it is natural to import scans as RGB. But comics are printed with CMYK inks. So at some point in colour restoration it makes sense to work in CMYK. But it is important to have your Photoshop CMYK set up properly. You should be using GCR, with “Black Generation” set to Maximum. The purpose of this is that greys are thereby generated only in the black (K) channel, not by various combinations of all the channels.
NB: In Photoshop 7, go to menu item: Edit > Color Settings. At the end of the CMYK field is a checkmark, clicking it causes a list of options to be displayed. Choose “Custom CMYK”.
In the CMYK Setup dialog, in “Separation Options” select “GCR” and in “Black Generation” select Maximum. Click “OK” (twice).
4. Auto Levels
Harry Mendryk (2006/01/19) [#4]:
“Auto levels” is a quick and dirty tool. I have tried it and compared the before and after. It seems to maximize the ranges of the CMY tones, but does not seem to do much to the K channel. This has the effect of giving stronger colours to the image. This is important because the tone range of the CMY channels is generally low and uneven. This is due to the original poor printing that comics received, the fading of the inks with age, and the limitations of the scanning. It seems to do a good job on the CMY channels, but I prefer the full control I get from working with the Level tool.
5. Curves to Blacken
Harry Mendryk (2006/01/19) [#4]:
Since the auto level did not do much to the K channel, I suspect you are using it to enrich your blacks. I remember an article I once read that recommended using the Curves tools to do colour adjustment. Curves does seem to provide the maximum flexibility. But again I prefer the Level tool, because it provides a histogram. I find this gives me a better insight into what is going on in the image, and what I should do to correct some of the problems.
6. Gaussian Blur each channel separately
Harry Mendryk (2006/01/19) [#4]:
This is step one of the de-screening process. I mainly de-screen to remove the Moire patterns that often show up when resizing an image. I do not resize all my work, and only use de-screening when actually needed. On the matter of blurring the channels separately, it shouldn’t matter. Certainly there is no harm in doing it. It might be possible to use different settings for the blurring of each channel, but I am not sure what the benefit would be.
Randolph replied (2006/01/19) [#11]:
It is *all* about using different settings on the blurring of each channel. I may have picked this up from Dan Marguiles, a photoshop expert, in a magazine column or website. Or some other photoshop experts forum.
Harry Mendryk replied (2006/01/20) [#12] [#13]:
I recall one guy who generally worked using the Curves tool. I followed that approach back then, but have since switched to using the Level tool. You get more control over how a channel is adjusted from Curves, but I find the histograms help by giving me better insight into the image itself. As for working on channels separately, when bluring during de-screening, now that you mention it, perhaps I can see some advantages. In a recent cover restoration I had to do some severe bluring to get rid of some Moire. Maybe working on the channels seperately would allowed a less severe blur, or require less post-blur re-touching.
7. Re-size
Harry Mendryk (2006/01/19) [#4]:
I’ve been re-sizing *after* de-screening, not in the middle. But, to be honest, I never thought about it. Since you did not get your de-screening technique from me, did the source give any reason for doing re-sizing here? This is something I definitely want to experiment with.
Randolph replied (2006/01/19) [#11]:
I came up with the order of these steps after considerable testing. I think it is about re-assuring that the blacks are where I want them to be after the bi-cubic resampling that takes place when resizing.
Harry Mendryk replied (2006/01/20) [#12]:
It’s always a good idea to keep track of what has happened with the channels as you work. I have only begun using de-screening for restoration recently. I’ll keep this tip in mind.
8. Curves on K channel only
Harry Mendryk (2006/01/19) [#4]:
Again I am not sure what to say about this step. Does your de-screening process somehow affect the black channel? Like I said, I’ve just started to experiment with de-screening.
Randolph replied (2006/01/19) [#11]:
As I noted above, this is just doing a little more adjustment after the resizing.
9. Sharpen all
Harry Mendryk (2006/01/19) [#4]:
I presume you are using the unsharp mask filter to do this, the last step in the de-screening process.
Randolph replied (2006/01/19) [#11]:
No, just plain old “sharpen” and “sharpen more”. I know the unsharp mask filter is a powerful tool, but have never put the time in to figure it out.
Harry Mendryk replied (2006/01/20) [#12]:
I admit I don’t really understand the Unsharp Mask tool. And having three adjustments to use makes it hard to just twiddle until you get the results you want. But I have found settings for two of the adjustment bars that seem to work pretty well. This leaves adjusting to just one (Radius). At that point it just becomes a “sharpen” adjustment. The re-sizing I do for covers is not much. I think that is the reason why I find that the “Sharpen” and “Sharpen More” tools don’t do much.
This does not include hardware issues.
Colour Correction : Conversion to CMYK alters colour
Topic: Loss of “out of gamut” colours
Dario [vulcaniano99] (2006/03/06) [#105]:
The colour adjustment suggested in this list involves, firstly, a rough level adjustment in RGB, then a finer one after a conversion to CMYK.
I am puzzled that when I convert colour mode from RGB to CMYK, the colours change a little (sometimes quite a bit).
Harry Mendryk (2006/03/06) [#106]:
I have seen a slight change when converting to CMYK. This is expected, because the “colour space” for CMYK is smaller than that of RGB.
But I am surprised if you see a significant change, except for purples in RGB changing to more of a grey in CMYK. I fix that by adjusting the middle triangle when I am working on the RGB correction.
Most browsers won’t display a CMYK .jpg file, but Photoshop will have no problem doing so.
Dario [vulcaniano99] (2006/03/08) [#107]:
That was the problem I observed, mainly with purples. Other colours seem to be okay.
David [betroot] (2006/03/09) [#108]:
A computer can display some colours that can’t be printed.
The exclamation mark in the colour pallette (the big one), in Photoshop, demonstrates this: a means of indicating out-of-gamut colours.
When you scan a comic, you convert a CMYK printed image to RGB, but Photoshop then converts the RGB image back to CMYK. The purple region of the spectrum is the part most likely to be altered by this process, because of incompatibilities in the filters used.
Harry Mendryk (2006/03/09) [#109]:
What you are saying is true, but is not the explanation for the problem I had with purples. In my case, when I went from RGB to CMYK the purples really did became greys. Converted back to RGB, they were still grey.
Once I made the proper adjustments, that was no longer true. The purple might change slightly, but it was still a purple.
David [betroot] (2006/03/09) [#110]:
When you scan the colours that are on the page, it takes the data and records it as RGB colours for screen display. When you make a conversion to CMYK (a sub-set of RGB), the RGB colour number of every single pixel ‘jumps’ to the value the Photoshop algorithm thinks is the closest CMYK equivalent: it just converts it to the nearest one that it was programmed to.
Select a colour, then check its compatibility with CMYK by clicking the ForeGround swatch in the Toolbar (that brings up the large window for colour choice, with shades of colour and numeric info).
Chances are the purple will force a ! (exclamation point) prompt, meaning “this colour is unprintable in CMYK”, and when clicked it ‘leaps’ to the nearest colour which DOES have a CMYK equivalent.
Often, the purple goes ‘grey’ (dull).
You can cheat this ‘leap’, by selecting an alternate purple colour that has a rich saturation; but by doing so you’ve probably shifted the red or blue component.
That’s why the current comic colourists work in CMYK and avoid RGB if possible (some filters don’t work unless RGB).
I’m not suggesting a method to avoid the problem, just stating the basic facts about RGB/CMYK.
Harry Mendryk (2006/03/10) [#111]:
You are writing about colour space in general. But my discussion with Dario about purple has nothing to do with that colour space issue.
If you change the Foreground colour, but do not use the swatch, but instead enter in the CMYK boxes 100 for C, 100 for M, 0 for Y and 0 for K you will get a purple. There will be no exclamation mark to indicate any colour space problem. Make a new RGB file and Fill it with the Foreground color, and you will get a purple. Convert the file to CMYK and it will still be purple. Convert it again to RGB and it remains a purple. If you are observant, you might have seen that the info box shows the CMYK is not exactly 100,100,0,0 it has shifted a little. But it is still purple.
Colour space difference may explain slight changes in colour when changing to/from CMYK and RGB. It does not explain the big shift to grey that I once had. That shift is an artifact of the scanning process and the settings used. And, at least in my case, that problem was correctable by proper adjustments during my RGB level adjustment step. Hopefully that will be the case for Dario also.
I wish that I could remain in CMYK mode for my restoration. But, unfortunately, scanners actually read the image using RGB detectors, most browsers will not display a CMYK jpeg, and my printer uses RGB.
I do most of my actual work in CMYK. I just have to return to RGB in order to do anything with it.
Dario [vulcaniano99] (2006/03/10) [#112]:
I found a problem even with blue.
When I convert to CMYK, Reed’s costume changes a bit, to a less bright blue
David [betroot] (2006/03/10) [#113]:
FFblues.jpg shows what happens when you sample the blue. The Color Picker says it’s an “out of gamut” (unprintable CMYK) colour – the exclamation mark shows this. If you click on the exclamation mark it will jump to what Photoshop decides is the closest colour numerically, trying to preserve the Hue, Saturation and Value. It’s a grey.
If you just do a CMYK mode conversion, that blue will shift to that grey.
Marked in FFBlues.jpg is an area of blue where, if you click, you WON’T get an exclamation mark warning: a nicer blue.
You are making a judgment here, and saying “colour saturation is most important, I don’t care about value shift”. Photoshop can’t do that, as it’s a subjective judgement.
So, by trial and error, you can find a nicer CMYK blue, and then substitute that. Basically, you would record the number of the ‘nice’ blue. Then, using one of the colour controls (such as ‘selective color’), shift the blues to your replacement colour. Well, that’s how you would evolve a ‘method’ — I can’t give you a step-by-step.
Remember: a comic is a CMYK entity. The scanner (nothing to do with Photoshop) scans the picture and converts it (using the scanning software’s algorithm) to RGB: so it’s a scanner problem. It makes the scan for on-screen representation in RGB. It looks good on the monitor. BUT when you convert it to CMYK, in Photoshop, it isn’t going “back” to CMYK – it’s a first time conversion for Photoshop, which uses its own algorithm to do the CMYK conversion.
David [betroot] (2006/03/10) [#114]:
[This section is valuable ONLY if you intend to print the comic to paper, not if it will only be viewed on a computer monitor.]
One other thing you have to consider. It may look grey on screen, but it’s not until you PRINT it that you can be sure if there is colour loss: it may look dull on screen, yet print perfectly well.
You are converting an RGB screen display to a CMYK printed image, and the concern is for the PRINTING, not for its representation on the screen. It’s the PRINTING that is now important.
He then gets lost in meaningless rambling
The only points he seems to be trying to make are:
a. Do two printouts, one as RGB and the other as CMYK.
b. Ignore entirely what you see on the computer screen, and instead compare the RGB printout with the CMYK printout.
c. Do the CMYK conversion in Photoshop, don’t leave it to the printer, because consumer printers do a crap job of this type of conversion.
David [betroot] (2006/03/10) [#116]:
If you scan a comic, then convert it to CMYK in Photoshop, although it may look duller on the screen it should print like the original!
It’s only the difference on-screen between RGB and CMYK that you’ve been noticing.
The diagram shows how different media react to colour (it’s from an old book, and newsprint, like colour photocopiers, has improved in the last few years). But, in the diagram, see how newsprint can’t handle rich purples, and which colours fall in problem areas.
Harry Mendryk (2006/03/10) [#117]:
What David has said is all true, but I have a different suggestion. The faces have way too much magenta. Was the comic really like that? If not, you have pushed the initial RGB level adjustment too far.
If the comic was heavy in magenta, you can try what David suggested.
Despite what David says, the Photoshop conversion in this case does not push to grey, it pushes to magenta (when I try it). After conversion there is 4% of black in Mr Fantastic’s suit, but 25% of magenta. It is the magenta that is causing the problem.
You can use “Apply Image” to correct that. First select only the magenta channel, and I would advise clicking the little box for combined CMYK. This allows operations to be done on magenta, but shows you how they will look. Then in the menu select Image/Apply Image. Select Cyan for the channel, select invert, and Screen for the Blending. This will remove magenta under any cyan. But there may be a slight shift in the background purples.
Colour Correction : Yellow & Magenta – Edit as CMYK
Harry Mendryk (2006/02/17) [#102]:
I generally do not discuss manual editing of scans. Photoshop provides the tools, but no magic solutions. You have to do a lot of tedious work. But working in the proper colour mode can make some corrections a lot less painful.
Since I work with scans of golden age comics (generally low grade ones), I often work with pages that have a browning problem. My colour correction technique can correct much of this problem.
But sometimes the browning is uneven, so after colour correction part of the page will have white paper, other parts of the paper will have yellow to magenta tones.
I recently scanned a Boy Commandos story from “Detective Comics” that had this type of problem. After colour correction some of the paper was a pretty good white, mostly in the center of the page. But other areas, particularly the left side, were still pretty ugly. Getting the yellow/magenta out of word balloons etc would take a lot of effort in RGB mode.
But I converted the image to CMYK mode using the GCR at Maximum setting. Then I worked on first the Yellow channel, and then the Magenta channel. Erasing unwanted tones out of word balloons becomes an easy task, as work done in the Yellow or Magenta channels does not affect the black lettering, which is in the Black channel. Browned paper like in this example will leave unwanted Yellow tones under some of the Magenta, along with some unwanted Magenta under some Cyan. I use low tone values, screen angles and subject to indicate what should be removed and what left. Low values of Yellow probably need to be removed. Higher values of Yellow that exactly match the Magenta screen angle and pattern, are also likely to be undesired. But a low Yellow associated with a strong Cyan in Brooklyn’s shirt makes it Green and should not be removed. Most skies are made using Cyan alone, the presence of Magenta in skies probably needs to be removed. That sort of reasoning. With practice, it becomes pretty much second nature.
If further cleaning up was needed, I would also work on the Cyan and Black channels. But in this case it looks pretty good with just the Yellow and Magenta work. I really don’t want to spend too much time on a non-Kirby work.
Colour Correction : Avoid the Red Halo
Tom Kraft (2006/01/28) [#55]:
I own a Microtech ScanMaker 9700XL.
I’m having a problem with scanning original art. Some of the finer black lines have a blue or red halo, usually 2 or 3 pixels above the black line or in some cased the entire line has a blue or red tint.
I tried scanning at a higher resolution. This diminished the halo but does not eliminate it. I recallibrated the scanner with the included Kodak recallibrator but observed little difference.
Is there something I can do to eliminate the halo?
Randolph Hoppe (2006/01/28):
What software are you using to scan? If it’s not VueScan or Silverfast, it might be worth trying their demos, although the IT8 colour
calibration is part of the paid versions:
VueScan http://hamrick.com
SilverFast http://silverfast.com
Harry Mendryk (2006/01/28):
I recently got a Microtek 9800XL, and do not have the problem you are reporting.
The problem sounds like one of two things:
1. Calibration – I know you said you re-calibrated it. But the Kodak calibration reference is probably smaller then the art page. If possible try calibrating with the Kodak reference placed midway on the glass.
2. Filter & Descreen – Make sure you have both of these set to none. The description of a halo sounds like a sharpen filter is in effect.
This does not include hardware issues. I used to take original art to a digital service in the city. They could scan art much larger than I could, for a relatively low fee. But one time, scans I got from them had the problem you describe. I tried to talk to them about it, and they re-scanned for me two more times. But the problem never went away. In the end they told me they did not want my business anymore.
Colour Correction : Greys
David [betroot] (2006/02/02) [#93]:
I used Harry’s method of digital bleaching, then some ideas of my own to try and get rid of the grey.
There was a post in the Kirby Group about the greys on covers.
Whoever did the colouring around the early Marvel in the transition to the Silver Age was fond of using Grey as a colour – presumably he/she thought it made the colours “pop” more.
I assume that a grey tone was added to a copy of the original art, either with Benday stick-on screens, or in some cases with watercolour, so that the photographed black plate had greys added.
He may of course have done it on the original art.
Harry Mendryk (2006/02/02) [#95]:
I tried to follow the discussion in the Kirby list about the use of grey on Atlas/Marvel covers. But I was never completely clear on exactly what was meant by using grey. In this particular cover, Journey Into Mystery #52, are you talking about the grey in the giant’s costume? If so, the low resolution of the scan makes it hard for me to give a definitive answer.
My experience with original art is that if the grey was added using Benday, the dot size would be different from the screen dot used in printing the colours. Generally, Benday dots are larger. And, since Benday is manually applied, there often are differences in the dot row/column angles from place to place on the image. Differences in dot size (but not angles) would also be expected with the special pre-treated boards that were sometimes used to achieve the greys. Water colour was also mentioned, but I have never seen it on original Golden or Silver Age art.
However, the JIM #52 scan’s resolution is too low to make such comparisons with confidence. But I will hazard a guess that in this case the greys were achieved just like the rest of the colours. That is, by the comic colourists, based on colour guides. They were not on the original art.
Greg T [Greg Theakston] (2006/02/02) [#96]:
Ben-Day, in my experience is a treated board with two lines at 45 degree angles, left and right. One set of lines is 30%, the other is 50%, so if both are used in an area, the result is an 80% tone.
There may have been a dot-pattern Ben-Day, but I don’t recall seeing it. Usually, the dot pattern grey is accomplished with Zip-A-Tone: plastic sheets with a sticky back, cut with an Exacto-knife.
The water-colour you are talking about was three shades of blue ink which were translated at the engraver’s into a dot pattern. The Marvel cover greys were produced by ink-toning a blue-line board: a fifth colour-separation.
Jack Adler and Jerry Serpe did the grey tones at DC. I suspect Sol Brodsky did them at Marvel.
Colour Correction : Colour Noise
Dario [vulcaniano99] (2006/02/17) [#100]:
Using Photoshop CS2, the filter “Surface blur” will remove colour noise.
Colour Correction : Limit Colour to 8 bit
Harry Mendryk (2006/01/25) [#52]:
Subtle colour differentiation has nothing to do with scanning resolution.
Rather, it is governed by the bit depth. Most people scan with 8 bits per pixel, i.e. for each colour channel, in the case of colour. But some scanners allow 12 or even 16 bits per pixel. 8 bits provides 256 tone levels for a channel, 12 bits provides 4,096 levels, and 16 bits 65,536.
Personally, I think 8-bit depth is sufficient.
Darci (2007/09/04) [#147]:
How many colours should a comic’s palette contain? It seems to me there’s no point in scanning for 16-bit colour, for example, if there are only 1,024 possible colours. What do three colours, times three screen sizes, plus one (for solid black) work out to be?
Harry Mendryk (2007/09/04) [#148]:
It seems to me you have reached the right answer for the wrong reason. If in fact we were trying to use a computer to produce new comic book art that uses a silver age palette, then you don’t need a lot of bits for each colour channel. In fact to minimize file size you would probably be better off using an Indexed Colour file format.
But that is not what we are trying to do. I am trying to restore, as close as possible, the original colours from scans of old comics. Primarily the problem is the page has yellowed, affecting the colours scanned. You need more bits per colour channel to make the distinctions, you simply are not dealing with just 1,024 different colours. Having said that, you don’t need to distinquish millions of colours either.
In my restoration techniques, I work with the individual colour channels. What matters to me is how many levels I can get from each colour channel. With 8 bits you get 256 different levels, with 16 bits you get 65,536. I find 256 levels is more than enough. 65K is overkill, and such overkill results in file sizes that are difficult to handle.
Darci (2007/10/26) [#149]:
Comics have 63 colours (plus black and white).
Steven [webster2000] (2009/05/09) [#152]:
There were no colour matching standards before Pantone. Individual printers provided designers with numbered swatch books, but these would vary from place to place.
Resizing : Moire Patterns
Harry Mendryk (2006/01/19) [#5]:
My restoration of Young Romance #6. The front cover is surprisingly well preserved. There were relatively few tears or creases.
I needed to slightly reduce its size. Moire patterns occured when I did. I had particular problems with patterns in the man’s brown jacket. So I had to do a special job on it.
I ended up with 3 layers: one with severe de-screen of the man’s jacket, another for the lesser de-screening of much of the figures, and a final layer for the solid colours that required no de-screening (mostly the background).
Harry Mendryk (2006/01/23) [#24]:
David wrote:
> surely in the “Tomorrow Man” cover
> there’s ways of getting out the grey
> other than the Eraser tool
The grey in the word balloon is close in tone to some of the grey in the background. Removing the grey from the balloon using the “Level” tool, will adversely affect the cover as a whole.
If for some reason I really did not want to use an Eraser tool, I would probably create a selection of just the word balloon. Lots of ways to make such a selection, perhaps the Lasso tool would do. That way I could use my Level tool on the grey without affecting the rest of the cover.
David [betroot] (2006/01/24) [#30]:
Your mention of moire in the “Tomorrow Man” restoration was of interest. My scanner has built-in filters to get rid of “dots” in printing — using ‘Magazine’, ‘Newspaper’ print (it doesn’t have an ‘Art magazine’ filter that I’ve seen in other scanners). I tried to scan an art picture from a library book and none of the filters (de-screeners) in the scanner was perfect, and left a diagonal line moire (they were very small pictures and I was enlarging them). Do you have an idea for getting rid of moire?
Harry Mendryk (2006/01/24) [#37]:
Moire problems are a recurring headache. The most general solution is to scan at high resolution. Usually the further the scanning resolution is from the comic’s screening density the better. When I work with 600 dpi scanning (and even at times 1200 dpi) I generally do not have any Moire problems. That is unless it becomes necessary for me to re-size. Then it may show up. Both Rand and I have discussed de-screening techniques, and some scanners already come with their own de-screening utilities. If you like we can go over that more carefully. But there is no magic bullet that prevents Moire at all times.
Rand once mentioned getting some of his procedures based on some columns by Dan Margulis. When I dug out some old magazines that helped me, when I first got into doing image manipulation in Photoshop, it turns out that they were also written by Margulis. Dan’s articles are well written and contain valuable info. But his writings are geneally for use with images ultimately used in commercial printing. I do have some articles by Margulis that talk about how to prevent Moire. But I want to experiment with some of his techniques to see if they are truly useful for our type of work.
Harry Mendryk (2006/01/26) [#54]:
Reducing Moire from scanning –
Previously I was asked about how to prevent Moire patterns when scanning from printed material. The short answer is that there is no way that is guaranteed to work in all cases.
However Dan Margulis wrote some articles on the subject that I recently re-read. He provides a shortlist of practices to follow. I’ve reordered them slightly, and added some notes in brackets.
1. Always scan printed material at the highest possible resolution. These scans can be resized down later. (Although a high-resolution scan is moire-free, this does not mean that the resized image will not have Moire.)
2. Don’t use a sharpening filter. (Most consumer scanners use automatic sharpening filters when scanning. To avoid this you would have to get into the setup for your scanner and turn off sharpening.)
3. Don’t use an automated descreening package. (Some scanners have descreening capabilities. Some are better than others. But even when they work they destroy detail. Dan advocates a manual approach in Photoshop. But Dan’s approach is complicated, and I have not actually used it. So I would say if your scanner has descreening, first try scanning without using it. If that does not work out well, try again with descreening.)
4. Learn to read the screen angles of the original. (For black and white, this is pretty easy. For colour there is a different angle for each CMYK ink. Generally it takes some effort to determine these individual angles. Converting the file from RGB to CMYK in Photoshop helps. But it still takes some practice.)
Dan Margulis also provides a 30-degree rule: to minimize Moire, scan the original at an angle 30 degrees from the original’s screen angle. For black and white prints this is not difficult to determine: most B&W images use a 45 degree screen angle. Using the 30 degree rule would mean scanning with the original at a 15 degree angle. Occasionally some B&W are screened at 0 degrees. That would mean scanning the original at 30 degrees. I have never come across B&W screened at any other angle; but if you read the screen angle, you can determine the best scanning angle in all cases.
Or if you cannot read the screen angles, scan a B&W image first at 15 degrees, and if that doesn’t work try 30 degrees.
After the image is scanned, you can use Photoshop to rotate the image back to the original vertical.
Attached are two versions of the same image, cropped to keep the files relatively small. The original was from a movie ad in a newspaper. View them at 100%, or at Actual Pixels: viewing at other than 100% makes it harder to see the Moire; viewing at a reduced size may show Moire on the monitor that is not really present in the original file. The first (angle_0.jpg) was scanned normally.If you look at the forehead of the actress, or in the background, you will see the Moire pattern. The second image (angle_15.jpg) was scanned at a 15 degree angle, then rotated back using Photoshop (Edit > Transform > Numeric). This second image, scanned at an angle, has no Moire.
I can’t say if following the 30 degree rule will always work perfectly. But it should always minimise the Moire pattern.
But things get messy when scanning colour prints. These prints have a different angle for each ink, and they attempt to follow the 30 degree rule themselves. But although a screen is said to have some particular angle, it really is composed of rows at that angle and columns 90 degrees to the angle. This means that only three colours can follow the 30 degree rule in CMYK, the fourth ink must be at some other, non-optimal angle. The eye is less sensitive to Yellow, so that is the ink that normally gets the poor angle.
For CMYK prints, the screen angles normally are Cyan (15º), Black (45º), Magenta (75º) and Yellow (0º). There simply is no perfect scanning angle available, the best that can be done is to be 30 degrees from two of the ink colours. Which two can vary depending of the particular image. But Margulis suggests that the best screening angle is 45 degrees. This is best for Cyan and Magenta, but not so good for Black and Yellow. In fact it is the absolute worst for Black, so I am a little surprised by his suggestion. So I would say try his 45 degrees first, then scan also at 15, 75 and 0 degrees. Use whichever one is best.
Years ago when I first started doing comic scans, I re-read Dan’s articles. But I never tried following them. One of the reasons is a practical one. Most consumer scanners scan up to about 8.5 by 11 inch images. This is fine for comics, until you try scanning them at an angle. Even at 15 degrees, a comic will not fit on this size of scanner. Dan Margulis’s advice is only useful if you have a large scanner, or for scanning prints smaller than comic books.
High Resolution scanning : Advantages
Harry Mendryk (2006/01/24) [#33]:
Scans obtained from eBay have severe limitations with respect to the colour correction method I use.
One major shortcoming is their low resolution (typically only 100 dpi). Golden and Silver Age comics typically are printed with a screen pattern of 85 lpi (lines per inch). At 100 dpi a screen dot pattern on the comic page does not sample well. In Photoshop, first view the attached file at 100% (x1) magnification. The dot pattern is readily seen. Now use Image > Image Size, making sure “Resample Image” is set; then set the Resolution to 100 dpi (at this point DO NOT SAVE). Look at the image again: you no longer can see the screening. (After doing this test, discard the Image without saving it).
My method works best when the scan is fine enough that the comic’s screening dots can clearly be distinguished from the paper background. I usually work at 600 dpi, the 300 dpi of this example is a compromise for email purposes.
The other limitation of files on eBay is that they are generally adjusted to look good. This is usually an auto-adjust. But when an image is adjusted it often loses data that would have been useful to my colour correction technique. Actually my example just barely escapes losing data.
Harry Mendryk (2006/01/24) [#34]:
If you have Photoshop and are going to try to use my technique on the Hi-Res scan I posted, you must have CMYK conversion set up properly in Photoshop. I have two versions of Photoshop.
In Photoshop 5, the setting dialog can be found using menu item File > Colour Settings > CMYK Setup.
In Photoshop 7, getting the dialog is a little more involved. First bring up menu item Edit > Colour Settings. At the end of the CMYK field is a checkmark, hitting it causes a list of options to be displayed. Choose “Custom CMYK”.
Once you get the CMYK Setup dialog, in the Seperation Options select GCR, and in the Black Generation select Maximum. Click OK (twice in Photoshop 7).
NB: The purpose is that now the black channel will have a better black, and there will be less black in the colours.
This CMYK setup is important in that it defines how greys are converted. Commercial printers often want part or all of the greys to be made with CMY inks. For colour correction we want greys to only be in the Black channel. This setup provides that.
David [betroot] (2006/01/24) [#39]:
The first image is an example of colour mis-registration: you can see the mid-ground girl’s lipstick colour is ‘off’ – is this solved by moving the red channel, so the red registers correctly?
NB: The red of the lipstick does not coincide, on the image, with the girl’s lips/mouth. This is due to a mis-alignment of the K plate (holding the line art) and the M plate (holding the Magenta ink), known in printer’s jargon as mis-registration.
Harry Mendryk (2006/01/25) [#53]:
I goofed, and failed to convert the file from CMYK to RGB.
The reason for this image is so that anyone following my procedure by themselves processing the original scan I posted could have something to compare their results with. If they use the same settings I did, they should be getting the same results.
But if they decided to use different settings (a valid thing to do, particularly when doing the CMYK adjustment) they could see if their version turned out better. I expect people will have different preferences on how the final image should look.
I purposely did not do any manual editing on this image. The image is posted as an attempt to allow members to understand my colour correction technique.
In Photoshop you can select the Magenta channel, then use the Move tool to shift it about (i.e. move the Magenta colour patterns to coincide more accurately with the line art: termed ‘registration’ correction). If shifting it up/down and left/right is not sufficient, you can also use Edit > Rotate to do rotation.
Unfortunately, fixing registration problems, particularly on interior pages, almost always ends with a lot of fixing and touching up. As you move the magenta into proper registration, areas which originally were under the black line art (i.e. were over-written by black) become exposed. These newly exposed areas will have to be re-touched.
Photoshop LAB Color
Harry Mendryk (2006/02/15) [#97]:
I am currently reading “Photoshop LAB Color” by Dan Margulis.
The subject of the book concerns adjusting photographs, but I am interested in adapting his ideas to digital comics restoration.
LAB colour is an alternate colour mode used by Photoshop. It provides some benefits as compared to RGB or CMYK, but is not as intuitive.
It consists of three channels: A, B, and Lightness. The Lightness channel is the easiest to understand. Its name pretty much covers what it shows, it is similar to the grayscale of the image. The A and B channels are colour channels. Both A and B show the range between two different colours, with absence of either colour indicated by a midway point. A-channel is for green (negative numbers) and magenta (positive). B-channel is for blue (negative) and yellow (positive). For the following simple adjustment, it is not important to know which colours are part of A and which are part of B.
I will describe a way to do a quick colour correction for a comics scan using LAB colour. This could replace the RGB level adjustment I described previously for colour correction (also for digital bleaching). Like the RGB adjustment, the LAB adjustment only makes an initial rough correction, which can be further improved by secondary adjustments in CMYK mode. I described these other adjustments in previous posts.
1. Convert the scan to LAB colour:
Image > Mode > LAB Color
2. Open the Curves tool:
Image > Adjustment > Curves
3. I work with the Lightness channel first, it should be the default when the Curves Tool dialog comes up. I mouse click the cursor over an area that should be white, in this case inside the word balloon in the center. While the mouse button is held down a little circle will appear on the Lightness curve. I note where it occurs, and then drag the nearby curve end horizontally to the right to match that location. If I now hold the mouse button over the same area, the small circle should be over the point where the Lightness curve starts rising from the axis. The info box will show the L channel in this spot to be in my case 94/100. The first value (94) depends on the particular scan’s white value, but the second (100) is what we are aiming for.
4. Still using the Lightness channel in the Curve Tool dialog box, I hold the mouse button down over an area in the image that is black. In my example I used the lettering inside the yellow heart. A gain a circle will appear on the curve to indicate where to adjust. This time I moved the nearby curve end to the left. When finished the Info Box shows the Lightness channel of these letters to be something like 18/2. Again the 18 value may differ for other scans, but the 2 (or 1 or 0) is my goal.
5. Having adjusted the whites and the blacks, I noticed that the image has become too dark overall. I click the mouse button on the Lightness curve someplace in the middle and drag the curve to the right. This dragging causes the curve to no longer be a straight line. You can tell if you are dragging the curve in the correct direction, because if you go the wrong way it has the opposite affect of what you want. I have attached an image of the Lightness curve having made the three adjustments to it.
6. I now select the “A” channel in the Curves Tool dialog box. For this channel I will only be adjusting the white. I hold the mouse button down inside the same word balloon. In my case the little circle shows up right in the middle of the curve. The Info box indicates the area has 1/1 for the a channel. 0 is the ideal value for no colour cast, but 1 is good enough. So in my case I make no adjustments to the A channel. Had this not been the case, the adjustment would have been similar to what I describe below for the B channel.
7. I next select the “B” channel. When I hold the mouse button down with the cursor in the word balloon the little circle appears on the curve. This time the circle is on the lower half of the curve. The Info Box shows values of something like 17/17. I drag the opposite end of the curve, in this case the top to the left. I keep trying different settings until the clicking the mouse in the word balloon has the circle showing midway and the Info Box showing for the B channel something like 17/0.
8. Having done all the adjustments, I click the “OK” button in the Curves Tool dialog box. I would now convert the image out of LAB colour mode to RGB (if all I wanted was a rough correction) or CMYK (if I wanted to get even better adjustment). I have also attached a before and after image of the cover I tried this on (“Young Romance” #4). Note this example only shows the results of the LAB Curve adjustment, no other work has been done on it.
I have just started experimenting with using LAB color adjustments. I do not yet know whether it provides any benefits as compared to the RGB adjustment I described in an earlier post.
I have also experimented with improving the colours in general (not just the blacks and whites) using LAB. But those maneuvers are a bit more complicated.
Digitally Colour the Lineart
Dario [vulcaniano99] (2006/02/17) [#100]:
In Italy they started to publish Marvel comics in 1971. The paper quality was much better than that used in the USA originally, so the pages are much better printed than in the American originals. However, due to the high cost of colour, they print only half the pages in colour, printing only the lineart of the others.
I would like to digitally colour the lineart. Do you have a process for that?
Harry Mendryk (2006/02/17) [#101]:
To digitally colour the pages originally printed as line art, I can made a few suggestions. I’ve done something similar, using line art that I digitally bleached from some Joe Simon covers.
The first step would of course be to scan a line art page. If the printing quality is pretty good, in Photoshop use first Filter > Noise > Median with a very low Radius setting (perhaps 1). Then use Image > Adjustment > Threshold to covert the line art to pure black and white. If the print quality is not good enough, you may have to just use Image > Levels or Image > Curves to improve it as much as possible. In either case, you now have the line art in grayscale.
Next open a new file that is the same size as your line art image. But this file should be in whatever colour mode you want to work in. I generally do my work in CMYK. Make a new Layer: Layer > New > Layer. Right now this Layer is blank, but eventually will hold the Line Art.
On the new file, create a new Channel, again for the Line Art. Now going back to the original Line Art file, Select > All and then Edit > Copy. Go to the Line Art Channel of the new file and Edit > Paste. Now Select > Load Selection, and in the Channel selection of the dialog box, choose the Line Art Channel. Also click on the Invert box. After clicking OK, go to the new Line Art Layer you created before. Make sure your foreground is pure black. Now use Edit > Fill with Foreground Color, 100% Opacity and Normal Mode.
You now have a Layer for the Line Art, and a Background Layer that you can use to do the colour work in. Working in the Background will not affect the Line Art. Use whatever tools you want: Pencil, Paintbrush, and Airbrush are commonly used. You may not need the Line Art channel any more. But I would keep it, in case you mess up your Line Art channel by mistake.
You’ll want to match the colour to the Italian comic’s coloured pages. Take a scan of one of them, and use on it Filter > Blur > Gaussian Blur. Set the Radius high enough to remove the screen dots. You can then use the Eye Dropper Tool to select colours from this file.
Modern Reprints : Colour Techniques
Davis Trell (2006/08/09) [#124]:
Some of the colouring in recent recreations, the colours are too lurid.
For the one of the Rawhide Kid (“Two Gun Kid” cover), the colourist even coloured the Kid’s hat yellow! It was really hard on the eyes. Your cover was okay, Harry; it was the insides I disliked.
Some on the Kirby group argued that we are used to seeing old yellows on Kirby’s pages, that are faded, but when first printed they weren’t!
Also the colourists back then had a limited range of colours available, most noticeably in value, and couldn’t overwhelm the black line art. With modern colour the lines seem less important, with the oversaturated colours fighting for attention.
Gregory A Huneryager (“Greg”) (2006/08/09) [#125]:
I agree. The stories need to be coloured with modern paper stock in mind. I prefer the look of “Batman Chronicles” to the Archives, for that reason. The paper on the former is cheaper looking, but it doesn’t hurt to look at.
Another prefered variation was the recolouring on DC’s hardcover “Best of the ’40s“, “Best of the ’50s” books, which was a less white stock with some nice colouring, particularly by Greg Theakston on the Lou Fine stories.
I really think the best way to do it is to photograph the story. I don’t know if that’s more expensive or time consuming, but I like the way it looks in the Marvel “Five Decades” book and the recent Krigstein comics book.
Harry Mendryk (2006/08/09) [#126]:
There is always personal preference when it comes to colouring. Particularly when reprinting Golden or Silver Age comics. The printing technology just isn’t the same. IMHO a bright red on flat paper looks very different than when printed on high quality glossy paper.
But still, common sense should prevail. Rawhide Kid with a yellow hat? Everybody knows that good guys wear *white* hats! As they used to say in old Westerns, when someone would not come out to fight, “yer yellow”.
Craig Ede (2006/08/09) [#127]:
The Will Eisner “Spirit” Archives do the best job matching non-slick paper and colour, improving on the originals.
But, of course, the original “Spirit” didn’t have glossy covers.
Harry Mendryk (2006/08/10) [#131]:
I agree. The Spirit archives are amazing.
Craig Ede (2006/08/09) [#128]:
There was a lot more restoration involved in the Krigstein book than just “phographing” the stories, as the article in the book makes clear. That book is my top choice as an example of how comics reprinted in hardcover should be handled.
Huneryager, Gregory A (2006/08/09) [#129]:
In the Krigstein volume, I’m sure the Marvel “Five Decades” stories — those in the back of the book — were just photographed, and they look great.
It’s amazing how sophisticated some of the old stories are in terms of their colour use, most of which gets lost in the reprint. The early Sub-Mariner stories in “Marvel Mystery” are sometimes quite exquisite, as are some of the Vision. “Marvel Mystery” #13, which has the first Vision, has very interesting colouring on both of them, especially on the clouds and smoke. I’m assuming that Marvel was so small that the individual artists did their own separations.
Marvel should have found some way to do a better reprint of Marvel #1-4. That may be the worst of the archival reprint books.
Matthew Moring (2006/08/09) [#130]:
The guides we were given on the “Captain America” Masterworks volume were pages sourced from the Microcolor microfiche sheets. The colour was way off on them.
On other books such as “The Rawhide Kid“, they want the colourists to follow the same colours as originally appeared in the original issues, albeit with proper trapping.
I agree. I’d like to see a wider range of colours & gradients used, then flat colours, as in the first wave of “Masterworks” from the 1980s.
Harry Mendryk (2006/08/10) [#132]:
Not all of that “Captain America” Masterworks volume, because I supplied them with good scans of “Captain America” #2.
Matthew Moring (2006/08/10) [#133]:
Some of the pages were fine, but most weren’t. I worked on a Tuk story for it (might have been the one in #2), and that was among the cleanest, easiest to restore stories I’ve encountered: good quality scans.
Harry Mendryk (2006/09/22) [#136]:
So far all the comparisons I have made between the original comics and the “Masterworks” volume, for Captain America, show that Marvel has done a great job in keeping to the correct colour.
The line art for “Captain America” #1 does not appear to be based on bleached comic pages like the rest of the volume. I have compared them to copies of the flats (a type of proof that uses line art and no colours) that Joe Simon has. They appear to be an almost perfect match, and do not show the type of blurring that occurs due to the original primitive printing techniques used.
The “Captain America” Masterwork volumes seem accurate and are great buys. I doubt many on this list could afford to buy the original comics: I do not have them! My only complaint is that I dislike the use of glossy paper for Golden and Silver Age comic art. I much prefer the flat paper used in DC’s “Spirit” volume, which I consider the gold standard for reprints.
Greg [Theakston] (2006/09/23) [#137]:
That book was photographed from flats in the Jack Kirby Collection. Wish I’d been using the computer to retouch the rest of the volumes, but those were more primitive times.
Note on other Methods
Greg [Greg Theakston] (2006/02/01) [#86]:
1. Destructive methods (Painting Covers)
There are times when the chemical approach simply won’t do. Bill Black asked me to convert a Frazetta GHOST RIDER cover for him, and the black plate just floated off of the page. Whatta mess. Ditto on Harvey covers of the same period. And Atlas.
Those covers were printed on “clay-coated paper.” A low grade paper, coated with a fine layer of clay for a gloss finish. Cheaper I suppose, but a pain in the neck for me. Charlton used clay-coat as well. When the paper gets wet, the clay-coat lets go, and the result is a mess.
Maybe I should have sun-bleached them, as I did during the 1970s, but that’s so time consuming, unless it’s summer. I’ve been searching for the perfect process for 30 years.
These days I paint bucket white for results, on covers, but it takes forever!
Alex Toth Reader Vol.2 is on the newsstands this week. I took great care in reconstructing the Ben-Day patterns on CRIME AND PUNISHMENT #66. Hours and hours spent unclogging lines, and reconstructing patterns.
Destroy a comic? HAW! I’ve done that to $150,000 worth of comics. As Spock said, “One must die, so that many will live”. Or, as Walt Simonson said about Theakstonising, “You gotta break some eggs to make an omlette.” F.Y.I., I use as low grade copies as I can get.
2. Non-Destructive method (Tracing)
Next up, RAWHIDE KID #24 for Marvel. Seems the proofs and film are missing. Gawd those covers are a bitch. I’ll probably re-ink it on vellum. Short-cut method used on some of the “All-Winners” covers, and interiors. So much faster to just trace them off at 300% than scrub, and scrub, and scrub.
I believe that it’s important for retouchers to understand how the inker worked, and his intent.
< finis >
Feedback – Contact Me :
Stephen Poppitt's Blog-o-Sphere 