Scanning and editing tips

[red exposure]

After finding my scans had slight overkill in the reds, and were sometimes low on saturation, I came to use Photoscape (I've posted a link in the Useful Programs thread). It is an excellent, low-resource freeware for photo editing, which I use for post-scan colour-adjustment issues. The menus are easy to use and comprehensive enough to get what you need, and the auto-settings aren't bad at all (if you aren't yet sure about using colour-graphs).

For colour adjustment in Photoscape, my usual process tends to be:

• Use contrast settings first. I usually find bringing out the colour via contrast actually makes it easier to work on and adjust.
  
• Low saturation scans can be given a boost with auto contrast (there are manual sliders also) settings, as well as the 'depth' function which seems to enrich the images a little. It's also possible to bring out shadow details with the 'backlight' function.
  
• Colour-adjust using the graphs in Photoscape (it allows you to separately adjust reds, greens, and blues). My scans tend to have a reddish finish, so I counteract this with reducing reds in the mid/low areas, and then adding a little luminance, which usually seems to work.

I'm learning as I go (evident from my scans in other threads), but I've found the use of colour graphs is a much better tool than using RGB sliders for any colour-correction issues. The major drawback with Photoscape is that it doesn't use layers, so stitching together double-spreads needs to be done in something such as Paint.net.

[howerd]

Originally Posted by Dekoda

Photoshop does a poor job when it comes to jpeg sizes. They have a 1 to 12 scale, while most other programs have a 1 to 100 scale. If you use PS to save an image in jpeg format, a 9 is the best number to use. Anything over that will just make a bigger filesize, but won't result in a better image. If you save the image with a setting of 12, then compare it to a setting of 9....there will be no visible difference.

There is a reason for this. The better way to save a jpeg is with a program that uses the 1 to 100 jpeg settings. With a program like that, I use a setting of 85. This is called the "Q" or quality. This is derived from the five factors of an image.....the red, blue, green, chrominance, and luminance of the picture. When you save a jpeg, you are compressing the image. That means that you are throwing away some of the information in the picture.

You can eliminate the chrominance and the luminance, since they are a product of the brightness of the colors in the picture. The chrominance and luminance make up 20% of the picture, so if you eliminate that, you are left with 80% of the information in the picture. To play it safe, I add an extra 5% of the information, making it 85% or a quality of 85. Anything over that will not be visible in the picture. Saving it at a 12 in Photoshop will not yield any benefits over saving it as a 9.

I agree that using the descreen function will result in a softer picture, but you can regain that sharpness in PS by using the unsharpen mask. If you don't use the descreen function, you will get moire, and that is hard to eliminate.

I also use a black background to eliminate the bleed through that you find in some magazines. I use a black foam board, but any black paper or cardboard will work.

I appreciate the tips ellias. All input is welcome.

I disagree: use max file sizes because of broadband. Programs like Photoshop allow you to use higher compression because of the dial up internet legacy. These days almost everyone has broadband & large Hard Drives for saving larger files. Any compression at all will have a subtile negative effect on images -some of which aren't always apparent in the small square sample Photoshop displays when it asks you to choose a compression level. I choose 'maximum' every time because maximum image quality is much more important than download speed in my humble opinion. There's far too much shoddy scanning on this website!

[effCup]

I'm having problems with "shadow" from the other side of the magazine paper I'm trying to scan. I've got a sheet of matt opaque brown paper to use as "backing", plus books to sit on top, but some scans still have "shadow" from black text or white areas on the other side. Is it worth trying to fix in, say, Photoshop? How?

Also: does anyone else use Photoshop's transform-skew to "square"-up scans? I've just tried a couple and it's fiddly/time-consuming--probably just because I'm a novice with graphics apps.--but I was pleased with the results. The main thing to beware of is if the image contains text--i.e. text is rectilinear so shows badly with any skew beyond marginal levels.

[Darth Joules]

Originally Posted by effCup

I'm having problems with "shadow" from the other side of the magazine paper I'm trying to scan. I've got a sheet of matt opaque brown paper to use as "backing", plus books to sit on top, but some scans still have "shadow" from black text or white areas on the other side. Is it worth trying to fix in, say, Photoshop? How?

I use card, rather than paper. Thin card from an art shop. It might be worth trying a different shade of colour card. I mostly use a dark choclate brown sheet, but also have use four other shades sometimes: black, dark gray, mid-grey, and light brown. If the reverse is still ghosting through, it could be the paper stock of the magazine is too thin. I get this problem somtimes with cheaper mags and there's little to prevent it from happening. Fixing it requires some time consuming editing in Photoshop layering/painting/cloning, but most of the time it isn't worth it.

Also, don't put too many books on top of the magazine or the scanner's lid or you could crack the glass platen.

Also: does anyone else use Photoshop's transform-skew to "square"-up scans? I've just tried a couple and it's fiddly/time-consuming--probably just because I'm a novice with graphics apps.--but I was pleased with the results. The main thing to beware of is if the image contains text--i.e. text is rectilinear so shows badly with any skew beyond marginal levels.

Use Edit > Transform > Rotate. Typically if I'm matching up two halves, the first half will be squared up using Rotate Canvas > Abritary. Then the canvas will be expanded to make room for the second half. The second half is copied and pasted onto the first, so it's on another layer. The second half is moved and rotated until it lines up with the first. When you select Rotate there's a crosshair in the center of the image you're about to edit, this can be moved to anywhere within the work space and acts as the pivot point.

[effCup]

Originally Posted by Darth Joules

I use card, rather than paper.
[..]
it could be the paper stock of the magazine is too thin. I get this problem somtimes with cheaper mags and there's little to prevent it from happening. Fixing it requires some time consuming editing in Photoshop layering/painting/cloning, but most of the time it isn't worth it.
[...]
Use Edit > Transform > Rotate. Typically if I'm matching up two halves, the first half will be squared up using Rotate Canvas > Abritary.
[...]
When you select Rotate there's a crosshair in the center of the image you're about to edit, this can be moved to anywhere within the work space and acts as the pivot point.

Yes, I tried some card and it made an improvement for some but not others re. ghosting. I agree, it's the thin mag. paper.

Re. skew, I was meaning its use *after* any rotation. I don't get crosshairs for rotate--must be my older version of photoshop (cs2). I add guide(s) against which to align the image, for rotate or skew.

[Darth Joules]

Originally Posted by effCup

I don't get crosshairs for rotate--must be my older version of photoshop (cs2).

Using the same version. When the layer is selected for Rotate and the box selection appears around it with the 8 little boxes (handles) around the edges, there should be what looks like a little tiny sniper sight in the center. That's the pivot point which you can move by holding down right click on it and dragging it about.

[effCup]

Here's two versions from the same 2 scans. It's a 2-page image & there's clearly problems with different sides of my (cheap) A4 scanner bed/mechanism creating varying tones/textures, etc. Suggestions for how to fix that would be great, but here I'm particularly interested to read suggestions for how to avoid the moire pattern effect.

Both pages were scanned to tif, 300ppi. One was converted to jpg using IrfanView batch process, the other in Photoshop, roughly halving original scan height to 1600, each set to 80% or 8/10 "compression"/"quality", i.e. not very compressed/"high" quality... yet resulted in quite different file sizes(?).

I would have thought the former would be fine but apparently not--or am I doing something (probably obvious) wrong? Colour banding (upper sand) and ghosting from reverse image showing through are unfortunate artifacts of the paper/print. I tried also reducing resolution to 150ppi but it didn't seem to make any (readily apparent, to me) difference:

Any suggestions where I'm going astray? (new glasses, you think?)

[haVEFun here]

Maybe You have already tried these:

Ghosting: Was scanning performed with Matt Black cardboard placed behind ( on top of ) the page that is to be scanned ? That thicker matt black paper that maybe obtainable in cardshops/ office articleshops?

Moire: Is it possible to set the scanner at low speed ? The slower speed at scanning gives less interference of light, therefore less "banding".
or does the software that steers the scanner contain a moire reducing program?

Was a heavier book ( besides the matt black paper) placed on the scanner? Sometimes the page is not pressed good enough onto the glass of the scanner ( do not overdo that ; a book with approx. 200/250 pages will do just fine )

The initial scanning is the the most likely place to gain performance, measures afterwards probably will help just only a little i am afraid.

Hope You can do something with my suggestions,

Awaiting Your reply, may take some time to reply again.

With kind regards, haVEFun here.

[effCup]

Sorry, I should have been clearer before: I don't think there's any moiré in the original scan, only the mag's halftone "screen" pattern/speckle--hope I'm using the technical terms here in the right, or at least a non-confusing, manner? The above moiré seems to be a creation of IrfanView when it is used to batch-convert the 300ppi tif to jpg. The obvious solution: use Photoshop instead to convert, but it seems peculiar that this step in IV creates a moiré.

Re. the scanner: Epson Stylus CX5900 printer-&-scanner. It has both "descreen" and "unsharp mask" filters--the latter I leave on "medium" having found no noticeable improvement on "high"... but perhaps I didn't experiment enough? The descreen filter has a "screen ruling" setting which I've left on "magazine (133 lpi)". I tried it once on "fine prints (175 lpi)" but no noticeable improvement... and I suspect nor should it have.

There are also settings for histogram adjustment, tone correction, image adjustment, and colour palette, but I leave it on auto exposure. Such tweaks seem very complex & beyond my limited knowledge/ability. I'd rather not spend "hours" tweaking such widgets for each scan, and I don't know what would be an appropriate "base"/normal set of settings to apply uniformly, so leave it on auto.

Colour/tone issues previously encountered seem more due to the scanner mechanism--i.e. different sides/edges of the scanner bed/plate/mechanism. I would not have thought those were correctable with these settings, but that may just be my ignorance?

My scans are (now) virtually all done with the lid removed & a matt brown piece of paper which is reasonably thick/robust (a former paper bag from a wine shop, as it happens) plus 2 books side-by-side to provide full "coverage" of the plate, & sometimes a third book if suitable/necessary--the aim is to ensure flattening near the mag. spine rather than simply more weight per se. The ghosting happens despite that, with some (thin) mag. paper types, although usually it would be worse without the brown paper.

I also have a similarly matt brown piece of cardboard to use instead of the paper, such as for some mags. that have had their spines "de-stapled", but it cannot be used with other mags. because of bending/creases near the (e.g. glued) spine, caused by the card's thickness.

Originally Posted by haVEFun here

The initial scanning is the the most likely place to gain performance, measures afterwards probably will help just only a little i am afraid.

I think you're right. I shall try experimenting further with the settings/options, but there is no way to slow the scanner as such... well, perhaps other than upping from 300ppi, which--not wanting to start an argument--again I suspect won't achieve much in terms of quality, just "bigger" files. I did try it, but remain sceptical as to its value.

Thanks for your further, well-considered & useful suggestions.

[effCup]

By which I mean curves, rather than ribbons.

I was having problems with joining "bowed" images--i.e. one (or both) of the two halves of the image had a slight concave or convex curve along its "to-join" side, making it difficult to join them together because it left gap(s) along that centre join-line, typically containing distorted-hue (i.e. highly visible) "edge" pixels such as bright oranges/rainbows, etc. Image rotation and Photoshop's Transform -> Skew tools do not fix this, though they do help straighten things beforehand, while the Filter -> Distort -> Lens Correction is too "gross" or "coarse" an effect to be useful here. Likewise with the Pinch filter.

Just now, however, I've discovered that Filter -> Distort -> Spherize has quite fine control--e.g. right down to just 1% effect, and can be made horizontal-only. For an image scanned at 300ppi, filter effects of ~2-4% can adjust for bows of ~1-3 pixels width, i.e. just enough to be pretty handy. The only possible loss is a couple of pixels on the "opposite"/outside edge, but that's often just innocuous "background".

/shrug/ Might be helpful for others to know.

Late addition: spherize does not affect the outermost row(s) of pixels in an image, i.e. use that tool before cropping the image... or paste it into a larger canvas prior to using the spherize tool.