Scanning Process

I've created this page to give a little bit of insight of how my scanning process works, so if you always wanted to know, here's your answer!

These example pages are from Strawberry Panic v01. Just because I use them here, doesn't mean I'm gonna scan it. Go read the scanlation from Dynasty Scans if you're interested.

Alright let's get this going. First I have to scan, obviously. The pages come out like this:
Even pages are upside down, uneven pages are normal. The reason for this is, I use a scanner with a special book edge, so it only scans one page at a time but it scans all the way to this edge, so the spine shadow is hopefully very small or non-existent. After scanning one side I'll flip the book over 180 degrees and scan the other.
These pages are completely unleveled and normally 3492x4956px large, always exactly the same (I believe I scan in an A5 area) and saved directly as uncompressed TIF-files. Grayscale pictures are 16.5MB big, color pics over triple that size, 49.5MB. Exceptions are color double/cover pages and rarely grayscale double pages, both pages scanned at once. Those are over 100MB and 35.5MB big respectively. Scanning a typical 170-200 page book takes about 1h-1 1/2h but sometimes I just can't get in a good "rhythm", always have to check if everything is flat on the scanner surface...stuff like this takes longer and feels even longer...

After scanning I usually run the pictures through one automated Photoshop action. This crops the picture a bit and levels. Regarding the levels, if you scan without any modifications, no auto-levels, auto-brightness, auto-whatever crap, just using the default settings the histogram could look like this. Now I have to find fitting settings. I usually take ~5 pictures from the book as a sample and just play around a little bit. In general there isn't much difference between all books so you could get away with using the same settings for many different ones. You have to look out what paper the book is printed on, i.e. are the pages really white or low-white and adjust everything accordingly. In my histogram pic above I've already put the sliders in an alright position and could probably use this on all pages but like I said I'd test these settings on some more pictures before I decide. Usually my black levels range from 80-100 and the whites are between 180-210. Color pics are usually leveled much softer, like 50-230 but I really have not much experience or knowledge on this (and there's quite a bit of bleed-through on my color pics (not covers) but they look ok). After all that the pages look like this:
While still not fully cropped it's pretty good already but the rest has to be done by hand. This "rest" is the small black border/line at the top of the page. The size can vary quite a bit, so I can't do it with an automated action. I've also flipped the pages over on the side. I do this because in the past I always worked on a small 15.6" widescreen laptop monitor and Photoshop always opens the picture in the biggest zoom factor possible, while still displaying the whole picture (well using these zoom steps: ..., 16.7%, 25%, 33,4%, etc.). So it was a huge difference working like this instead of this. While I'm still on a laptop, I now have a second 27" screen so it wouldn't matter too much, still it's just more comfortable.
Now the black spaces are getting cropped by hand, nothing spectacular really and I'm flipping the pages again so they are all right-side up. The pages are now 4X00px high, in this case it's 4250px. No other size matters to me except this one, so I can modify my action which saves the pictures in different sizes (because I don't want to enlarge it to 4400px). Right now I'd decrease it to 4200px just to be on a nice, even number and I don't have to name the files something like "Manga_v01[sg2]4519h" which would be just ridiculous.
The end results looks like this plus an example of a double page:
If there's a double page I just expand the image canvas by 205% to one side and copy/paste the second page (sometimes I just use like 201-202%, when those pages were scanned pretty much at the spine and it would just look weird with a bigger white space in the middle).

Recently I began to scan double pages both pages at a time, like you would on any normal scanner but only those that are at the very beginning or end of the book, where I can keep the spine shadow very small (Yotsuba&! was probably be the first one, so really not too long ago). Here's an example for this:
If I scan those pages by themselves a small area in the middle (where the spine is) would have been cut off (probably about 3-5mm on each side) so it wouldn't fit together as good as this. But if I always do it like this the spine shadow for double pages in the middle of the book would just be too big. Also I don't want to press my books down with all my might to reduce it and probably damage it (I did it in the past but what's the use of a book edge scanner if I did it anyway...)

If you want to know which settings to use for saving files, just search online for a scanlation tutorial, it's what I did. I'm saving as PNG-8 with 17 colors. I don't really know why 17, it just was written somewhere ~.~ the pics look good and have an acceptable file size so I don't see a reason why I should change it. Color pages are saved as JPG and right now I'm using medium quality, 50 I believe. I came to this through some testing and I don't see any artifacts, so it's good enough too. In the past I may have used some higher quality, whatever 9 or 10 is if you use the normal "Save As..." and not the "Save for Web & Devices..." but it just isn't necessary for the stuff I'm doing.
I don't use any filters, plug-ins or other programs whatsoever. I used PNGOUT a couple of times in the past, especially on my first 4X00px releases because it crushes the PNG files and reduces their file size about 10% it just takes sooooooooooo damn long which is why I don't use it anymore.
I don't really know anything about filters and from what I've seen from normal scanlations, mostly results of overusing this really popular one, Topaz?, it just doesn't look really good. Maybe some things could be improved if I used something but I don't think it's necessary.

All in all a complete book takes 4-5h, from scanning to the finished upload though about 2h of that I don't have anything to do because either Photoshop is working through some automated actions or the book is being uploaded. This can be shortened if I work on stuff in batches but I'd probably still need at least 3h per book.

Even though this page was created to give people a small insight on how I do my things normally I realize not many people have a book-edge scanner and if you are interested in scanning your own books, chances are you're using a normal flatbed scanner and scan two pages at a time. If that's the case, the following paragraphs might interest you. I'll go over my process on how I'd do things if I still scanned like that or if I'm editing existing scans on the web (e.g. scans from Random Scans/Kickthekitty).

So you scanned your pages and they might look like this. Nothing special, it's not straight or fully cropped. You can either level first and straighten/crop later or the other way around, it doesn't really matter. I'd probably do some batch leveling first because that way I'm working with smaller files (smaller file size) in Photoshop which could possibly be opened faster and stuff like that but that's for you to decide. Straightening and cropping should be self-explanatory, either use the Crop Tool for both or the Ruler Tool to straighten first and then whatever you like to crop (Marquee Tool, Crop Tool, reducing the canvas size, etc.), if you have problems at this point you should probably read some general guides for Photoshop and/or scanlation first. Alright, the page is leveled, straightened and cropped and looks like this:
The spine shadow is pretty big and could have been reduced significantly if I used more force to press the book down while scanning but I just decided not to for this example. Check out scans from Random Scans/Kickthekitty for good examples on how it should/can look. Next is already the splitting of the pages, I have a small action to help me with that so I have to do very little work myself (if you are scanning or editing hundreds or thousands of pages everything that shortens the time you have to do it is great!). I start by selecting one page, on normal right-to-left manga I'd go with the right (the first) one. Now comes the PS action! It crops so you are left with just your selected page, opens a "Save as" dialogue where I usually just add a '-1' to the filename to know what order they should be in, next it goes back one step in your history to undo the cropping, inverses the selection so you have the other page selected, the left one in this case, crops again, opens another "Save as" window where I change the name from 'filename-1' to 'filename-2', closes the file and opens the next page via a script. Just rinse and repeat until you are done or fed up and throw everything into a corner. Your results will look like this:
Now you're either done already and save the pics or do a bit more cropping. The large dark spine shadow in the right picture is unnecessary, you can just crop it.

...and you're done. Congrats! Now you just release the results somewhere and gain the love and admiration of your peers.

I hope this is enough to explain how stuff gets done here but if you have any questions or suggestions, just leave a comment.

Update 09/16/2011: I put some more details about leveling in.
Update 11/09/2011: New section on editing normal "2-page-scans"


  1. thanks this will help alot when I scan my books.

  2. Does this damage the book? I'd like to scan some of my own manga, but I'm worried about breaking the spines.

    Also your scanner, is that a plustek opticbook?

  3. No, the book doesn't get damaged normally. You can push the book so hard against the edge that even with this scanner it's possible to leave traces or damage it but that shouldn't be necessary to get good results in 95% of the time.

    The scanner is an Avision FB2080. I had a Plustek Opticbook for a day or two (dunno which model, probably the 3600 or 3600 Plus) but didn't like the results, that's why I returned it. I know of someone that used a Opticbook and the scans look good, so maybe I did something wrong but it's been too long.

  4. what resolution do you scan pages

  5. 600 dpi and mostly at a fixed size of A5 and crop everything later, so depending on the physical size of the book the height varies from 4100px - 4900px.