Comic Book Plus Forum
Help And Support => Digital Comic Help => Topic started by: Vangald on June 22, 2011, 02:38:56 PM
-
I know that the most common way of going about a restoration is usually based on healing what already exists through fixing discoloration or rips and smudges in the original scan but is there any projects out there going about recreating the image from scratch?
My thoughts were to remove all color information from a page and re-ink over the existing black inks on the scanned page using a digital illustration program. Then going back and using as close to the same colors as they were printed to recolor the page. Of course through the process I would make sure that the images fit uniform dimensions as well since many scans have a portion of the books images scanned/saved at a different resolution. Then after the process is done maybe go back and add the printed dot pattern to the page.
Is there a project for this already in existence? If so can anyone direct me to it?
Thanks ahead of time for any information.
-
I know that the most common way of going about a restoration is usually based on healing what already exists through fixing discoloration or rips and smudges in the original scan but is there any projects out there going about recreating the image from scratch?
My thoughts were to remove all color information from a page and re-ink over the existing black inks on the scanned page using a digital illustration program. Then going back and using as close to the same colors as they were printed to recolor the page. Of course through the process I would make sure that the images fit uniform dimensions as well since many scans have a portion of the books images scanned/saved at a different resolution. Then after the process is done maybe go back and add the printed dot pattern to the page.
Is there a project for this already in existence? If so can anyone direct me to it?
Thanks ahead of time for any information.
Certainly sounds like a worthy project. I bumped up your downloads so you can download good entries for such an endeavor.
B.
-
Oh wow. Thank you. ;D
-
Ok. Just want to throw out my experience with my first attempt at doing this. I took the cover image from centaurs man of war #1 and basically desaturated it and adjusted the levels until I could get good clarity while making the black and whites on the page pop out. Then I saved that page as a new copy and popped open inkscape. I then tried to use the bitmap to vector tool. To get a clear image I had to set the settings for grays. Then when I hit ok it made so many vector points that when it tried to display them all the program crashed. :D
So now I need to figure out how to reduce the information from the image going into inkscape while reducing the number of generated vectors. Also with that particular image there was lighting coming in from the top of the book changing the color values for the top of the book.
Now you may wonder why I was doing things this way. The idea is to make as much of the process automated and then go back and finesse the vector inks till they looked good before flattening the image and bringing it back into gimp for the color process. I will keep playing with this method for a bit to see if I can get it to work. However, the closer I can get an image to just black and white color info in gimp the easier this process will be. If anyone know of different methods of approaching that I would appreciate the information.
Also if this looks like a jumbled mess of text then chalk it up to my scatterbrained nature or the fact that I am toying with this at 5AM. ;D
-
Ok. Just want to throw out my experience with my first attempt at doing this. I took the cover image from centaurs man of war #1 and basically desaturated it and adjusted the levels until I could get good clarity while making the black and whites on the page pop out. Then I saved that page as a new copy and popped open inkscape. I then tried to use the bitmap to vector tool. To get a clear image I had to set the settings for grays. Then when I hit ok it made so many vector points that when it tried to display them all the program crashed. :D
So now I need to figure out how to reduce the information from the image going into inkscape while reducing the number of generated vectors. Also with that particular image there was lighting coming in from the top of the book changing the color values for the top of the book.
Now you may wonder why I was doing things this way. The idea is to make as much of the process automated and then go back and finesse the vector inks till they looked good before flattening the image and bringing it back into gimp for the color process. I will keep playing with this method for a bit to see if I can get it to work. However, the closer I can get an image to just black and white color info in gimp the easier this process will be. If anyone know of different methods of approaching that I would appreciate the information.
Also if this looks like a jumbled mess of text then chalk it up to my scatterbrained nature or the fact that I am toying with this at 5AM. ;D
I know nothing about image manip so forgive the stupid question: isn't there a way to tell a graphics program to just change an image to black and white, or for every line that isn't black, make it white?
-
Its easy to make an image black and white but I don't know how to get just the straight blacks without alot of details being lost. Alot of the times it will shrink the black for one reason or another. But I may be doing it wrong. :P
-
Going to try this method in Gimp and see what happens.
http://www.teamphotoshop.com/articles-Quick-photo-to-lineart-%28with-hundreds-of-variables%29-5,8,168a.html
^
No luck.
So since I can't seem to get the image down to just pure black and white without any intermediate grays without loosing a great deal of data, I am just got to take the desaturated black/white/gray image and trace over it from scratch. :-\
It is a shame I can't get the auto trace feature to work out as it would help save a great deal of time. Oh well.
Also if anyone has a suggestion for the final size of the images that go into the final .cbz/.cbr please chime in. I would like to have an idea size that is most reader friendly.
-
1280 in width is a good starting point. It's what I save my scans at when finished.
Geo
-
pure black and white is called bitmap . It's zeros and ones, on or off, and unless you're doing VERY high resolution (1200 pixels per inch or more) scans, you're going to end up with jagged edges:
|_
|_
|_
as the program tries to decide what is an edge and to turn the angled edges into smooth lines. If you go vector, rather than pixel-based, you're either going to end up with tons of vector points to capture the detail or fewer points and less detail.
The devil is always trying to automate the decision of what is black and what is white. Too many shades of grey and too much purple and brown colors that read as black. Better men than I have bumped up against this problem. Greg Theakston takes his comic book pages and soaks them in a light bleach to eliminate the color and then scans them at high resolution for his reprint books. Bitmap web images are always a battle between size and clarity.
Good luck and let us know if you succeed. Just be aware that what you're trying to do is NON-trivial.
Peace, Jim (|:{>
-
Bill Black used to use Greg but switched to a digital process of his own. Contact him at AC comics and he may be able to help
-
Well still doing research and experimenting on auto tracing methods while working on tracing inks from scratch. In doing so I came across this which some of you folks might find interesting. http://vectormagic.com/home