Comics/Manga on the kindle - the right way (using PS)

[curstpriest]

OK, using the PDF viewer sucks, the margins are large, and the background is nonwhite. Using mangle is bad, because it dithers images horribly (images have a weird pattern and look sandy/cloudy). If you don't dither, comics take up MASSIVE amounts of space because they are converted to PNG (>1mb each!). We will be using GIF (~150k each).

Now this is a very very very bad/stupid idea.

So let's get into some basics.

The kindle uses a 16 shades of gray, I printed out a screenshot (it encodes it as gif) I extracted that pallet. I have attached this pallet to the document. You will need this.

So now, we need to know the max height of your screen:
DX: 1200 x 824, so we will use 1200
K3: 800 x 600, so we will use 800

Extract your comic book/manga (cbz/cbr rename to .zip/.rar and extract with winrar to some directory)
Now open ONE SINGLE IMAGE in photoshop, this will be a test image for recording an action, you will only need to do this one time EVER.

-The first thing we need to do, is go into Photoshop, and record an action.
-Window -> Actions
-Click on the bottom right of the pane (new action)
-Name your new action something like "create comic"
-Now, click the record button

*note everything you do will be recorded, if you mess up you can delete individual actions, but "undo" won't be reflected in the recorded actions.

-Go to image -> image size, chose 1200 for dx, 800 for k3, select (bicubuc sharper).

*If you need to zoom/pan you can change the WIDTH to 1200 for DX and height to 800 and go landscape but the viewer will probably crash and I do not reccomend it.

-Go to filter blur -> smart blur, chose 6 radius and 6 threshold, high quality, normal mode.

*This will remove artifacts from image compression, and allow us to convert down to 16 colors without creating artifacts. If it causes too much loss of detail, or if you have really high quality scans, reduce these values accordingly, but 20/20 or 15/15 are very small and you should not notice these values in the end and they should not degrade readability of text.
*Also at this step you can add sharpen filters in or whatever else you want if you need extra post processing, image brightening, or whatever you want.

-If you want to control dithering for better results go to http://www.ximagic.com/g_download.html and get the free Ximagic plugin. The image quality will be greatly improved. Find a dither to your liking.

-Go to image -> mode -> index color. Under palette select custom, and then chose LOAD, use the attached file kindlecolors.act.
-Enter 16 colors, NO TRANSPARENCY, no dither.

-Go to file -> Save As -> Compuserve GIF -> Normal

Now You can stop recording your ACTION

Your action should look like this:


Good job! Now you are set for evar!

Go to File -> Automate -> Batch
For action, select Convert Comic
Chose the directory of the unzipped images you want to convert
Select a destination directory to export to, or select (save and close) to put the gif images in the same directory as the original images.

Here is an example of a hi-res cover I found @ amazon:
Resized IN PS

Coverted IN PS

Converted IN KINDLE Viewer


For more help with making actions go here:
http://www.creativemac.com/2002/04_a...ls/psbatch.htm


Plugin Dithering Samples:
Threshold (this is more or less the method kindle uses)


Dispersed 8x8 <-- This is the pattern that I am using, because I believe it to look the best!


Checks


Random method (it's its own mattern, not just a random of the other options)



You really need to try and see these images on your kindle. They look A LOT DIFFERENT than they do here. The kindle does not show low shades of gray very well for example the clouds are nearly white on a computer monitor, but on the kindle they are dark grey + white, and you can see a crosshatch grid with some dither patterns. Test them out, fiddle with settings. Regardless, using any methods above will get you amazing results.

White backgrounds WILL BE WHITE, text will be crisp and readable, and images will look a whole hell of a lot better.

Edit: Also noticed another good option, the Error dithering option in the plugin is quite good if you are getting two tones/patterns on page backgrounds and don't want to use threshold mode. You can eliminate these patterns by using error dithering instead of threshold dithering, and using a small value such as 30-40%. It's either this, or go into thresholds and up the brightness by merging lightness together a bit.

As I said earlier, the kindle doesn't handle display of off-white well, so killing them by editing levels will make force it to use those 16 colors of gray on better things...

[curstpriest]

Oh now an additional comment before everyone chimes in. I realize that the image isn't going to be 1:1 to the original. Realize that the kindle does not dither ~ the same as you may expect. If you like dithering, configure it. I think kindle dithering looks bad, so I don't use it. Different types of dithering may give you better image quality depending on the media/content you are viewing. You can tweak smart blur as I said and dithering to get better quality/artifact removal. However. The kindle uses a hybrid dithering system, where it does not dither, but it uses a high threshold for colors/contours to exaggerate detail. Using the image manipulation techniques here I believe you can get a very similar result to putting in a source image (unscaled, full color) and taking a screen shot with the kindle (alt+shift+g) to produce a gif. Compare that GIF to this converted image and you will see that they are very much the same.

I still need to tweak it a bit to get more detail in general, but I think the palette is correct. Any feedback will be helpful.

The kindle still does a Slightly better job at converting a color image to 16 shades of gray, but I think that's just because I don't have the filters set up correctly.

*Edit - with the PS plugin you will get about the same dithering pattern as the kindle, better quality if you tweak it.

Denoise+Sharpen filters will give you some interesting results on line art. You can significantly sharpen/increase readability and contrast on simple art with this. Also tweaking LEVELS can help achieve better contrast with some black/white works (manga).

Edit: OK found a free plugin that does exactly what I need will update guide after I get it working and example:
http://www.ximagic.com/g_download.html

Here is what Dispersed Dot Bayer 8X8 looks like after these transformations int he picture viewer on my DX

[curstpriest]

Another thing I have noticed using the built in image viewer -
Using gifs/smaller image files does not seem to stall/crash the kindle.

When I was using raw JPEG images the kindle was crashing every 5-10 images.
With rescaled ping images it was crashing every 20 or so, or when I would switch from fullscreen, or rotate, or do any strange manipulations.
I have not yet experienced any crashes when working with these 100-200k 16 color gif images. I don't know if it is their small size or format, or perhaps just blind luck, that makes the image viewer appear to be more friendly with them.

They also load significantly faster than the larger files, but I assume that is because the device isn't working hard to scale them down and dither and such. Also note that you can go to menu, and turn dithering off. You don't need it. The images are already dithered and set to display properly on native kindle display.

[badbob001]

For giggles, lets compare your images to the ones created by:
Mangle (http://foosoft.net/mangle/)
Canti (https://www.mobileread.com/forums/showthread.php?t=75628)

I used the first color image as the source image and processed it through each respective app at DX and K3 sizes and with various settings. Mangle has an option to turn off dithering but it turns out to just disable conversion to grayscale. Canti defaults to 256 colors (grayscale I think) but you can also set to a lower amount such as 16 in my tests.

The Kindle viewer does support dithering but I turn it off since it slows down the viewing of images. You should also make sure all your images fill up the resolution of the screen since viewing images of different sizes will result in annoying after-images. The add border option in these apps resolves that.

Mangle_DX.png


Mangle_DX_NoDither.png

Mangle_K3.png


Mangle_K3_NoDither.png

Canti_DX_256.jpg


Canti_DX_16.jpg

Canti_K3_256.jpg


Canti_K3_16.jpg

I took all four K3-size images + your Dispersed 8x8 image and viewed them on my K3, full screen and with dithering disabled.

• Mangle_K3_NoDither: Outlines of characters are black and sharp. Obvious color banding on face. Sky is grayish and smoke is lightly visible. Castle is detailed. Text is sharper.
• Mangle_K3: Outlines are grayer and overall image has a blurrier look. Face has smoother color gradients but overall is darker. Sky seems whiter but could be an illusion since the smoke is now darker due to dithering. Castle is gray-filled with little detail. Text is good.
• Canti_K3_256: Similar to Mangle_K3_NoDither, but outlines are slightly less dark, grandients on face is better but complexion is a little darker, sky is similar but smoke is darker, and castle is detailed but a little less sharp. Text is slightly blurry.
• Canti_K3_16: take the canti_k3_256 image and make it darker / have less contrast. Face gradients is okay but probably masked by the overall darkness. Sky is much grayer and smoke is hardly visible. Text is slightly blurry.
• CurstPriest's Dispersed 8x8: viewed it on the K3 without downsizing it beforehand. Similar to Mangle_NoDither but slightly less contrast. Gradients on face is better but complexion slightly darker. Text is good.

I would rank them as such:

1. CurstPriest's Dispersed 8x8
2. Mangle_NoDither
3. Canti_256
4. Canti_16
5. Mangle

The top three are pretty close and 'better' is subjective and depends on what you favor: good contrast but banding vs good gradients but darker

What is clear is that the default settings for Mangle are pretty bad for color images. It may be tuned for high-contrast black and white manga instead of full color western comics.

For now, I prefer Canti because it's a lightweight app that can process folders of comics at once. The author is trying to add auto-level and auto-sharpening options so maybe it can get better results. My K3's screen is probably too small for me to sweat over the tiny details anyway.

[curstpriest]

Like i said, if you want to change brightness or levels (image -> adjustments -> brightness+contrast / levels), just do so where I say to for post processing, add the adjustments, and they will be done in the profile for batch mode. Also, you can do entire directories+subdirectories with photoshop batch mode

/comics
/comics/xmen/
/comics/xmen/1/
/comics/xmen/2/
/comics/HOD/
/comics/hod/1/
/comics/hod/2/

Now when you select batch mode, simply select include all sub directories. It will process them all.

Also note that these images are native rez for the DX not the K3, so taking these and putting them on your k3 won't give you good quality, you need to convert them properly to k3 native rez. You can't do that by resizing an already dithered image down to the k3's native rez, you lose a lot of quality (if you view them full screen, if you view them at their native rez (c i think) you'll get an idea of what they look like). =X

[curstpriest]

Here is a basic dark/sharp enhancement:

Levels (midtones darker)
Edges sharpened



I still can't really see the smoke on my kindle, unless it refreshes incorrectly and it catches the ghost =X
I think that shade of gray is so faint that it's almost white

[badbob001]

Originally Posted by curstpriest

Also, you can do entire directories+subdirectories with photoshop batch mode

Good point and batch mode does allow you to specify a target directory so you're not altering the originals. But alas, I'm too lazy to extract all my cbz and cbr's.

I mainly view non-color manga on my kindle. I hear that color comics are usually too dark.

[curstpriest]

Yeah, you need to lighten mid-tones and sharpen if you use some of the darker/marvel comics, but ps does a good job of that.

I noticed that canti/mangle produces a strange banding effect in the backgrounds of images that is quite annoying to me anyways. Also mangle always sets itself to kindle 3 and won't save its preferences so I just deleted it and use PS instead. The PDF reader does it on the kindle too for some reason. I like to see plain white or w/e.

But using Photoshop works good, you just have to get it set up and then it takes like 2 seconds to do your conversions. Touch ups aren't too hard to do either.

[el_brio]

I read comic books when I was 5.

[curstpriest]

Originally Posted by el_brio

I read comic books when I was 5.

I used kai's power goo when I was 5

[Rinzwind]

Excellent post and I will keep an eye on this as my kindle was suppose to be my book reader mechanism with an option for manga viewing.

Photoshop is not an option for me. I guess that should also be possible with say... Gimp or ImageMagick but those are not easy accesible programs
Any chance in someone typing a small how to in doing this with either of the 2?

[NiLuJe]

Let me just say that I wholeheartedly approve of your reading material! \o/

And, yeah, for those of us using GIMP, I usually also prefer the 'Positionned' dithering alg, which more or less looks like the 'Dispersed' method shown here.

[curstpriest]

Felicia Day is amazing.

[el_brio]

Originally Posted by curstpriest

I used kai's power goo when I was 5

I'd rather not talk about your yaoi/hentai collection thank you. Heh. Just kidding of course.

[badbob001]

Originally Posted by Rinzwind

Photoshop is not an option for me. I guess that should also be possible with say... Gimp or ImageMagick but those are not easy accesible programs
Any chance in someone typing a small how to in doing this with either of the 2?

Canti uses ImageMagick, but it certainly would be easier to fine tune to your tastes if you use ImageMagick directly.