FBInk: A small tool/library to print crap on your screen!

[NiLuJe]

And finally released v1.0.0 .

Mainly making sure Mark 7 devices are handled correctly, and tweaking a few things for KOReader & KFMon .

[NiLuJe]

Released v1.2.0 .

This one adds a major feature: displaying images! Most common formats should be handled, and you can adjust the exact positioning in all kinds of crazy ways .

As usual, for best results, properly dithered PNGs are your friend .

If you have access to ImageMagick v7, my current rescale + letterbox + grayscale + dither pass of choice looks something like this:

convert input.png -colorspace Lab -filter LanczosSharp -distort Resize 1080x1429 -colorspace sRGB -background black -gravity center -extent 1080x1429! -grayscale Rec709Luminance -colorspace sRGB -dither Riemersma -remap eink_cmap.gif -quality 75 png:out.png

1080x1429 being the effective resolution of an H2O .

With the cmap attached here.

EDIT:

An approximately 10 times faster alternative, if you *really* need on-device processing:

convert input.png -filter LanczosSharp -resize 1080x1429 -background black -gravity center -extent '1080x1429!' -colorspace Gray -dither Riemersma -remap eink_cmap.gif -quality 75 png:out.png

[NiLuJe]

Released v1.2.5 .

Expands on the image display feature of v1.2.0 by adding proper transparency support (i.e., if you feed it an image with an alpha channel, it'll do the right thing with it) .

You can opt to disable this via a flag, because it comes with a bit of a performance penalty .

[NiLuJe]

Released v1.2.6 .

A few code cleanups & tweaks, and a fix for potentially broken behavior at high font scaling values.

[NiLuJe]

Released v1.3.0 .

This one comes with a crapload of changes .

I won't rehash the changelog, but, basically, everything's faster, we've got some new fonts to play with (c.f., attached screenshots), and there's even more control over image & text positioning.

This is also the first real introduction of the button_scan thingy we've talked about in the kobo-rclone thread, but that'll be the subject of a dedicated thread later tonight .

And, just for shit'n giggles, there's now a linux target to bypass a few eInk specific checks, so that you can use this on any kind of Linux fb without spurious warnings (c.f., a snapshot from my desktop's EFI framebuffer ).

[NiLuJe]

And released v1.4.0, a bit earlier than I initially planned, because there was a stupid build issue in 1.3.0 that basically broke this Kobo package... (thanks to @codex-hex for the heads up).

Which means you get the performance improvement in text rendering a bit early, as well as the pixel-perfect text positioning options .

[sherman]

For those who want a Go wrapper, I have just updated go-fbink to support FBInk 1.4.0

[NiLuJe]

Released v1.4.1 .

With a few fixes to the X/Y offset features that got in a bit too early in the rushed 1.4.0 release .
Also stdin handling on the CLI tool, and lowercase glyphs in the block font .

And in Kobo-land, button_scan *should* now work neaaaarly everywhere (the H2O²r1 is still out, while the Aura Edition 2 *might* work, at least the original hardware revision).

[NiLuJe]

Released v1.5.0 .

Adds support for custom pen colors (both fg & bg), as well as drawing progress bars!

[sherman]

Originally Posted by NiLuJe

Released v1.5.0 .

Adds support for custom pen colors (both fg & bg), as well as drawing progress bars!

Hmm. Progress bars could be useful...

Anyway. go-fbink has been updated again to 1.5.0. Although I haven't tested it, so YMMV...

[NiLuJe]

Released v1.6.1 .

Minor bugfixes around the progressbar feature, but mainly in Kindle-land .

[NiLuJe]

Released v1.6.2 .

Some more tweaks/fixes related to progress bars .

[NiLuJe]

Released v1.7.0 .

A bunch of changes in this one, most notably one trying to make row 0 more usable, in particular when there's empty vertical space left (i.e., when we can't fit a final line on screen because its height is larger than the space left).

In the same vein, this one finally honors the crazy viewport on the H2O, with its 11 top pixels hidden behind the bezel.
I'm fairly sure the same sort of thing should apply to the Aura, and possibly the H2O², but I'd need an fbgrab screenshot of Nickel to confirm that.

There's also a bunch of new stuff around button_scan handling that @sherman will make use of .

[frostschutz]

Originally Posted by NiLuJe

In the same vein, this one finally honors the crazy viewport on the H2O, with its 11 top pixels hidden behind the bezel.

More like 9¾ on my H2O. When there is an all black image I always see a white stripe at the top since that part usually always stays white.

But that might be because I had it opened up and all and things might have shifted around a little.

It's just impossible to handle. In a picture viewer it might make sense to just draw-repeat the first line of the picture into the top area just in case it's peeking out for someone else too

[NiLuJe]

@frostschutz: Yeah, even on my unopened one, the fit of the top edge of the bezel/IR array is atrocious :/.

11px is the value Nickel itself blacks out, to avoid any possible nitpicking on that front (i.e., I shift the blame to Kobo ;p).

That works out fine on my end, I can maaaaaybe see one or two of those if I look at it at a reaaally shallow angle (like, err, when it's sitting next to the keyboard during tests ;p). . But then you start to see fairly weird crap everywhere at that angle, plus the reflections off the IR casing .


In any case, that fudging around is togglable, to restore the previous behavior .

----

EDIT: Stupid idea of the night: one could use the nightmode hack as a skeleton on which to base something to forcibly black out those pixels in Nickel: interpose updates w/ top=11 & height=1429, and possibly add a bit of targeted memset if Nickel also skips writing to those entirely, and not just refreshes .