Native FBInk: Another small tool to print crap on your screen!

[PoP]

Originally Posted by NiLuJe

@PoP: Just to double-check: did it ever work as you wanted it with earlier versions?

EDIT: Okay, tried it, no, don't think that's doable, that's an artifact of how eInk handles the update. I'll see if I can fiddle with different waveform modes .

EDIT²: Yep, it's a byproduct of GC/GL waveform update modes . And those that don't work like that do not support so 'many' colors .

No, never tried with earlier versions.

Waveforms: Ah, thanks for checking, no problem, eink's animations is not their forte anyway

[NiLuJe]

Released v1.14.0!

Quite a few front-facing changes this time around...

On the CLI, the various wfm= suboptions have moved to a global -W, --waveform flag. This means that the syntax for printing images and doing screen refreshes has been altered a tiny bit. On the upside, this is now honored *everywhere*, so you can, for instance, enforce A2 updates when printing text.
Because we can, in *addition* to the existing software inversion support, you can now toggle the eInk controller's hardware inversion (i.e., "nightmode"). Really just because we can, as there's almost no performance impact for our current use-cases. Not available on legacy devices (< K5).
Performance increase when printing images.
Performance increase when decoding images (especially JPG/PNGs). If you suddenly encounter broken/buggy image rendering, give me a shout.
Don't crash the CLI when passing a suboption without a value, emit a proper syntax error instead.

See the release notes for all the gory details .

[Sol Arkite]

Originally Posted by NiLuJe

Released v1.14.0!

Because we can, in *addition* to the existing software inversion support, you can now toggle the eInk controller's hardware inversion (i.e., "nightmode"). Really just because we can, as there's almost no performance impact for our current use-cases. Not available on legacy devices (< K5).
.

Does this mean an "invert screen mode" Kual extention can be made from this concept?

[NiLuJe]

@Sol Arkite: Not really, no, the inversion flag is on a per-update basis, it's not an instruction to ask it to invert *every* update until someone asks it not to.

If you *really* wanted to, you'd have to go the interposition way, like the Kobo nightmode hack.

There may be a way to tweak the ligl awesome stuff on FW 5.x, too. Hell, it may already be supported but hidden away in an obscure debug-only feature, like there is to disable REAGL.

[parkerlreed]

I made a thing do a thing




Custom Python script to push a section of the screen to the Kindle and use Fbink to display the image

KindleTermPV connecting to netcat to send the input events back

And Android x86 running in QEMU with video=600x800 passed on the kernel command-line

screen.py

import sys
import time
import paramiko
import pyscreenshot as ImageGrab
import signal

hostname = '192.168.86.125'
password = 'pass'
username = 'root'

def sigint_handler(signum, frame):
    print('Exiting')
    time.sleep(2)
    ftp.close()
    ssh.close()

signal.signal(signal.SIGINT, sigint_handler)

ssh = paramiko.SSHClient()
ssh.load_system_host_keys()
ssh.connect(hostname, username=username, password=password, compress=False)
ssh.exec_command("lipc-set-prop -- com.lab126.powerd preventScreenSaver 1")
while True:
    im = ImageGrab.grab(bbox=(1320, 280, 1920, 1080))
    im.save('/tmp/screen.png')
    ftp=ssh.open_sftp()
    ftp.put('/tmp/screen.png','/dev/screen.png')
    stdin, stdout, stderr = ssh.exec_command("fbink -g file=/dev/screen.png -x 0 -y 0")
    stdout.readlines()

mouse.sh

#!/bin/bash
sudo netcat -T -l -p 23 | while read -n 1 c; do
    case "$c" in
        ($'\033') 
            read -t.001 -n2 r
            case "$r" in
                ('[A') xdotool mousemove_relative -- 0 -10 ;;
                ('[B') xdotool mousemove_relative -- 0 10 ;;
                ('[D') xdotool mousemove_relative -- -10 0 ;;
                ('[C') xdotool mousemove_relative -- 10 0 ;;
            esac;;
        "")  xdotool mousedown 1 && xdotool mouseup 1  ;;
esac; done

[NiLuJe]

Released v1.15.0!

Remember when I said I wouldn't add image scaling? I lied. .

Besides a few bugfixes, this adds scaling & dithering support when printing images .
Dithering is cheap, scaling is not. So while this not an excuse to forget to properly preprocess, it does give you more options .

[PoP]

Small lies often lead to big lies.

[root@kindle us]# fbink -c -g file=NiLuJe.out.png,w=600,halign=center,valign=center
[FBInk] Detected a Kindle PaperWhite 3 (0G1 -> 0x201)
[FBInk] Clock tick frequency appears to be 100 Hz
[FBInk] Screen density set to 300 dpi
[FBInk] Variable fb info: 1072x1448, 8bpp @ rotation: 3 (Counter Clockwise, 270°)
[FBInk] Fontsize set to 24x24 (IBM base glyph size: 8x8)
[FBInk] Line length: 44 cols, Page size: 60 rows
[FBInk] Vertical fit isn't perfect, shifting rows down by 4 pixels
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 6782976 bytes & line length: 1088 bytes
[FBInk] Pen colors set to #000000 for the foreground and #FFFFFF for the background
Displaying image 'NiLuJe.out.png' @ column 0 + 0px, row 0 + 0px (scaling: 600x0, H align: 1, V align: 1, inverted: N, flattened: N, waveform: AUTO, HW dithered: N, SW dithered: N, nightmode: N, skip refresh: N)
[root@kindle us]# screenshot

[PoP]

Rebember piczat? Well, here is piczat2 using the image scaling new parameters. It is smaller and can resize text to any number of pixels.


announce.sh:

#expects PATH=$PATH:.:..
cd /mnt/us/piczat
fbink -c -B graya -q '' #change background color for transparency to show
piczat2 x= 250 y= 100 -S 6 -T "piczat2"
piczat2 x= 180 y= 300 -S 3 -T "\"Picture That Too\""
piczat2 y= 400 -S 3 -T "Using this Pants Patrol font for ASCII with transparent characters in any size:"
piczat2 x= 600 y= 700 -S 1 -T "Size 1"
piczat2 x= 500 y= 800 -S 2 -T "Size 2"
piczat2 x= 400 y= 900 -S 3 -T "Size 3"
piczat2 x= 300 y= 1000 w= 64 -T "16 x 64"
piczat2 x= 200 y= 1100 h= 64 -T "64 x 16"
piczat2 x= 100 y= 1200 w= 100 h= 50 -T "100 x 50"

screenshot

fatall.sh:

#expects PATH=$PATH:.:..
cd /mnt/us/piczat
fbink -c -B graya -q #change background color for transparency to show
piczat2 -T SAME -S 6     halign= MIDDLE   valign= TOP
piczat2 -T FAT w= 160  halign= MIDDLE   valign= MIDDLE
piczat2 -T TALL h= 160 halign= MIDDLE   valign= BOTTOM

screenshot
on K3  on PW3 

[PoP]

Beyond any reason... « 8x8 is already mind-boggingly tiny on a >250dpi screen » and « width must match the width of a C data type in bits », I was thinking that even titty's 5x3 font wasn't small enough for real geeks.

Seeing

prompted me to code that 3x3 font using the single atomic "█" character

ascii.sh:

fbink -c
3x3 ALT -x 0 -y 0  -S 1 -T "abcdefghijklmnopqrstuvwxyz0123456789"
3x3 ALT -x 0 -y 5  -S 2 -T "abcdefghijklmnopqrstuvwxyz0123456789"
3x3 ALT -x 0 -y 12  -S 3 -T "abcdefghijklmnopqrstuvwxyz0123456789"

And why stop there! I also updated piczat2 so you can read the entire United Nation's Human Rights Declaration on 1/3 of a PW3 screen.

HumanRights.sh

cd /mnt/us/piczat
file="./HumanRights.txt"
Y=0
while IFS= read -r line
do
piczat2 x= 0 y= 0  -Y "$Y" w= 3 h= 3 -F 3x3 -T "$line";Y=$((Y+3))
done <"$file"

Of course this font is best viewed at 3x3 (preferably with glasses). If you scale it, antialiasing ruins the effect.

piczat2 -M -m w= 100 h= 100 -F 3x3 -T "a b c" -c

@NiLuJe, would implementing pixelated scaling be a potential option?

[NiLuJe]

You'd just need to implement it as a proper bitmap font, as that requirement has since been relaxed to "width/height must *at most* match the width of a C data type" .

scientifica is 5x12, for instance .

[PoP]

Originally Posted by NiLuJe

You'd just need to implement it as a proper bitmap font, as that requirement has since been relaxed to "width/height must *at most* match the width of a C data type" .

scientifica is 5x12, for instance .

Ah, nice to know.

No need to bloat fbink with geeky stuff though. Thanks for its True Type support many alternatives exist, such as the 5 pix tall Zepto, the 4 pix tall Tiny, the 3 pix tall Tinier, and ultimately the 3x3 Font for Nerds.

And if our screens were not eink, possibly the 1x5 Millitext sub-pixel font.

[EDIT] Furthermore, the incredible 1x3 Bee-Knees sub-pixel font.

[NiLuJe]

Released v1.16.0!

Mostly a minor maintenance release, with a few tweaks & new features. It also handles the few new PW4 variants that popped up in FW 5.11.2...
Let's shine a light on the amazing lab126 codenames they dub their boards & platforms with .

[PoP]

Originally Posted by PoP

Maybe not so much

What could I fbink today? ... ...Maybe adding "Squashed" and "Spread" effects:

...
#SQUASHED text
fbink -c
T="SQUASHED"
i=1;x=0
while [ $i -le ${#T} ]; do
  t=$(expr substr "$T" $i 1)  #next byte of T
  fbink -t $tt,size=86,left=$x "$t" -M
  fbink -t $tt,size=86,top=3,left=$(($x + 3)) "$t" -o -M
  i=$(($i + 1));x=$(($x + 110))
done
screenshot #save for posting



#STRETCHED text
fbink -c
T="SPREAD"
i=1;x=0
while [ $i -le ${#T} ]; do
  t=$(expr substr "$T" $i 1)  #next byte of T
  fbink -t $tt,size=20,left=$x "$t" -M
  fbink -t $tt,size=20,top=3,left=$(($x + 3)) "$t" -o -M
  i=$(($i + 1));x=$(($x + 200))
done
screenshot #save for posting


...

[NiLuJe]

Released v1.17.0!

A fair amount of goodies in this one .

Error handling in the CLI tool is now extra verbose.

In the OT/TTF font rendering codepath, you can now get information about the line-breaking computation pass (c.f., the docs and the discussion in the Kobo FBInk thread).

Also on the OT/TTF side of things, the font size can now be specified as a decimal, which should allow finer-grained control over the final rendering .

On the fixed-cell font rendering side of things, Spleen has been updated to 1.0.5, and a couple of Amiga fonts have been imported (Topaz+ A1200 & MicroKnight+), as well as the VGA variant (8x16) of the IBM font.
The LF glyph is now always replaced by a space, to account for fonts where it is visible.

There's a bunch of minor performance improvements all over, but especially for the fixed-cell font rendering.

Of particular interest to users of legacy (< K4) Kindle devices (hi, @PoP! ), this one finally fixes a bunch of issues related to the quirkiness involved with dealing with a 4bpp framebuffer.
The most visible example I could find was overlay mode, in the fixed-cell font rendering, with an odd font scaling multiplier. It'd lead to bogus colors on the edges of glyphs.
(While it affected more stuff, the most common codepaths were using a couple of workarounds to paper over the underlying issue ;p. This release fixes it for good).

[NiLuJe]

Released v1.18.0!

A few fixes, most importantly the massive breakage on some specific recent Kindle variants, and added the possibility of specifying OT/TTF font size in pixels (which is how I'm drawing the icons in MRPI ).