Current Server Time: 03:49 (Central Europe)

#apertus IRC Channel Logs

2020/05/08

Timezone: UTC


02:15
calle__
joined the channel
02:19
mumptai_
left the channel
04:48
metal_dent[m]
Hi, morning!
04:51
metal_dent[m]
i think the first step for T1172 (making dimensions dynamic) is solved when BAndiT1983 added the Icon struct
04:54
Bertl_oO
off to bed now ... have a good one everyone!
04:54
Bertl_oO
changed nick to: Bertl_zZ
05:41
BAndiT1983|away
changed nick to: BAndiT1983
05:43
BAndiT1983
hi metal_dent[m], image button has still fixed numbers, my adjustments was only for casting to Icon struct, so the numbers can be used easier and under common data structure
05:47
metal_dent[m]
the fixed numbers are the x and y positions, right? the size is now not fixed i guess
05:49
BAndiT1983
size of button is still fixed, icon is now according to the icon data
05:49
BAndiT1983
also the positioning of icon is fixed
05:52
metal_dent[m]
do you want a different size than the size in the icon data?
06:00
BAndiT1983
why?
06:06
metal_dent[m]
if not then the size issue is solved, the button is taking the size from the image headers
06:06
metal_dent[m]
now I need to solve the fixed-positon problem
06:09
BAndiT1983
position and dimensions of button and position of icon (should be centered for now)
06:11
BAndiT1983
you should also avoid force push in git, it can have bad consequences on the repo when done incorrectly
06:15
metal_dent[m]
I force push only after rebasing :/
06:16
BAndiT1983
nope, never force push without a good reason, are there merge conflicts after rebasing?
06:18
metal_dent[m]
no
06:19
BAndiT1983
then force push is not required, as it can introduce bigger problems if the code which is pushed has errors or even worse merge conflicts, as there are special markers and when it lands in the repo then the developers have a lot of "fun" to clean up the mess again
06:21
metal_dent[m]
ohh, I have cleaned a lot of mess initially when I started using git (bluez_ knows ':D )
06:22
BAndiT1983
git is rather intelligent and prevents quite some stuff, when used in proper way, but you can also create a bloody pile of bones, it's like with every tool
06:23
metal_dent[m]
yes, will avoid using force-push from now :)
06:25
BAndiT1983
ok, just google it, wasn't expecting that -> https://www.youtube.com/watch?v=y0tbVCvX4gE
06:28
metal_dent[m]
Wait, have I already created the pile?
06:28
BAndiT1983
https://stackoverflow.com/questions/19933127/imagemagick-force-output-fileformat
06:28
BAndiT1983
not yet ;)
06:29
BAndiT1983
i suppose it should be -> convert image.svg ... xbm:image.data
06:40
metal_dent[m]
convert .svg file did not work so I converted with .png
06:49
metal_dent[m]
> https://stackoverflow.com/questions/19933127/imagemagick-force-output-fileformat
06:49
metal_dent[m]
okay, thanks! let me try this
06:50
BAndiT1983
have you done this for png? -> https://stackoverflow.com/questions/9853325/how-to-convert-a-svg-to-a-png-with-imagemagick
06:51
BAndiT1983
i mean the inkscape export through command line?
06:51
metal_dent[m]
no, I used im
06:52
BAndiT1983
and which problem came up?
06:53
metal_dent[m]
i noticed after removing "-monochrome" the image looked better
06:53
metal_dent[m]
but it's not the ideal solution
06:54
BAndiT1983
i mean SVG conversion
06:58
se6ast1an
good day
06:58
BAndiT1983
hi
06:58
BAndiT1983
have just checked and i can convert without big problems from SVG to xbm
06:59
BAndiT1983
e.g. magick -background black -monochrome Crusader-1973.svg xbm:crusader.h
06:59
BAndiT1983
GIMP can even open the file, but it has to end with XBM
07:00
BAndiT1983
this would reduce the extra steps in your conversion script
07:15
BAndiT1983
can someone try http://www.imagemagick.org/discourse-server/viewtopic.php?t=30986, looks like their server has problems at the momen
07:15
BAndiT1983
topic is Jagged edges after converting to XPM - ImageMagick
07:16
BAndiT1983
ah, now it works, probably a hiccup on their side as other sites were ok for me
07:17
se6ast1an
yes works here as well
07:19
BAndiT1983
this one does not have jagged edges, but the roof is thinner -> magick home_icon.svg -threshold 50% -scale 24x24 home_icon.xbm
07:23
illwieckz
left the channel
07:56
illwieckz
joined the channel
08:02
panintended
joined the channel
08:03
BAndiT1983
metal_dent[m]: here is the current command which is okay for further tests -> magick -density 800 home_icon.svg -filter point -threshold 50% -resize 24x24 home_icon.xbm
08:03
BAndiT1983
point filter is necessary to reduce the jagged edges, otherwise imagemagick tries to do some filtering, probably linear or bicubic
08:07
BAndiT1983
ah and you should add +negate (!) not -negate
08:07
BAndiT1983
otherwise i've got black image
08:09
BAndiT1983
http://www.imagemagick.org/Usage/filter/
08:14
metal_dent[m]
i'm getting same image in both + and -negate
08:16
BAndiT1983
had a black image, as the conversion kicks in and then it can't differ between the images
08:16
BAndiT1983
i mean pixels
08:16
BAndiT1983
ah, the order is necessary, it should be before threshold it seems
08:17
metal_dent[m]
yes, I unknowingly did that ':D
08:17
BAndiT1983
good, but un- should vanish later ;)
08:17
metal_dent[m]
yup yup!
08:19
BAndiT1983
the tricky part with such small images is the scaling and the pixel position, you could read about scaling pixel art in imagemagick, as it's targeting almost same approach
08:19
BAndiT1983
or maybe the order has to be adjusted, like scaling then negate and threshold and so on
08:21
metal_dent[m]
order is very imp, I read that when was working earlier
08:21
BAndiT1983
just by looking at the small icons i would almost say that they have to be adjusted by hand in gimp
08:21
BAndiT1983
this is what i've forgotten yesterday, that it can open XBM files, when se6ast1an has asked about it
08:22
metal_dent[m]
but still they look much better than the earlier conversion
08:22
BAndiT1983
or you can go fancy here -> https://xbm.jazzychad.net/
08:23
BAndiT1983
yes, but at some point we have to reach a polished version, without edges, problem is the scaling and the angles of the lines (think of Bresenham)
08:23
BAndiT1983
https://en.wikipedia.org/wiki/Bresenham%27s_line_algorithm
08:24
BAndiT1983
SVGs are nice for most stuff, but the goal are 24x24 pixel perfect icons, then the things start to look much different
08:25
BAndiT1983
there are nice filters for console emulations like SNES or so, 2xSuperSai and whatnot, but i don't know if imagemagick has similar ones to straighten out the edges
08:26
BAndiT1983
few clicks in the online editor and i have perfect home icon
08:55
futarisIRCcloud
left the channel
09:10
illwieckz
left the channel
09:10
illwieckz
joined the channel
09:49
se6ast1an
<BAndiT1983> or you can go fancy here -> https://xbm.jazzychad.net/ <- great tool, bookmarked
09:52
BAndiT1983
i miss some basic tools there, so it's okay tool
09:53
BAndiT1983
krita allows to export XBM and has grid for pixel perfect drawing
09:54
BAndiT1983
haven't found the pixel grid in gimp yet
09:54
BAndiT1983
it should work -> https://www.youtube.com/watch?v=oAaoh49yf5A
09:58
BAndiT1983
grid can be configured under image -> configure grid -> pixels -> 1 / 1
09:59
metal_dent[m]
> metal_dent: here is the current command which is okay for further tests -> magick -density 800 home_icon.svg -filter point -threshold 50% -resize 24x24 home_icon.xbm
09:59
metal_dent[m]
with this command are you getting white image with black background?
10:01
BAndiT1983
yes, if you add negate
10:01
BAndiT1983
after SVG file ;)
10:02
BAndiT1983
nice gimp works fine with pixel-perfect drawing, the tool i miss in the online version is the line drawing
10:02
metal_dent[m]
and without negate a blank image?
10:02
metal_dent[m]
*black
10:03
BAndiT1983
nope, black house, white background
10:04
BAndiT1983
this is the final command for me: magick -density 800 home_icon.svg -filter point -negate -threshold 50% -resize 24x24 home_icon.xbm
10:05
BAndiT1983
but afterwards the pixels have to be adjusted a bit, as SVG doesn't have proper angles or the conversion is mangling them, but 45 degrees is not jagged, also 22,5 or so, at least gimp allows to draw straight lines with fixed degrees and they look just fine in pixel art
10:06
BAndiT1983
ah, it's multiplies of 15, so 15, 30, 45 etc.
10:07
BAndiT1983
another thing to check is the SVG itself and it's data in inkscape, maybe we have to scale it to pixels in the original file, then the conversion will be clearer, but angles have still be followed to get crisp image
10:11
metal_dent[m]
the command is working differently with different images
10:18
BAndiT1983
what do you mean?
10:20
metal_dent[m]
this is working like you said with the home_icon (white image, black bg) but not with up and down icons (black image, white bg) and it's giving a black image for the apertus_text and ring
10:24
BAndiT1983
interesting, as down arrow SVG is like the home SVG, transparent background and black icon
10:24
BAndiT1983
maybe you have to add like -background white before SVG
10:25
BAndiT1983
nope, still no problem with my line
10:25
BAndiT1983
magick -density 800 down_icon.svg -filter point -negate -threshold 50% -resize 24x24 down_icon.xbm
10:25
BAndiT1983
black background and white arrow down
10:30
metal_dent[m]
but I'm getting this :/
10:30
metal_dent[m]
-> https://pasteboard.co/J7pebBV.png
10:31
BAndiT1983
https://pasteboard.co/J7pfoXS.png
10:34
BAndiT1983
which imagemagick version do you use?
10:34
BAndiT1983
andi@apertus-dev î° ~ î° magick --version
10:34
BAndiT1983
Version: ImageMagick 7.0.10-10 Q16 x86_64 2020-04-28 https://imagemagick.org
10:35
metal_dent[m]
mine is 7
10:35
metal_dent[m]
same!
10:36
BAndiT1983
would have wondered otherwise, as you also have manjarpo
10:36
BAndiT1983
*manjaro
10:40
metal_dent[m]
found the problem!
10:41
metal_dent[m]
my input svg(s) somehow changed while I was experimenting multiple commands for the conversion
10:50
panintended
left the channel
11:05
metal_dent[m]
can you check whether you're getting proper logo images with these commands:
11:05
metal_dent[m]
magick ApertusLogo.svg -crop 645x186+1+15 -density 800 -filter point +negate -threshold 50% -resize 214x64 TextLogo.xbm
11:06
metal_dent[m]
magick ApertusLogo.svg -crop 41x41+649+0 -density 800 -filter point +negate -threshold 50% -resize 14x14 RingLogo.xbm
11:17
BAndiT1983
just a moment
11:18
BAndiT1983
yes
11:18
BAndiT1983
some pixels are jagged, but ok all in all
11:27
BAndiT1983
this crop argument is a bit wonky, as SVG can change in size
11:27
BAndiT1983
inkscape can convert objects by id, bit IM not
11:38
metal_dent[m]
so should I use inkscape?
11:39
BAndiT1983
it was just a note about possible problems in the future, can only guide there
11:44
BAndiT1983
checked apertus test and there is black line at the top above "t", this is one of the things where the process is showing it's shortcomings, but was expected
11:47
BAndiT1983
as we know the positions of elements in the firmware, maybe we can split the logo in 2 SVG files and work with them
11:50
metal_dent[m]
> checked apertus test and there is black line at the top above "t", this is one of the things where the process is showing it's shortcomings, but was expected
11:50
metal_dent[m]
how about when you write this:
11:50
metal_dent[m]
-crop 645x186+1+20
11:51
BAndiT1983
i don't like the fact that we try to find absolute positions in some SVG file, which is a vector format, not absolute pixel
11:57
BAndiT1983
se6ast1an: are you still with us?
11:58
Bertl_zZ
changed nick to: Bertl
11:58
Bertl
morning folks!
11:58
BAndiT1983
hi
11:58
Bertl
note: you can run inkscape on the commandline as well to separate 'objects' by id
11:59
BAndiT1983
yes, tried that today, but the question is if we should also add it, but in my opinion inkscape is rather common
12:00
Bertl
probably a sed command would suffice for 'clean' SVGs
12:00
BAndiT1983
the process would be: export 2 objects in separate file at higher resolution, convert each file to an icon
12:00
BAndiT1983
Bertl: please elaborate
12:00
Bertl
let me take a look at the SVGs in question first
12:01
Bertl
https://apertus.org/sites/default/files/apertus%C2%B0_Logo.svg this one?
12:01
BAndiT1983
let me compare
12:02
BAndiT1983
hm, this one is 5,1KB and in the repo 4,5KB, can't say if they are equal, have to check their content quickly
12:02
Bertl
at least that is the official one :)
12:03
BAndiT1983
viewbox is equal
12:03
BAndiT1983
so i suppose yes and maybe just the one in repo is cleaner
12:04
BAndiT1983
nice tool for cleanup of SVG -> https://jakearchibald.github.io/svgomg/
12:04
Bertl
of course, online, as usual :)
12:05
BAndiT1983
when i want to see quick results i use online tools, can still move the process to the command line for the build pipeline
12:05
BAndiT1983
it's much quicker to uses toggle buttons and see the compression results
12:06
Bertl
so you want to split out path36 from the rest, right?
12:07
Bertl
or more precisely, you want to split by color
12:07
BAndiT1983
i thought by ID, but haven't looked at the objects yet
12:08
Bertl
so, my suggestion would be the following:
12:08
Bertl
there are two 'known' colors in the SVG, the text and the ring
12:08
BAndiT1983
i think they should be regrouped, as dot is in the same group as letters
12:08
Bertl
basically you can simply change one of the 'colors' to transparent
12:08
Bertl
(or background)
12:09
BAndiT1983
https://www.imagemagick.org/discourse-server/viewtopic.php?t=15675
12:09
BAndiT1983
"crop to content" like in gimp
12:09
Bertl
and you will get two identical images (size and rendering) except that one is 'missing' all objects from one color and the other from the other color
12:10
Bertl
this not only allows you to have consistent rendering, it also allows you to locate the ring in relation to the rest
12:12
BAndiT1983
sounds good
12:14
BAndiT1983
http://www.imagemagick.org/discourse-server/viewtopic.php?t=29720
12:14
BAndiT1983
how to change the color ^
12:28
Bertl
sed 's/#c2bfbc/#ffffff/g' ?
12:30
Bertl
but it might be even simpler to change the fill-opacity from 1 to 0
12:31
Bertl
something like sed '/#c2bfbc/ s/fill-opacity:1/fill-opacity:0/' might do the trick here
12:34
BAndiT1983
nice, this would avoid quite some stuff and also can be done automatically in the build pipeline
12:35
BAndiT1983
thinking about generating hash, so the build knows when image file has changed
12:35
BAndiT1983
but it's a "nice-to-have" for later
14:05
BAndiT1983
metal_dent[m]: what's your state at the moment?
14:12
BAndiT1983
changed nick to: BAndiT1983|away
14:12
BAndiT1983|away
changed nick to: BAndiT1983
14:22
metal_dent[m]
sorry took a break... right now I was reading the logs, so are we going with what Bertl suggested?
14:22
metal_dent[m]
it sounds good but a little long process..
14:27
Bertl
hmm? please elaborate?
14:27
BAndiT1983
with sed it would be a quick process, you can also try the command line of inkscape if you want to test things
14:28
BAndiT1983
quick reboot
14:28
BAndiT1983
changed nick to: BAndiT1983|away
14:28
BAndiT1983|away
changed nick to: BAndiT1983
14:29
metal_dent[m]
okay, will first try on one image then will change the script
14:30
BAndiT1983
regarding conversion in general, maybe you can add a script which loops through all SVG files, at least the ones ending with _icon, then you wouldn't need to have that many lines and it would process all available files
14:32
Bertl
Makefile
14:33
Bertl
this way it only converts stuff when the image changes
14:33
Bertl
(or the output product is not present)
14:34
BAndiT1983
the task is not for a faint heart, a lot of trial and error is involved, but maybe it was only me with the firmware and bootlaoder ;)
14:34
calle__
left the channel
14:35
Bertl
'the task' being?
14:36
metal_dent[m]
the image conversion, right?
14:36
BAndiT1983
makefile adjustments
14:37
BAndiT1983
you can try to extend available one, but i would suggest to create an empty makefile in the icon folder and try with it first
14:37
BAndiT1983
later we can merge it in the firmware makefile
14:38
metal_dent[m]
yes, obviously
14:42
Bertl
it's perfectly fine to have a bunch of makefiles distributed over several sub directories
14:42
Bertl
no need to 'merge' them, just call them from the main makefile
15:28
megora
joined the channel
15:34
anuejn
or even include them :)
15:34
anuejn
... depends
15:53
BAndiT1983
changed nick to: BAndiT1983|away
15:57
Dest123
joined the channel
16:16
BAndiT1983|away
changed nick to: BAndiT1983
17:09
BAndiT1983
changed nick to: BAndiT1983|away
17:29
se6ast1an
<BAndiT1983> se6ast1an: are you still with us? <- I am now
17:29
se6ast1an
how can I help?
18:30
megora
left the channel
18:32
megora
joined the channel
18:35
BAndiT1983|away
changed nick to: BAndiT1983
18:38
BAndiT1983
se6ast1an: was just wondering that you were suddenly not active anymore
18:46
se6ast1an
yes, the unexpeted happened, a social life :P
18:47
BAndiT1983
so unusual for you? ;)
18:47
BAndiT1983
quick reboot, sometimes it's really difficult to get some devices to run under linux, but this usb device runs with my android without problems
18:47
BAndiT1983
changed nick to: BAndiT1983|away
18:48
BAndiT1983|away
changed nick to: BAndiT1983
19:07
megora_
joined the channel
19:07
megora
left the channel
19:15
se6ast1an
just pushed a larger commit to add full screen parameter list menu
19:15
se6ast1an
https://github.com/apertus-open-source-cinema/AXIOM-Remote/commit/9690200aba4d1455319c45ceb69f3e9580d9f9be
19:15
se6ast1an
note that the button assignment in visualizer for testing changed
19:15
se6ast1an
Up = Btn 10
19:15
se6ast1an
Down = Btn 13
19:15
se6ast1an
sorry
19:15
se6ast1an
Down = Btn 12
19:15
se6ast1an
select = Btn 11
19:16
se6ast1an
which basically mimics turning the knob or pushing the knob
19:17
megora_
left the channel
19:20
Bertl
because the knob is not implemented yet?
19:25
BAndiT1983
the knob is implemented, but no button there yet in visualiser, panintended cliamed the task
19:25
BAndiT1983
*claimed
19:26
BAndiT1983
serial connection is also working on my side, have to check with real board data exchange
19:27
se6ast1an
great
20:47
se6ast1an
off to bed for today, good night
20:48
BAndiT1983
good night
21:19
BAndiT1983
changed nick to: BAndiT1983|away
22:03
futarisIRCcloud
joined the channel
22:06
aombk
left the channel
22:08
aombk
joined the channel
23:22
illwieckz
left the channel
23:33
illwieckz
joined the channel