<J24k69>
alternativ we could use look_at *-1 and location the same + <0,0,50> then rotation needs +<0,0,180> to match and would allow to change the distance easy ( 50 )
mmu_man has quit [Ping timeout: 255 seconds]
mmu_man has joined #openscad
lockywolf has joined #openscad
lockywolf has quit [Client Quit]
lockywolf has joined #openscad
<f0lkert>
J24k69: i'm working on using the actual viewport
<f0lkert>
J24k69: I just got sidetracked with other projects
<J24k69>
f0lkert no problem - i just wonder which approach is more convenient to use the look at or rotation - using rotation with a top down view makes it easier to change distcance
<f0lkert>
Yes but I need to re-calculate that into a look-at as that's the povray-way afaik :-)
<J24k69>
f0lkert from my tests - if you use locaton <0,0,50> look_at <0,0,0> then rotation<-55,0,25+180> works
<J24k69>
also using povray as print service is strange - pov ray show the rendered image - but not the file and if you hit run again it will render the previous file - so you can not change resolution.
<J24k69>
also the image is saved in temp
<J24k69>
maybe there are some command line options to make this nicer
<J24k69>
so we could leave the export geometry only and include a selectable light/camera file
<f0lkert>
yup
<f0lkert>
but
<f0lkert>
hmmm, and then also a template for 'selected-viewport.inc' or so, so that via a #define the calculated/selected viewport is used.
<J24k69>
I noticed the position of the light needs to be relative to the object size - so maybe best placed two Arial lights beside the camera location ( like it is done with real cameras)
mmu_man has quit [Ping timeout: 246 seconds]
<J24k69>
If we have a nice working template - we don't need any selections .. the pro user will know how to change the script - and dummy pov user like me are just happy what they get
mmu_man has joined #openscad
<J24k69>
f0lkert btw when using more complex geometry i get a lot "polygon not closed - closing" warnings in povray
<f0lkert>
that's a thing I don't know how to fix unfortunately
<J24k69>
ok
mmu_man has quit [Ping timeout: 265 seconds]
deathonater has joined #openscad
Smeef has quit [Ping timeout: 252 seconds]
mmu_man has joined #openscad
Guest28 has joined #openscad
Guest28 has quit [Client Quit]
<buZz>
J24k69: oooo that second one looks so much better, nice
<buZz>
J24k69: i'd say one behind the camera, one behind the camera and 'high up a lot' maybe 2x Z of object?
<buZz>
so you get a bit better ambience lighting
<J24k69>
yes but needs to be before the object as the areal light casting shadows but lighting an object like a point .. bit strange. Also you probably want them not directly in line with the camera - bit left so shadows show
<J24k69>
also making something that works for a cube(10) and for a cube (100) is quite difficult
aiyion1 has joined #openscad
<buZz>
hmhm yeah i can imagine
<buZz>
maybe f0lkert has a brilliant idea for it ;)
<J24k69>
i think if the camera position is calculated we get a good result by positioning the light relative to the camera and scale the intensity and size by distance
<buZz>
i hope someday we can use the animation features of openscad with the povray export :P
<J24k69>
probably not so difficult as the $t and clock for pov is not so different - easiest would be to use openscad make the animation - else you need to export the $t which is not part of CSG-tree afaik. But using scad for the animation will leave no control over lights material etc.
<buZz>
unless its a 'theme' for the export
<J24k69>
yes
<J24k69>
but for windows seems difficult to get animation render in povray
<buZz>
windows?
<buZz>
the OS?
<J24k69>
yes
<buZz>
i wouldnt have expected windows povray to behave differently from all others :)
<J24k69>
can't find any documentation but seems the animation need scripting that is OS dependent as povray it self is a comand line tool
<J24k69>
there is no official tutorial to learn povray .. smh
snaked has quit [Quit: Leaving]
<J24k69>
oh y is z in povray that explains somethings
J24k17 has joined #openscad
J24k69 has quit [Ping timeout: 256 seconds]
nmr_scientist has joined #openscad
<nmr_scientist>
hello. Anybody has experience with using BOLS2. It's a great library but scarcely any tutorials
JakeSays has joined #openscad
ubuntourist has joined #openscad
<pca006132>
anyone here is interested in joining a discord server about openscad, manifold, jscad and potentially other code cad stuff?
<teepee>
no
<J24k17>
no
<teepee>
nmr_scientist: for very specific BOSL2 questions, the mailing list is a good place, as at least BOSL2 devs are reading there. good chance for detailed answers
<pca006132>
I think my major complaint with irc is that it doesn't work across multiple machines (kind of works with multiple accounts), and recently the latency is getting higher...
<pca006132>
it may take up to several seconds for the msg to display on screen, like slower than rendering :P
<teepee>
I guess we could try installing a client on our fileserver
<teepee>
no such issues here with the delay, not sure why that happens, maybe one of the libera servers is acting srtange
<pca006132>
and the weirdest thing is my lab network just outright blocked irc
<pca006132>
no idea why, maybe manager hates irc
<teepee>
hmm, you are connected via tantalum.libera.chat which is supposedly in Amsterdam
<pca006132>
I guess all their servers are in Europe?
<pca006132>
maybe US as well?
<teepee>
I don't know if they have a list somewhere
<pca006132>
now the network is actually fine, I just get that high delay quite often recently
<teepee>
they block networks with open relays as those are potential spam entry points
<pca006132>
I see
<teepee>
what was that client name? thelounge or something?
<teepee>
it claims "always connected" mode
thelounge7162 has joined #openscad
<pca006132>
makes sense, just feels like a bot for your account, and you connect to the bot
<thelounge7162>
hello from the "thelounge" demo instance....
<teepee>
uh, that demo instance is running in germany?
<Scopeuk>
When I was first playing the this studd I used to export stl, pull into a blender model and light it, then render with indigo (before it went proprietary)
<Scopeuk>
Pov ray is much quicker and lighter
<Scopeuk>
I do still like the unbiased physical renderers, there is magic in watching the image form before you as it adds extra projections and the details fills in
<nmr_scientist>
pov integration?
<J24k17>
you can export povray and config it as print service - so with a button click (F7) it will render the image
<peepsalot>
teepee: found a little bug via clang-tidy, which was complaining because a number instead of bool literal is used here, but I think beyond that the logic is inverted too (it should return false ?)
<nmr_scientist>
there's a workshop I am planning to do.
<teepee>
cool, good luck with the workshop, so you teach other people or got some homework?
<nmr_scientist>
I am supposed to be teaching. it wouldn't do me good if I barely know how to use fusion 360. I supposed I could demonstrate OpenSCAD instead.
<teepee>
unfortunately we don't have easy to use material for that ready. I'd always love to have that on the website
<teepee>
there's the tutorial of course, which is *very* basic stuff but that's probably not simple to convert into a workshop/presentation
<teepee>
a couple of days ago we talked about having some web based introduction, but that's going to be quite some work
<teepee>
J24k17: nice render!
<nmr_scientist>
I am going to be going to makerfaire soon. So I may have to postpone the workshop....Hmm.
<peepsalot>
GLAD support is supposed to be the most cutting edge option, right? is that still WIP?
<teepee>
yes, GLAD is the way to go, it does allow to have both EGL and X11 in one application
<peepsalot>
i got a bunch of redefinition errors after attempting USE_GLAD
<teepee>
I'm not sure if the windows build is already active with GLAD
<teepee>
oh?
<teepee>
I thought that's default on the CI
<teepee>
maybe it needs to be enabled only via USE_INCLUDED_OPENCSG (or whatever the exact name is)
mmu_man has joined #openscad
<teepee>
hmm, no, CI has none of the flags. my local setup uses -DUSE_BUILTIN_OPENCSG=ON though
J24k17 has quit [Quit: Client closed]
<peepsalot>
with -DUSE_GLAD=ON also?
<teepee>
no, not in my script calling cmake, and I can't find in the cmakelists that it forces that
<teepee>
so I'm a bit confused what the current combination is
<teepee>
[cmake] -- OpenCSG submodule: 1.6.0
<teepee>
[cmake] -- USE_GLAD/USE_GLEW not specified: Defaulting to GLAD
<teepee>
ahh, there it is
<teepee>
if (${OPENCSG_VERSION_STRING} VERSION_LESS "1.6.0")
<teepee>
set(OPENCSG_GLEW ON)
<teepee>
so glad is default, *but* only if opencsg is >= 1.6.0
<peepsalot>
ah ok, i had to wipe my build dir after adding USE_BUILTIN_OPENCSG and its working now
<peepsalot>
still wish i knew how to avoid these issues which seem internal to boost: /usr/include/boost/function_output_iterator.hpp:14:1: note: ‘#pragma message: This header is deprecated. Use <boost/iterator/function_output_iterator.hpp> instead.’
<InPhase>
nmr_scientist: I was running a physics course for non-scientists (mostly majors like business and accounting), and I managed to wedge OpenSCAD into the lab portion. Lab 3 at this link was my crash course in OpenSCAD, designed to give the most time-efficient background and basic competence in OpenSCAD that I could to people who have never heard of a syntax error. Then 4 was design the rockets, 5 was an
<InPhase>
nmr_scientist: You can see the structure and content of that lab03 was extremely arbitrary, but it was a sufficient background for all those non-programmers to get going quickly on the rocket design.
<InPhase>
nmr_scientist: When I was advising on the content for the official tutorial, it was the experience of having taught this crash course on it that motivated much of my advice about how the tutorial should work. The tutorial is of course "better" because it's much more thorough and comprehensive, but I'm linking to this anyway to show how these notions can be made very compact with simple reference
<InPhase>
examples following highly specific instructions that draw attention to the things they need to understand.
<InPhase>
nmr_scientist: Of course, you can go have your workshop participants play with Fusion 360 too. :-P But with the right introduction to it, people can handle OpenSCAD regardless of their background.
UltimateCodeWarr has quit [Quit: Client closed]
mmu_man has quit [Ping timeout: 245 seconds]
<peepsalot>
teepee: i'm looking at some more clang-tidy checks that are triggered only once or a few times, and there is this one in some FontList code which I'm confused about https://bpa.st/PTZA
<peepsalot>
the message seems like its saying that idx.model() is definitely null? or is it just there's a possibility for it to return null?
<teepee>
that file only has 499 lines here
<teepee>
oh, probably the header changes
<peepsalot>
a bunch of header stuff was edited recently
<peepsalot>
yeah
<teepee>
hmm, idx can't be null, it's a reference
<teepee>
so model() returning has to be the issue
<teepee>
default constructor does set it to nullptr
<peepsalot>
i could see adding a conditional in case idx.model() is nullptr, but not sure what is appropriate to return; empty string for colStr? no idea for colIdx
mmu_man has joined #openscad
<teepee>
if idx is only used from the str call, the str check might already take care of it too