<InPhase>
teepee: Crap... We can't define modules inside of if. :(
<InPhase>
I guess I will have to define N different modules to do this incrementally.
J24k71 has joined #openscad
J24k34 has quit [Ping timeout: 240 seconds]
teepee has quit [Remote host closed the connection]
<InPhase>
Tutorial 13 is done. Will push in like 30 minutes.
teepee has joined #openscad
guso78k has quit [Ping timeout: 240 seconds]
kintel has joined #openscad
<kintel>
pca006132 I was trying to understand how Manifold uses Clipper2 differently from OpenSCAD's usage of Clipper1, and this is my understanding:
<kintel>
Both Clipper1 and Clipper2 operate in quantized 64-bit integer space
<kintel>
Manifold quantizes global space in to units sized 10^-8
<kintel>
OpenSCAD quantizes the space occupied by the bounding box of the operands of any function so that 31-bit space is used for the bbox.
<kintel>
The latter is obviously far more precise, but we also run into the issue of losing precision when converting to double-prevision world space, causing occasional topology changes during conversion
<kintel>
* not actually 10^8 but its nearest power of twp
<kintel>
I'm considering aligning OpenSCAD with Manifold on this one, but it could also make sense to make this configurable in Manifold
snaked has quit [Read error: Connection reset by peer]
snaked has joined #openscad
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<InPhase>
J24k71: Is Tutorial 12 complete? I put Tutorial 13 in place on Day 18, but it can be moved to day 17 and we could swap tutorial numbers if yours is not ready.
<InPhase>
J24k71: Although it looked ready at a glance, it just wasn't in the index and readme.
<InPhase>
J24k71: Ours have no interdependencies that require any specific ordering.
<InPhase>
J24k71: I'm done pushing, and shutting down for sleep and then work, so if you need to swap and need more time, go ahead.
<pca006132>
kintel: where is that 10^8 coming from?
<pca006132>
oh, cross section
<pca006132>
I was thinking about 3D
<pca006132>
to be honest, we did not spent too much time working on clipper2 integration, it was worked on by one contributor a year or two ago, in a short amount of time without too much testing
<pca006132>
our tests were relatively simple, with small polygons where the coordinates are quite small, so we are not anywhere close to the precision limit
mtm has quit [Read error: Connection reset by peer]
guso78k has joined #openscad
mtm has joined #openscad
guso78k has quit [Quit: Client closed]
<J24k71>
inPhase Tutorial has 8 steps and is complete (not checked nor gif made)
<J24k71>
teepee now i understand replacing colors with size - well done! .. (strangely i didn't got a github mail for the update)
aiyion has quit [Ping timeout: 264 seconds]
califax has quit [Remote host closed the connection]
aiyion has joined #openscad
califax has joined #openscad
<teepee>
J24k71: yeah, I never got mail notifications for the private repo either
<teepee>
changing to translate([i,i]) makes it a bit more obvious what's going on for steps 1-3 I think
califax has quit [Remote host closed the connection]
califax has joined #openscad
califax has quit [Remote host closed the connection]
califax has joined #openscad
teepee_ has joined #openscad
teepee has quit [Ping timeout: 264 seconds]
teepee_ is now known as teepee
<J24k71>
f0lkert did this look right for you in povray export https://bpa.st/2F7Q
<J24k71>
teepee can you please check part 14 (day 19) and part 12 for tomorrow (day 17)
<teepee>
I had a look at part 12 already, looks nice. as mentioned above, I'll probably move the cubes around a bit so it's clearly visible there are multiple cubes in step 1 and make it a bit easier for the animation generator
<teepee>
like adding -1 to the list and dropping 10 so it does not need to zoom out as much
<J24k71>
ok feel free .. it is just a proposal
<teepee>
it's great, we almost made it, that leaves only 2 slots? one tutorial and the big 24 door
<J24k71>
i am more worried that part 14 is too complex
<teepee>
I'd say it's "advanced" week now, so it's fine even though I have not looked at it yet :)
<teepee>
if we integrate it into the app and/or maybe get a web version, we can always extend/modify/add more details
<J24k71>
it's about polyhedron and i think it would be easier to use 9 points for an n=8 gon - but i used 8
<J24k71>
because you don't need to warp (modulo) around for the last face
<J24k71>
i assume that is super confusing for new user
<teepee>
we can check with InPhase later, his teaching experience can probably give some advise
<InPhase>
Guest0101: When seeking to understand something like that, peel it off from the right, and run the pieces separately, adding things onto the left until the combination of things makes sense to you.
<J24k71>
nope the change is 4*(1/cos(atan(13/30)))
Guest43 has joined #openscad
Guest43 has quit [Client Quit]
<Guest0101>
InPhase: makes sense
<InPhase>
J24k71: Yeah. That expanded enough to get it right that nested cylinders is almost simpler here, except for the nuance of handling the epsilon exactly if you wanted exactly 4mm width.
<J24k71>
InPhase - if you have time, have a look at day 19 and if that can be improved pedagogically .. thought about calculation an ngon with n+1 points instead
<kintel>
pca006132 Ah, thanks for the clarification on the cross section history. Not sure what the best solution is in terms of precision, but I'll start by implementing the same precision as Manifold in OpenSCAD, then run all our tests and see what happens
<teepee>
J24k71: yep, I'll run formatter and gif script later
<kintel>
Of course, to make things fun, the core operators in OpenSCAD use one precision, svg import uses another, and roof uses yet another. Fun times..
<InPhase>
teepee: I made my own snowflake gif, because it has randomness. So I ran it a bunch of times until the "representative results" looked pretty in the way I wanted. ;)
<InPhase>
But I had found your script this time, and used it. Although switching the background scheme so that the white snowflake stood out better.
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mtm has quit [Remote host closed the connection]
mtm has joined #openscad
guso78k has joined #openscad
guso78k has quit [Quit: Client closed]
teepee_ has joined #openscad
teepee has quit [Ping timeout: 264 seconds]
teepee_ is now known as teepee
juri_ has quit [Ping timeout: 252 seconds]
juri_ has joined #openscad
<teepee>
J24k71: gif is generated now
<teepee>
note that the README is generated by scripts/generate-readme.sh from the index.js file, so it's always overwritten (except the header)
<J24k71>
+
<J24k71>
index.js was the one that needs to be changed .. i am sure i forget this till next year - Ü
<teepee>
just look in the scripts folder :-)
<teepee>
you don't need to run them, but they show what's automated
<teepee>
last one got me the "render aborted warning", so I added a note about that too
<teepee>
whew, and daily update posted even before the next door opens ;-)
<teepee>
that ALT text really takes a lot of time and I'm sure the result is not exactly good. but I'm trying
<teepee>
also I still have not found a way to actually get normal notifications. The only thing that works is the setting in the repo, but that's an admin config with a list of mail addresses
<UltimateCodeWarr>
Does OpenScad take advantage of the NVida GPU / CUDA?
<teepee>
no
<J24k71>
manifold uses multiple CPU cores though
<teepee>
and cuda was *removed* from manifold as it did not bring much improvement if at all
<UltimateCodeWarr>
I was looking a lot of OpenCV, DarkNet, FFMPeg Libraries, and you can download them and compile them with CUDA for a 2000x boost
<teepee>
also cuda is not happening anyway, freaking proprietary nonsense
<teepee>
image is very different from 3D CSG
Guest0101 has quit [Ping timeout: 260 seconds]
<UltimateCodeWarr>
Are there any "Rendering Benchmarks" with OpenSCAD that show various Computer Configurations (AMD vs Intel etc) to get the best benchmarks?
<teepee>
I have not seen any
<teepee>
in general a fast memory interface is a big win
<J24k71>
i think i have seen some complex test scripts