--- Log opened Wed Jun 16 00:00:39 2021
01:51 < L29Ah[XMPP]> > it doesn't automatically set the output resolution
01:51 < L29Ah[XMPP]> it does, you only have not to forget to set it
01:51 < L29Ah[XMPP]> according to your process' precision
01:53 < L29Ah[XMPP]> you throw it the max deviation of a circle() from an ideal circle, and it does everything by itself
01:53 < srk> hehe, it's a tradeoff.. rendering time, output file size
01:53 < srk> ideally there would be no discretization of the code until it is *really* required
01:54 < L29Ah[XMPP]> ideally there's no discretization anywhere except preview and g-code interpreter
01:54 < L29Ah[XMPP]> but we're not even nearly there
01:54 < srk> yes
01:54 < srk> implicitcad & hslice has a lot of potential regarding this :)
01:55 < L29Ah[XMPP]> still a printer would happily chew 100MB of gcode so the discretization is impossible to see in the end result
01:55 < srk> (also g-code is kinda superfluous as well)
01:55 < srk> "just" feed the motion controllers with functions as well and let them interpolate :)
02:04 < fenn> what part of "automatic" didn't you understand
02:04 < fenn> "you only have not to forget to set it" is not automatic
02:04 < fenn> and obviously it doesn't work because nobody ever does it
02:04 < fenn> marked "wontfix" by the dev
02:06 < fenn> and the defaults are really cruddy
02:08 < srk> and automatic is pretty hard
02:12 < fenn> no it really isn't
02:12 < fenn> if it were actually hard, there wouldn't be any reason for outrage
02:13 < fenn> you apply all the scaling factors, calculate the deviation from roundness, and if it's greater than your allowed error value you add more segments
02:14 < fenn> i'm sure there are more efficient algorithms
02:14 < fenn> maybe a lookup table
02:16 < srk> for a single cylinder it sounds easy, sure
02:17 < fenn> also fwiw g-code tangent biarcs can be very close to splines, while also producing a smooth movement
02:18 < fenn> approximating a spline with a small number of arcs, with smooth transitions between them
02:19 < fenn> it really cuts down on the data rate, or conversely enables a much higher precision in g-code
03:07 < maaku_> kanzure: that NURBS vs subdivision surface thing seems like BS.
03:08 < maaku_> NURBS are mathematically precise curves. subdivision surfaces is a fancy word for point interpolation
03:09 < maaku_> by information theory so long as your original mesh surface is sampled on at least twice the resolution of the feature size you care about, it will print correctly
03:10 < maaku_> i think that criticism boils down to "inaccurate sampling of a curve is insufficient to reconstruct the curve" ... which is a boring, obviously true statement
03:10 -!- maaku_ is now known as maaku
03:38 < fenn> subdivision surface probably is referring to https://en.wikipedia.org/wiki/Catmull-Clark_subdivision_surface
03:41 < fenn> typically people complaining about subdivision surfaces don't know anything about how they are used in practice
03:46 < fenn> solar panel installation EVA on ISS https://www.youtube.com/watch?v=gCKsedpraVg
05:30 -!- thedragon [~thedragon@user/thedragon] has quit [Quit: Leaving]
06:01 -!- thedragon [~thedragon@user/thedragon] has joined #hplusroadmap
06:09 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0:ecbc:81b9:438c:404b] has joined #hplusroadmap
06:27 < fenn> .title https://www.youtube.com/watch?v=nTEOTsMtw4M
06:27 < saxo> Home-made 'Nose Pillows' for Ventilator/CPAP use. - YouTube
06:28 < fenn> some random blender CAD thing that passed across my screen and would be rather hard to do programmatically (the purple air tube part, not the mold part)
07:15 < L29Ah[XMPP]> 12:04:02]<fenn> what part of "automatic" didn't you understand
07:15 < L29Ah[XMPP]> the part where you think that openscad somehow can know what kind of machine you're making your model for without your assistance
07:17 < fenn> all it should need is a precision value
07:21 < L29Ah[XMPP]> 11:53:07]<L29Ah[XMPP]> you throw it the max deviation of a circle() from an ideal circle, and it does everything by itself
07:21 < L29Ah[XMPP]> what part of it didn't you understand?
07:21 < fenn> does it actually do that? or do you actually have to mess around with $fn or write your own macros
07:22 < fenn> i haven't used it in a few years
07:23 < fenn> also $fn needs to be specified in multiple different locations, basically every time you use a cylinder
07:23 < fenn> and god forbid you do some kind of loop to approximate a curved surface
07:37 < L29Ah[XMPP]> fenn: $fs
07:38 < L29Ah[XMPP]> $fn is a low level stuff you rarely need
07:38 < fenn> $fs is the length of the segment which is still not really what i want
07:38 < fenn> but better than nothing i guess
07:39 < fenn> also this is all really badly documented
07:40 < fenn> also with no units specified, they have some highly questionable minimum values for these numbers
07:40 < L29Ah[XMPP]> thanks gentoo that patches $fs minimum away for me
07:40 < L29Ah[XMPP]> ∞ wgetpaste  /etc/portage/patches/media-gfx/openscad/openscad-lower-fminimum.patch   
07:40 < L29Ah[XMPP]> Your paste can be seen here: http://dpaste.com/7VSWGC4QA
07:41 < L29Ah[XMPP]> but indeed it isn't friendly for some strange reason
07:50 < L29Ah[XMPP]> ah and you also need to set $fa to 1 or smth
07:51  * L29Ah[XMPP] feels he should add ~/.openscadrc support or smth
08:05 < kanzure> portage? we have a living gentoo user?
09:04 < superkuh> I miss sci-hub so much.
09:04 < superkuh> It's like losing a body part. Suddenly everything I want to read is inaccessible.
09:14 < jrayhawk> only the scholarly priesthood is qualified to read and interpret the holy texts
09:54 -!- SDr [~SDr@li1189-192.members.linode.com] has quit [Ping timeout: 240 seconds]
10:03 -!- SDr [~SDr@li1189-192.members.linode.com] has joined #hplusroadmap
13:25 -!- catalase_ [~catalase@user/catalase] has quit [Remote host closed the connection]
16:41 < fltrz_> the whole idea of subdivision is that you can split a curve / surface in a way that the remaining parts have the exact same shape as the original part of the shape before splitting. brute force reducing to linear or planar segments means that further subdivision doesnt refine to the original circle in case of a circle. But with a ratio of polynomials of power 2 you can have perfectly circular arcs that are subdividable..
16:43 < fltrz_> superkuh, I ve been to busy to use sci-hub lately, but https://sci-hub2.org/ seems to work here? didn't actually try inserting a DOI though
16:44 < superkuh> It's just lacking new articles. Anything from the last month or so. Or, that's how long I've noticed new articles never resolving.
16:44 < fltrz_> ah
16:44 < fltrz_> https://www.bbc.com/news/education-56462390
16:45 < fltrz_> "Police warn students to avoid science website"
16:49 < L29Ah[XMPP]> Sean is an award-winning education correspondent for the BBC News website. As well as covering news about schools and universities in the UK, he is editor of the BBC's international education online series, The Knowledge Economy, which looks at the impact of education from a global perspective and how it is shaping the economies of the future. He is author of several books, including a cultural h
16:49 < L29Ah[XMPP]> istory of sleep.
18:19 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0:ecbc:81b9:438c:404b] has quit [Quit: Leaving]
19:13 -!- ^ditto` is now known as ^ditto
20:28 -!- mrdata [~mrdata@user/mrdata] has quit [Quit: Leaving]
20:41 -!- mrdata [~mrdata@user/mrdata] has joined #hplusroadmap
21:54 -!- drethelin [drethelin@096-042-035-082.res.spectrum.com] has joined #hplusroadmap
21:56 < drethelin> Anyone have any opinions on Kent Kemmish? He recently made me aware of him and demonpore and he seems like he's either insane or a grifter 
22:25 -!- drethelin [drethelin@096-042-035-082.res.spectrum.com] has quit [Ping timeout: 268 seconds]
23:13 < fenn> anyone have an opinion on X? here's my opinion on X.
23:13 < fenn> seems like a bad way to get an unbiased sample
--- Log closed Thu Jun 17 00:00:40 2021