<Jack21>
hmm door opens but picture and files are not accessible
Colt has joined #openscad
othx has quit [Ping timeout: 256 seconds]
othx has joined #openscad
sauce has joined #openscad
Xeha has joined #openscad
<teepee>
fixed
<teepee>
I missed that date +%d gives zero padded result
<teepee>
now it's: grep $(printf "setDay.%d," $(date +%d)) ...
<teepee>
fun :)
<Jack21>
nice
<teepee>
indeed, knowing all the cool stuff to come does not limit the excitement at all :)
ochafik has quit [Ping timeout: 252 seconds]
<teepee>
I have not looked at how the scripts work, so still lots of cool stuff to see for 22 days
<teepee>
something is confusing that webp generator, it finished after like 2 hours or so for 500 frames
<teepee>
and 3mb seems good
<teepee>
but it glitches at the beginning for some reason, later it's fine
<Jack21>
i think there was a glitch in the source png from scad
<teepee>
good point, let me check the pngs
<teepee>
no, the pngs are fine
<teepee>
maybe it's confused by those flickering dots
<teepee>
I can try a tiny blur over the images before running through the webp encoder
<Jack21>
InPhase said that he minimized this but still had it at the beginning - but probably you seeing something else glitching
<teepee>
did you see the link in PM?
ochafik has joined #openscad
<Jack21>
ah now (always these strange sounds and never find the source) .. these are compression artifacts that is different to the glitch
ochafik has quit [Ping timeout: 256 seconds]
<teepee>
good thing there's enough time to get some sleep first
castawayc has quit [Ping timeout: 256 seconds]
castaway has quit [Ping timeout: 245 seconds]
<Jack21>
imagine he had put the tree from last year into the middle ..
castawayc has joined #openscad
castaway has joined #openscad
ochafik has joined #openscad
<InPhase>
Jack21: Those compression artifacts are so extreme it hides the glitching. ;) But that's not a worthwhile trade I think.
<Jack21>
no it isn't
<InPhase>
Hmm. kdenlive's appimage with the recent webp support appears to have its webp support broken.
<InPhase>
Ah, there's a channel on here.
<Jack21>
so lets see if my render finish in some hours - or if i bluescreen .. damn i already opened the door .. well i can do it again when i wake up - Ü
<InPhase>
Almost every door opened can be closed. You just might not be able to put back what escaped. :)
<Jack21>
that is what she said (pandora)
<InPhase>
Boxes you have to be more careful about...
<Jack21>
at least we keep the hope ..
<InPhase>
Oh...
<InPhase>
teepee: Next year, replace doors with Christmas gift packages. ;)
<Jack21>
or balls that scatter and shatter
<InPhase>
By then maybe we'll understand animation well enough to make an unwrapping effect.
ochafik has quit [Ping timeout: 256 seconds]
<InPhase>
Jack21: And when you click again they reassemble themselves?
<Jack21>
magick balls
<Jack21>
and more fireworks and glitter
GNUmoon has quit [Remote host closed the connection]
GNUmoon has joined #openscad
ferdna has joined #openscad
ochafik has joined #openscad
ochafik has quit [Ping timeout: 256 seconds]
* JakeSays
has magic alls
* JakeSays
balls too
ochafik has joined #openscad
ochafik has quit [Ping timeout: 256 seconds]
ochafik has joined #openscad
ochafik has quit [Ping timeout: 252 seconds]
ochafik has joined #openscad
ochafik has quit [Ping timeout: 252 seconds]
<ali1234>
if you want to get rid of the glitches, just re-render the images at like 4x the size and then scale them down
arebil has joined #openscad
Jack2165 has joined #openscad
Jack21 has quit [Ping timeout: 256 seconds]
<InPhase>
ali1234: I poke at some image processing ideas, but for the intended purpose as a demo, it's only fair to create what OpenSCAD itself produces.
<InPhase>
So on principle I thought it best to leave them.
<InPhase>
It's possible there's a tweak to the design that would remove them.
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
<InPhase>
In general though it's tricky to get clean previews when you smash a lot of tiny stuff together.
<InPhase>
And it's not really possible to do a multi-color polyhedron, where a polyhedron would be my normal solution to this issue.
<ali1234>
well antialiasing is not cheating... everyone does it
<lf94>
InPhase: how do you do icosahedron in scad again? IIRC sphere with low fn
<lf94>
day 1 almost seems like a give away, I'm a little sad
<lf94>
in Curv it's just `icosahedron diameter` lol
<lf94>
To derive icosahedron though is just a tedious process of intersecting a ton of half-planes
<ali1234>
hmm do i understand this right? it makes 3d "planes" with liner_extrude, then minkowskis them with spheres, then hulls the result?
<ali1234>
why not just place the spheres directly where you want them and avoid the linear_extrude/minkowski?
<ali1234>
oh i see, the idea is to make the non-rounded version first, and then round it
<InPhase>
lf94: The best icosahedron code is in peepsalot's icosahedron subdivision code I suppose. :)
<InPhase>
Although when I did a very careful inspection of those spheres, it turns out that the subdivided octahedron is the most superior choice. It is a polygonal sphere which has the fascinating property that every one of the 3 axes has the faces line up exactly with a cylinder cross-section, so there are no jagged edges when you join up spheres and cylinders.
<InPhase>
This is the sort of maximal facet economy that you can't get with SDF. :)
<InPhase>
I was sufficiently enamored with this that I've proposed that our built-in sphere should be replaced with a subdivided octahedron, giving us proper axial symmetry.
<InPhase>
One can always achieve the old behavior with rotate extrude anyway, if it's truly desired.
<InPhase>
As I interpret the guarantees of the language, we've never guaranteed the polygonal layout of the sphere, so I do not consider this a breaking change. I see it as a straight-up upgrade of 3D integrity.
<InPhase>
An illustration of what I'm talking about. The foreground is the subdivided octahedron, and on the right it's unioned with some cylinders. https://i.imgur.com/2U9pMov.png
<lf94>
> This is the sort of maximal facet economy that you can't get with SDF.
<lf94>
Not true? I could create the exact same type of icosahedron
<lf94>
You can in fact create any N sided surface as exactly as OpenSCAD can.
<lf94>
I think you're associating SDF with only "smoothed surfaces"
<lf94>
You can create hard edged cubes and everything
<lf94>
:D
<lf94>
Sorry - I was a little hostile there. *reels himself in*
ochafik has joined #openscad
<InPhase>
lf94: Well, I'm not sure how you're marching your rays to do that. But if you can show me some math description that I'd read when less sleepy, I'd appreciate that.
<lf94>
We use libfive to do it
<lf94>
When meshing
<lf94>
(only thing that matters when exporting)
Zauberfisch_ has quit [Quit: No Ping reply in 180 seconds.]
<lf94>
Alternatively no meshing is required if using a voxel format, which we support too
<teepee>
the detail screen has a footer, we could add some small links there :)
<Jack21>
I wonder if ppl like to always get download links or just want to see the code in the browser - or link to the scad via cadhub
ochafik has joined #openscad
PovilasCNC has joined #openscad
<Scopeuk>
it varies, in previous years several designs have been "screen art" rather than printable (I don't know any of this years in advance)
ochafik has quit [Ping timeout: 252 seconds]
<teepee>
yeah, I brought that up with cadhub, even thinking having a calender with multiple projects, but that did not move
<Jack21>
hmm the links can't be used for cadhub external script but via the RAW it works https://cadhub.xyz/draft/openscad#fetch_text_v1=https%3A%2F%2Fraw.githubusercontent.com%2Fopenscad%2Fopenscad-advent-calendar-2021%2Fmain%2FoSCADvent%2FIcosahedron%2FIcosahedron.scad%3Ftoken%3DAWIM52FVTOPVRURB5UGAOELBU5F4Y
<teepee>
ah, good idea
<teepee>
hmm, getting an error in main.scad for the link
<Jack21>
i getting a timeout .. rendered a different complex file before ..
<teepee>
it allows 30 seconds
<Jack21>
hmm strange another browser totally failed
<teepee>
maybe not quite ready :)
<teepee>
but yes, a code view would be nice
<Scopeuk>
currently testing a print of todays scadvent in tpu
<teepee>
cadhub will have an embed card at some point
<Scopeuk>
presumeably it cant get into the private git hub repo to grab that file?
<teepee>
probably not
<Jack21>
yeah the token of the github raw iseems to have a short lifespan
<Jack21>
it was working for some seconds though and i got the proper code in cadhub
<Jack21>
i wonder why it didn't work with the file from file.openscad
ochafik has quit [Remote host closed the connection]
arebil has joined #openscad
ochafik has joined #openscad
ochafik has quit [Ping timeout: 252 seconds]
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
arebil has joined #openscad
<InPhase>
Jack21: Yeah, sinusoidal beating gets close to what I was looking for. :) I was just trying initially (and failing) to avoid the uncanny valley of that. But it's still a cool effect. :)
<Jack21>
all random numbers with interpolation
<InPhase>
Ah.
<InPhase>
So the thing I had in my mind was where the transition rates themselves would be randomly varying.
<InPhase>
Such that the eyes go to one state in a non-linear manner, seem to hover in a particular mode for a while, then transition more rapidly to another as if it had a sudden thought or was distracted by something (depending on what piece changed), and so on.
<Jack21>
a faster or slower transition will be the result of how far the values are appart
<InPhase>
It was just tricky to generate that sort of pattern without spending a lot of time remaking it each frame. :)
<Jack21>
you could make a sub recursion so the eye shape changes .. stay and only the view changes (maybe with blinking)
<Jack21>
so every shape had 5 view changes
<Jack21>
you just need to create the new shape if count%5
<InPhase>
teepee: Well that webp is certainly less compact than webm.
<teepee>
what's the frame count of the webm?
<InPhase>
teepee: At those settings the files are 4 times larger, and have a tiny bit of residual background ghosting. Definitely some inferior algorithms in this file format. But, the quality is at least good enough for a display.
<InPhase>
30fps
<InPhase>
I set -d 33 for my img2webp line.
<InPhase>
That was the only thing I changed.
<teepee>
being an image helps a lot :)
<teepee>
I guess we can try, maybe it just works with webm too
<InPhase>
I remember the promise. html5 was supposed to make movies integrate seamlessly something like more than a decade ago! Not sure if that promise ever really manifested.
<InPhase>
There are a lot more movies around though. :)
<InPhase>
I'm trying a run through with -m 6 to see how small it gets. It's sure taking its time compressing though.
<Jack21>
some of the gif pages converting uploaded gif into movies and using a html player to display when the gif file is called
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
arebil has joined #openscad
ochafik has joined #openscad
ochafik has quit [Remote host closed the connection]
ochafik has joined #openscad
ochafik has quit [Remote host closed the connection]
ochafik has joined #openscad
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
<teepee>
yeah -m6 takes hours :)
<teepee>
oh, I take it back, it finished
<teepee>
and it's slightly bigger
aiyion has quit [Remote host closed the connection]
aiyion has joined #openscad
<teepee>
nope, looks like webm is not working
<teepee>
just pasting the urls via firefox debugger shows the webp
ochafik has quit [Remote host closed the connection]
<InPhase>
teepee: 40 minutes. But -6 took it from 19.3MB to 18.8MB. Definitely not a good return on investment for this purpose. :)
ochafik has joined #openscad
kyki has joined #openscad
arebil has joined #openscad
arebil has quit [Client Quit]
<cbmuser>
knielsen: you got any feedback from NM yet?
EkpyroticFrood has joined #openscad
ochafik has quit [Remote host closed the connection]
ochafik has joined #openscad
la1yv_b has quit [Read error: Connection reset by peer]
la1yv_b has joined #openscad
la1yv_b has quit [Remote host closed the connection]
la1yv_b has joined #openscad
PovilasCNC_ has quit [Ping timeout: 252 seconds]
ochafik has quit [Remote host closed the connection]
<InPhase>
The pupils seem to be fluttering a lot with an +x+y oscillation, but that is right in line with the type of thing I pictured.
<Jack21>
still some room for improvement like keeping eye pos in shape size .. and the animation had an issue with the focus (pupils closer than IPD)
<InPhase>
If that +x+y oscillation were more left-right, it would be pretty excellent as is.
<InPhase>
People do a lot of looking left right already.
<Jack21>
yeah i had extra x and y seperated (instead of rands [1]) but when having more x the eyes left the shape more often ..
<InPhase>
Yeah, scaling it down in amplitude for that would be fine.
<InPhase>
One doesn't typically scan all the way to the end of the eye anyway as that hurts to do too much. ;)
<Jack21>
there is that garage scene in Terminator 1 where he (schwarzenegger) carefully paid attention to move his eyes to the max and then turn the head as little as possible like a machine would do (as we would expect it )
<Jack21>
the top google results include my reddit post and prusaprinters post
<InPhase>
lf94: I challenge you to mimic each one as they are revealed. It will actually be an intriguing test to see if cases arise that don't translate well. Keep me updated if there are issues with one, as language comparison is an interest of mine. :)
<InPhase>
Day 1 was certainly an easy one. :)
<teepee>
it's not quite the same though, missing the roundings
<teepee>
but I would expect that to be not so complicated to add
<InPhase>
lf94: To be a proper mimic you should also smooth the outside, although this should be super easy with sdf, if the default icosahedron is defined radially.
<teepee>
:)
<InPhase>
lf94: We should aim for details too, to see if there are edge cases that become tricky. This is where it will get interesting.
<Jack21>
if he can not simply copy the code - this would be lots of work for the later designs
<InPhase>
Well he's allowed to look at the code as a reference as much as he wants.
<InPhase>
For example the snowballs should be easy to emulate following the mathematical model as these basically define a radial function internally already, and then plot it.
<Jack21>
yeah and as long Boolean operations work the same way - but still he need to rewrite about 200 lines of code
<Jack21>
1200
<InPhase>
It should actually be a lot shorter in an sdf approach for that one.
<InPhase>
Size comparisons will be part of the intrigue.
<InPhase>
Decomposition will be interesting though. :)
<InPhase>
Probably still doable, but maybe tricky.
<InPhase>
Is color implemented in curv?
<Jack21>
color is but not sure how and alpha
<InPhase>
lf94: Feel free to get help from others in the curv community if they're up for this style of challenge. What I'm most curious about is, what's the best possible counter-solution in the other approach?
Jack21 has quit [Quit: Client closed]
Jack21 has joined #openscad
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
<lf94>
Yep, that's the plan InPhase
<lf94>
Yes, curv has color.
<lf94>
Ah didn't know you wanted rounded corners. That is no problem. Just jump 1 isosurface :)
<lf94>
>> offset 1
ochafik has quit [Remote host closed the connection]
ochafik has joined #openscad
<Jack21>
scad should have a 3D offset where every point gets moved along his normal
<lf94>
I thought it did?
<lf94>
I guess I'm thinking CADQuery.
<lf94>
You can offset on the 2D plane
<lf94>
Jack21: how;d you find my repo so fast? XD
<Jack21>
lf94 you were first result on ddg
<lf94>
Wait what do you mean Jack21 , I'd need to rewrite 1200 lines of code?
<lf94>
Every single icosahedral will be based off the GDF paper. This user simply adds a pow() operation around the planes
<lf94>
You could do other interesting combination too actually.
<lf94>
Not just rounding
<lf94>
I'll have to play with this later
<lf94>
This is exactly why I wanted to do scadvent :)
<lf94>
(Thank you again for making something so awesome guys)
<Jack21>
still doesn't look like an equal radius - for art and animation this is great - not so much if you construct a part that should fit anywhere
<InPhase>
linext__: Well you don't need to copy and paste. You can do that with children.
<InPhase>
Or a submodule.
<InPhase>
Oh, covered already.
<lf94>
Maybe I can get an equal radius
rvt_ has quit [Ping timeout: 250 seconds]
rvt has joined #openscad
<lf94>
I should absolutely be able to get an equal radius. just imagine two planes in 3D space - you can move them outward to make space for a cylinder or sphere or whatever, or combine them using a radial equation
<lf94>
I'm just trying to think of how I could do this in a general way. All I can think of is for all GDFs, have an option specify roundness and if it's set, then use that specific combination operation
<lf94>
I think this could even work for 2D shapes too
<lf94>
Well, for all 2D SDFs you just offset to the next isosurface...
<lf94>
I guess same for a lot of 3D SDFs too.
<lf94>
cube >> offset 1 works as you expect: rounded edges and corners
ochafik has quit [Remote host closed the connection]
LordOfBikes has quit [Quit: Ex-Chat]
LordOfBikes has joined #openscad
Guest21 has joined #openscad
Guest21 has quit [Client Quit]
LordOfBikes has quit [Remote host closed the connection]
GNUmoon has quit [Ping timeout: 276 seconds]
LordOfBikes has joined #openscad
GNUmoon has joined #openscad
ochafik has joined #openscad
ochafik has quit [Remote host closed the connection]
ochafik has joined #openscad
ochafik has quit [Remote host closed the connection]