|
that's 1/2/3. at g0 you don't have tangent continuity, so there is a visible edge where the surfaces join, and the curvature graph is broken. here is another way of showing it
|
# ? Sep 21, 2021 21:17 |
|
|
# ? May 25, 2024 12:43 |
|
are the inverse(?) of these ever used? like the inverse of g1 would be with the hill of inverse of the radius in the middle
|
# ? Sep 21, 2021 21:22 |
|
How do you do this in Blender? I know there's CAD software with tools for it but I didn't think Blender did.
|
# ? Sep 21, 2021 21:23 |
|
Elukka posted:How do you do this in Blender? I know there's CAD software with tools for it but I didn't think Blender did. strictly you can’t, it requires nurbs, and while blender has some nurbs functionality it’s minor and approximately .. idk how possible it is either the closest thing i’ve found is in the bevel tool or modifier you can change the profile, with a 0.5 profile being an ordinary fillet and somewhere between 0.6-0.8 looking like it approximates one of these g2 curves
|
# ? Sep 21, 2021 21:37 |
|
I mean if all else fails you can always generate the geometry with a python script but I doubt that Sagebrush would go there for a class exercise
|
# ? Sep 21, 2021 21:52 |
|
Trabisnikof posted:are the inverse(?) of these ever used? not quite sure what you mean. like a graph that shows radius directly instead of curvature, so it's smaller at the sharper edges? i guess you could do that, but this way seems more practical since most surfaces are generally flat-ish (i.e. very large or infinite radius) so you'd have huge graphs on the flat parts you don't care about, and tiny ones in the tricky details. there are all kinds of other diagnostic shaders though, like the continuity analyzer (0.000 is good) and the zebra stripes (slippery snakes are good) and curvature analysis (oof, don't like that blue sink mark in the middle of the blend) and just plain old reflection mapping and looking for weird stuff i know this isn't blender stuff but idk, i like surfacing and the extremely deep dive you can do into the exact shape of an artifact Sagebrush fucked around with this message at 22:23 on Sep 21, 2021 |
# ? Sep 21, 2021 22:21 |
|
those look cool and i wasn't really sure what i meant either, just noticing the shape of the graph and wondering if the curve that would have the opposite pattern of linear change would ever be used or if the answer is "nah" like the underlying curve that would cause this graph
|
# ? Sep 21, 2021 22:28 |
|
ah i see. well that would just be a g2 curve, because you don't have that sudden jump in curvature at the edge. it would have a slightly rounder/pointier profile from the other g2, but it would still be continuous.
|
# ? Sep 21, 2021 22:55 |
|
gotcha that makes sense
|
# ? Sep 21, 2021 23:01 |
|
good times, Cycles X just got merged into the non-experimental daily branch of Blender 3 https://twitter.com/BlenderDev/status/1440325606386769932 which I think means it will now be guaranteed to be included in the mainline release of 3.0 in December, if I understand the general software development workflow right.
|
# ? Sep 21, 2021 23:11 |
|
does cycles x bring anything new to the table or just faster the greatest thing I miss which is basically a niche process is bi-directional path tracing. shame have to use luxrender to get that. welp
|
# ? Sep 21, 2021 23:41 |
|
echinopsis posted:does cycles x bring anything new to the table or just faster cycles x also progressively renders the whole frame instead of using tiling, and it’s optimized much better for the viewport. supposedly they’re going to fix volumetrics as well? and improve shadow catchers somehow fart simpson fucked around with this message at 03:18 on Sep 22, 2021 |
# ? Sep 22, 2021 03:16 |
|
for a challenge i made the yospos meme in the same curvature-continuous style as the R, but it's amazing how indistinguishable it is at this scale from a regular old extrusion + fillet that would take 8 seconds. lol
|
# ? Sep 22, 2021 03:27 |
|
fart simpson posted:cycles x also progressively renders the whole frame instead of using tiling, and it’s optimized much better for the viewport. supposedly they’re going to fix volumetrics as well? and improve shadow catchers somehow progressive rendering is available already .. and I hate it often I want to see how something is going to come out at a certain sample depth and tiling allows that but full frame rendering means waiting for the whole thing to finish not just a small square. I have my tiles set to 8x8 it’s cute
|
# ? Sep 22, 2021 04:05 |
|
echinopsis posted:progressive rendering is available already .. you can still use render regions for that, also i think if you're using gpu small tile sizes like that slow it down
|
# ? Sep 22, 2021 04:07 |
|
the sweet spot for pure speed is around 32x32 on my machine but a few extra seconds over the whole pic isn’t a big deal and the small wee pixel crunchers are cute i have one of those 24 thread cpu’s so have a whole lot of tiles at once still. suppose it doesn’t make a massive difference .. but whenever I turn on progressive i’m always gutted at how bogus it is to watch do it’s job. if the end result is the same it’s hard to complain about tho
|
# ? Sep 22, 2021 05:33 |
|
echinopsis posted:the sweet spot for pure speed is around 32x32 on my machine but a few extra seconds over the whole pic isn’t a big deal and the small wee pixel crunchers are cute progressive rendering in cycles x is way faster and as said above you can ctrl + b out a render region as a test. it’s straight up better than tile based imo
|
# ? Sep 22, 2021 05:36 |
|
current cycles progressive rendering is way slower than tiles which is the main reason nobody uses it
|
# ? Sep 22, 2021 05:37 |
|
better =/= better
|
# ? Sep 22, 2021 05:48 |
|
Sagebrush posted:for the in-class exercise today i made a bubbly letter R with full curvature continuity (G3, actually, which is one better) everywhere. actually just seen this on big screen and this is so nice that t intersection in the middle of the letter is done so absolutely loving well
|
# ? Sep 22, 2021 11:41 |
|
Sagebrush posted:for the in-class exercise today i made a bubbly letter R with full curvature continuity (G3, actually, which is one better) everywhere. Wait, is this some crazy subdivision or nurbs or what?
|
# ? Sep 22, 2021 22:28 |
|
pure NURBS baby
|
# ? Sep 23, 2021 00:08 |
|
would love to get it pissed off and acting irrationally
|
# ? Sep 23, 2021 01:23 |
|
Hell yeah curvature talk give me more of that poo poo, it's fascinating
|
# ? Sep 23, 2021 15:20 |
|
brushing up on my nodes for nodevember. watched that erindale video i posted. now i figured out how to do this: https://i.imgur.com/AEBSv5u.mp4
|
# ? Sep 25, 2021 09:07 |
|
idk man seems like it violates some kind of fundametnal law I'm sticking to my tried and true method.... something spinning https://giant.gfycat.com/BountifulRawChital.mp4
|
# ? Sep 25, 2021 09:25 |
|
echinopsis posted:idk man seems like it violates some kind of fundametnal law i like the analog warmth of the lack of samples
|
# ? Sep 25, 2021 09:28 |
|
I tried to make my spaceship spin and that exposed how terrible the geometry actually is so I stopped. Subsequent ones should have less terrible geometry. I have embraced the sports car paint of shame, the shiny red matcap which exposes the slightest flaw in surfaces.
|
# ? Sep 25, 2021 10:12 |
|
fart simpson posted:i like the analog warmth of the lack of samples it's as many samples as I normally use, but it must be because a lot of it is lit from the sky rather than the sun. also, I realise now, if I had turned off random seed for the sampling, it would still be as noisy but the noise would be moving. ALSO.. orthographic projection makes poo poo look weird. im downloadning 3.0 and seeing if cycles x wil do me any favour. if I can suffer through the progressive rendering Elukka posted:I tried to make my spaceship spin and that exposed how terrible the geometry actually is so I stopped. you betta fuckin post it haven't you seen the garbage I am prepared to post? you're in good hands
|
# ? Sep 25, 2021 10:48 |
|
anyone ever done much with blender and python I've asked in the past if geometry nodes can be programmed and someone said nah you can't make loops or actually make it execute a program, can you use the python environment to create a shader program? a geometry shader program?
|
# ? Sep 25, 2021 11:18 |
|
i used python to make some manually-iterated fractals when i first started using blender and it was fine but it's basically just typing the things you would be inputting into the boxes on the nodes anyway so i'm not sure it would be more useful than the nodes themselves unless you find yourself making a bunch of nested/iterated nodes by hand
|
# ? Sep 25, 2021 16:34 |
|
Corla Plankun posted:unless you find yourself making a bunch of nested/iterated nodes by hand have you heard of nodevember lol
|
# ? Sep 25, 2021 16:42 |
|
this was a node group inside a node group i did for nodevember last year
|
# ? Sep 25, 2021 16:47 |
|
fart simpson posted:this was a node group inside a node group i did for nodevember last year so much tidier and prettier than my last big node project
|
# ? Sep 25, 2021 18:38 |
|
lol
|
# ? Sep 26, 2021 01:22 |
|
I’m uh, “codate”
|
# ? Sep 26, 2021 01:55 |
|
so this noisy image I've been loving around with cycles ; 4:30 cyclex x : 2:03 at least twice as fast. I like it a lot tbh
|
# ? Sep 26, 2021 06:17 |
|
echinopsis posted:so this noisy image I've been loving around with i told you bro
|
# ? Sep 26, 2021 06:35 |
|
you had my back I know I was being a stick in te mud. pathetic
|
# ? Sep 26, 2021 06:38 |
|
|
# ? May 25, 2024 12:43 |
|
echinopsis posted:you betta fuckin post it haven't you seen the garbage I am prepared to post? you're in good hands However in the interests of posting here's another spaceship that's probably a bit cleaner. Also one of the few things I've finished. Wish I had more time for art. It also comes in red. Then someone suggested to me I should go full Foss so I did that
|
# ? Sep 26, 2021 09:15 |