>>1031316
Don't extrude the faces, cut loops into both sides and build faces from that

File: IMG_1322.png (650 KB, 1136x640)

650 KB PNG

>>1031316
Blender can’t fill in the holes while it’s a common feature in every 3D software.

IT'S BEEN 3 FUCKING YEARS GET A NEW FUCKING JOKE
I JUST RETURNED TO THIS GODFORSAKEN SITE AND THIS SHIT IS THE FIRST THING I FUCKING SEE
FUCK OFF

>>1029101
Sounds like you need my BWC to calm down. I'll stretch your bussy good.

File: Slopjak_contacts.jpg (705 KB, 1299x1071)

705 KB JPG

>>>1029101
>Sounds like you need my BWC to calm down. I'll stretch your bussy good.

>>1025602
Whenever I think my art is bad and I'm about to cry I look at some of cris' work. It's therapeutic seeing my progression versus his

File: 1764133954265699.jpg (92 KB, 1220x521)

92 KB JPG

How many years has it been?

>need a part for a project
>'hmm... I'm going to need like four different variants of this part, and I don't want to spend an hour re-building it each time'
>use geometry nodes
>takes 12 hours to make
why do I do this to myself?

>>1031284
if you only made it to specifically make those 4 variations then I'm sorry for your loss but at least you had fun making it right?
>built geonodes asset that excels at a very specific niche purpose
>never have to rebuild it again as long as you don't have multiple storage drive failures

File: Wrinkles 01.png (1.25 MB, 1920x1080)

1.25 MB PNG

What are some good methods of making things look like cloth without actually using the cloth physics system? Ideally compute-lite.
Squashed Voronoi with smooth F1, distance fed through a float curve, used for displacement and bump gives a vaguely cloth-like wrinkle effect, but only looks at all good if the mesh density is fairly high (typical face size smaller than each Voronoi cell) and the object is tube-shaped.

>>1031087
>>1031077
I worked on something similar to this before. >>1028032
Using geometry nodes, I got as far as creating spikes that seamlessly conform to the base mesh. That much is doable. To take it further beyond is likely possible. But I'm not sure I have the capabilities either. Looking back at the node group now, I forgot some of the functions.

The broad concept, is that you want to convert the base mesh(or scalp) into the profile for the curves. Think about it like this: if you draw a line from one edge to another edge, it will always be flush with the face.(ignoring the hidden triangulation of the face)
Now imagine drawing a circle around the curve. Each edge that you crossed while drawing, should have a point on it. And if you can connect those points with edges, you will have a flush profile.

I kind of forgot how I managed that. But I did. You can start by having the base mesh sample the nearest curves. When a mesh samples a point cloud, it creates a vague sketch of a voronoi pattern. Each section of the mesh knows which point it's closest to. Creating the notion of boundaries where points meet halfway. And the thing about those boundaries, is that often they cross through edges. One side of an edge is within one boundary, and the other side of an edge is within the neighboring boundary. These edges that exist in both boundaries, are the edges you want to convert into points to create your profiles.

Because an edge can sample the position of both points, it can subtract one point from the other to get the normal of the boundary. And with that normal, you can find the position where the boundaries intersects with the edges.(I used higgsas's "line plane intersection" node to do the math for me)

There is a lot more to this that I'm omitting for brevity's sake. But I hope that will give you some insight.
My node can generate 1,500 spikes in 500ms.(think light yagami haircut length) It's not good for real time. But it's usable.

File: Blender Tank Tracks 01.png (1.26 MB, 1733x777)

1.26 MB PNG

How to make actually good tank tracks? Ideally as procedural as possible.
Wheels aren't that hard, although making them procedural without distorting features like bolt heads and circular cut-outs is tricky.
The track is difficult. Can't just array and fit to curve, because that causes distortion. Instead need to make a curve, resample curve, curve to mesh to make a ribbon of faces, then instance links on faces and connectors on points. Actually creating each link probably means modelling by hand rather than procedural, because they vary so much. Resample curve squashes things a little, but links can be slightly scaled in one dimension without it being noticeable.
Generating the curve to place the links is another challenge. Easy enough if the track is taut and on level ground (i.e. entirely convex around the wheels). A bit more complex to fit to terrain. Haven't figured out how to make it sag in a plausible way without breaking out the simulation stuff.
Suspension is a nightmare. Multiple different kinds, but the main one in the modern day is torsion bar. Torsion bar doesn't go straight up and down, but swings on an arm. Got to figure out how far to swing each arm to keep the wheel above ground. Fortunately I don't need to animate this, so I can just do it by hand if absolutely necessary.
Drive sprocket would be a headache if animating, but fortunately able to just do it by eye for now.

>>1027625
>I used to place hundreds of lights by hand
Does anyone have his portfolio?

>>1027615
https://www.artstation.com/artwork/JbVxA
Kek he is a grifter and a faker.

>>1031198
>https://www.artstation.com/artwork/JbVxA
get help

>>1030960
This guy gets it. AI is good and bad at various things as it mostly depends on the data. If there is lots of data around for what you want to train, then your model will be very good at that task, but if you want your AI model to be superhuman in a task you need to have a task with verifiable reward, and 3D to some extent is that. You can have a picture of a thing, then picture of the same thing from multiple angles and then you make it generate the thing until all the photos from those angles match the model. Idk how it could do superhuman topology, but this method is how AI became so good at math and chess.

>>1030970
I agree, I think a lot of people think that AI is a zero sum game and that every single ai asset is an asset that would have been made by a human, but to your point, if an indie dev or animator doesn't have the time to make a fire extinguisher, or the money to hire an indian to make a fire extinguisher, and has a style incongruous with all of the fire extinguishers on the Unreal store, they're either going to use ai or simply not add a fire extinguisher.

File: old.gif (1.31 MB, 498x270)

1.31 MB GIF

>>1029647
I have a feeling youths these days no longer create geometry vertex by vertex. I don't think they know how to massage vertices, even

>>1029649
stick to digging holes in the ground

File: 1532062319724.jpg (46 KB, 720x480)

46 KB JPG

I recently been interested in this art style not for nostalgia but because I want to make a game that reaches a wide audience by having low system requirement.
Can a game on this style run on a laptop with integrated graphics or what should I aim for?

>>1031295
No, it is impossible for modern GPUs to run the visual quality from 30 years ago

>>1031295
I remember playing a homebrew version of quake on the nintendo ds. I don't know what the modern standards are but if your hardware is better than the nds i think you're good

>>1030296
just subdiv, sculpt and retopo of course!
you can also add a boolean and some hair particles if ya nasty
glad i could help.

been lurking in and out since i don't even know how long, didn't you have threads in early 2025? Either way, awesome to see you make progress over time. I admire your art. Do you post anywhere else online?

>>1030058
hair seems odd, like it's weighted to random bones

do you not want to use a bone chain with physics?

>>1030055
yum, blueberries...

>>1030055
>>1030058
>>1030059
im glad to see your thread still up after all this time, its been several months since i last checked this board and its nice seeing that you're still doing this.

>>1030911
you are a cris enjoyer and don't know about /agdg/ on /vg/?

>>1030911
it's so sad that cris only got the recognition and appreciation he deserved after he died. many such cases
a lesson to us all. say something nice about a fellow anon's art while you have the chance

>>1030911
he's been spamming 4chan for over 10 years

>>1030031
Wow. I was last in here several years ago and you actually have improved. Arguably you've doubled your skill in the same time frame an ordinary artist would. Your rate of progression is actually normal; you just started ~4 magnitudes behind everyone else so it took you until now to appear as this wasn't your first honest attempt.

File: abcfb057-6662-42c3-85c7-a(...).png (118 KB, 645x302)

118 KB PNG

Chris? You ok? It was rough in Cali. Hope you and your family are safe.

>>1031292
i remember anon post it here on previous thread, just hope someone repost it again here

>>1031298
Thanks, I'll check the archive

>>1031299
my bad, it's "few previous thread" not previous thread

File: file.png (19 KB, 924x337)

19 KB PNG

>>1031300
Found it but not available... >>998164
But I'm pretty sure I remember that post so I might have downloaded it
I'll re-upload if I have it somewhere in my pile of downloads

File: file.png (2.79 MB, 1360x1920)

2.79 MB PNG

>>1031301
>>1031300
FUCKS SAKE I NEED TO ORGANIZE MY SHIT I HAD IT THE WHOLE TIME
I only have the free mega account, can someone upload it to a better place
https://mega.nz/folder/IaZ3xaAR#tHiVyskrmRPZ5pSzSirvuQ

>>1030874
can you upload it to other site? mega have download limit

>>1031185
nvm, found it on tele

>>1031190
nvm, 2nd part dead

watching alive course, after the squirrel part, the rest of the video is "using your feel and eyeball to make it look right", he teach you everything you need to know on the squirrel part, the rest is a good watch if you want some tips and trick

>>1030995
Looks interesting! Do you have a torrent for it?

>>1031223
Oh no not military bootcamp the thing that gets you in top physical shape while teaching you dozen of skills in record time

>>1031273
US Marine Corps bootcamp, at least, isn't really meant to get recruits in physical shape from nothing. They're are expected to get there with already a certain degree physical and mental preparation, and the "training" is mostly intended to break recruits out of their civilian habits and way of thinking with a deliberately intense, high-stress, demanding environment. Anyway, that's off-topic here.

To get better at drawing/painting, you need to iterate quickly, be motivated, practice seriously and consistently, incorporate theory into your art (not just studies for the sake of them), try to make finished works and not only "practice runs", look for feedback from people who don't have a vested interest in slowing or dragging you down. Just practicing doing old muscular men in stiff academic poses won't teach you how to make appealing anime/sexy girl artwork. Some skills will transfer, but others even will work against that goal.

>>1031276
You'd argue anything in order to not better your skills

>>1031231
You only get garbage 8 week courses for $3k where they show you the basics and "critique" the few beginner projects you manage to do, while also listening to the retarded shit that the other victims ask.
Books have infinitely more value. If you're at such a high level, that books can't help you anymore, you need to fuck around and find out for yourself.

>>1031280
which books would you recommend?

File: new bottle 02.png (1.41 MB, 1385x899)

1.41 MB PNG

This whole thing started as me trying to make a good version of these paint bottles. I did them but didn't really like how fake/glassy the bottles looked. And after a lot of screwing around with shaders I have one that I think is a lot better.
The label can rotate around the bottle so that you can rotate the plastic part and see the mould line some of the time. I also changed to a more straightforward copy of the real paint's label so it looks less odd just because it's unfamiliar. Right now there is just a bit of a crease in the label but I have to add a couple more weathering variations, make the color name a mask, and make the red part of the label driven by the central color like my last version.

Old ones on the right, new one on the left.

File: 1784078835139826.jpg (1.65 MB, 3840x2652)

1.65 MB JPG

>>1031002
Added some stuff

File: test.png (342 KB, 847x852)

342 KB PNG

I'm trying to recreate a 90s CGI feel. How does this look? Is she cute or is something fucked up with her face? I've been working on it almost nonstop since yesterday so I genuinely can't tell anymore.

>>1031228
Looks great, anon. Are you using geometry nodes for the vines?

File: 1767077786334434.jpg (1.96 MB, 3840x2600)

1.96 MB JPG

>>1031286
I'm using Ivy Generator
It's a free addon by antoine bagattini

https://www.youtube.com/watch?v=TAMY3FfED6g

File: GGST_A.B.A_2S.png (268 KB, 484x870)

268 KB PNG

>>1031270
Analyze the character design. They're horns treated like floppy ears, model them as such

File: file.png (371 KB, 822x1038)

371 KB PNG

>>1031274
well the end goal is to 3d print it as a solid piece and im struggling to think of the proper technique to make them taper into the main face.
maybe just manually extruding edges to match the reference would be the best way

also i feel bad you for guys in here i had no idea cris spammed this board so much.