[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/3/ - 3DCG


View post   

File: 550 KB, 988x988, Blender-Splash-Fox.jpg [View same] [iqdb] [saucenao] [google]
973562 No.973562 [Reply] [Original]

Previous: >>966941

>> No.973564

Blender Today is ogre.
Fucking bummer because I enjoyed that show

>> No.973567
File: 861 KB, 1124x1928, Wooden-Mannequin.jpg [View same] [iqdb] [saucenao] [google]
973567

Still looking for rigged mannequins I can use to pose and display people to then use as templates to draw over.
1. where can I get a good asset for blender that is already rigged and allows me to put them into any number of positions?
2. Should I draw in blender or just stick with krita?
3. No I am not looking for plugins like MLab or something to create photorealistic models, just a single asset with roughly correct proportions i can download and put into blender

>> No.973570

>>973567
Just learn to draw tracing nigger

>> No.973580

>>973570
I literally want to create images to trace you retard.

>> No.973584

>>973567
https://www.mixamo.com/

>> No.973586

>>973567
https://www.cgtrader.com/free-3d-models/character/anatomy/wooden-mannequin-for-artists
Took me like 2 minutes

>> No.973588
File: 197 KB, 1280x1280, 8306a675949b6f699db1f26371f21bca99b1fc036a8ea122f5fd92d9e79db7db.jpg [View same] [iqdb] [saucenao] [google]
973588

>>973562
So Blender hired this week the Cuban that made the Dyntopo mode 200 years ago to work in the sculpt mode and the mac version, he has worked on 3d coat and Zbrush too, Farsthary.
https://www.youtube.com/@farsthary1984

>> No.973589

>>973588
Holy fuck that's great news

>> No.973590

>>973588
This is huge

>> No.973592

>>973589
More news after the sculpt meeting on Tuesday :^) , maybe tomorrow in Blender today too.

>> No.973594

>>973592
>tomorrow
Monday

>> No.973595
File: 45 KB, 408x612, 1659140555181430.jpg [View same] [iqdb] [saucenao] [google]
973595

>>973594
The day Anon discovers timezones

>> No.973596

>>973589
It's great news for those who care about sculping in Blender. For those 5 people.

>> No.973597
File: 23 KB, 474x546, 1600513320244.jpg [View same] [iqdb] [saucenao] [google]
973597

>>973596
Yes, and?

>> No.973603

>>973597
Nothing. I'm sure >>939000 is going to benefit from it greatly. And that's a good thing.

>> No.973615

But most importantly, if the guy expands the scope and use of Dyntopo, it's not like you're going to complain because it's not ALL QUADS? Are you?

>> No.973620

>>973588
ahh i remember that guy

>> No.973621

>>973567
you should just make one yourself using subsurfed cubes and the rigify basic human template, but some twitter grifter's been peddling a premade one lately; here's a copy from cgpersia: https://files.catbox.moe/kog7zi.zip

>> No.973888

Is the new Beta stable enough?

>> No.973890

>>973888
No, crashed 4 times today while making hair cards, going back to an stable version

>> No.973908
File: 67 KB, 1123x286, 1694396600980704.png [View same] [iqdb] [saucenao] [google]
973908

What does this mean. I just went from 3.4 to 4.0 by the way. Losing my shit over here

>> No.973909
File: 5 KB, 136x88, 2024-02-13-1707844290_136x88_.png [View same] [iqdb] [saucenao] [google]
973909

>>973908
>BlenderGIS
You're trying to use a lazily maintained addon that got updated last almost two years ago and a lot has changed under the hood of Blender with version 4.0.
If you want to keep using it you need to keep using your old version of Blender or get someone to fix its incompatibility with 4.x for you.

>> No.973912

>>973909
Yeah, I just downloaded 3.6.8 and the addon works like it used to. thanks

>> No.973914

>>973912
Hey no prob

>> No.973928

Is there a way to change which direction Blender thinks is the front? I want it to be positive X because it makes the most sense.

>> No.973984

>>973928
A took a quick look, but nah, I don't think so.
Personally, don't think +X being the front makes any kind of sense at all. +Y is forward, -Y is back. It'll save you a fuckload of troubleshooting down the line when you start fucking around with objects on paths.

>> No.974254

starting to sculpt within blender, but I'm wondering if there's another piece of software similar to these videos that allows me to use booleans to carve out shapes using another shape, and sort of 'stamp' shapes down
I know you can use booleans and metaballs in blender, but it's not quite as fast as this
also, the way you can get a basic form, stretch it out and curve it very quickly is pretty appealing, but the piece of shit is annoyingly limited to the PS4

https://www.youtube.com/live/q8iJpblGowY?si=h_U8QpeXT19QoWGr&t=1079

https://youtu.be/pNcv6QtUcmA?si=IfWEqpvivOGarsEv&t=45

>> No.974256

>>974254
actually ignore that, looks Substance 3D Modeler is what I'm looking for. oddly enough I never ever see anyone discussing this here

https://youtu.be/HkVReuxEpq4?si=rGFJ_-0X06oA8sc6&t=344

>> No.974259

>>974256
all that is based on signed distant fields, you have a fuckton of programs, a paid addon and a couple of nodes in blender, experimental
https://www.youtube.com/watch?v=9V-IwE586m4
https://www.youtube.com/watch?v=kBSlTWb-uN0

>> No.974267

>>974256
>I never ever see anyone discussing this here
Boy, I wonder why.

>> No.974272

>>974267
>>974256
We have a thread here >>972553

>> No.974336
File: 933 KB, 128x128, 1702466099350834.gif [View same] [iqdb] [saucenao] [google]
974336

>blogpost on animation 2025
>big design info dump in the repo
>quickly becomes obvious they're going to dick around reinventing the wheel for 2 years instead of just making the NLA a linkable datablock and fixing the pose breakdown tool to work with "combine" strips

>> No.974354

>>974272
Looked more like a 3d printing thread, going by the OP image. Figured the "smoothing" was more about smoothing Resin/PLA prints, rather than making smooth CG objects.

>>974336
NOOOO you're supposed to look at the pretty graphics, not want actual feature improvements to dead aspects of the software

>> No.974361
File: 128 KB, 1010x1080, Camera Raw.jpg [View same] [iqdb] [saucenao] [google]
974361

Hey dudes, I'm doing something retarded and I'm trying to "convert" (using that term very loosely) some Adobe Camera Raw presets to Blender's compositor.
90% of it is pretty self explanatory and have a parallel, RGB curves to RGB Curves node, HSL Mixer to the Hue Correct node, that's like most of the heavy lifting.
Whether or not the numbers match up is another story. The story is, that they don't.

Pic related though are two sections that I can't really find a good parallel for.
Notably the "Highlights, Shadows, Whites, Blacks" in the basic, and all the sliders in the "Calibration" section.
For the first few, I thought the "Color Correction" node would be the analog, but I don't think so now, as it doesn't really do the same thing at all. I've tried searching it up to see if anyone has tried it, but all I get are PS tutorials by Indians and stuff.
For the second one with the RGB primaries, I don't think there's a node that handles it, but I think I can pull something by separating the RGB, and adjusting that way.

I guess if the math is available for the highlights/shadows stuff, I might be able to hack something together though if there's not a node that does that.

>> No.974503
File: 1.04 MB, 1920x1803, Compare.jpg [View same] [iqdb] [saucenao] [google]
974503

>>974361
Alright, a little progress on this, but somehow I've dug myself into a hole. The biggest factor in how the CR presets look is down to the RGB curves (what a surprise). Meaning all the extra shit is pretty much a non issue.
Apparently curve adjusters aren't built equal, though, even when you use the exact same settings/curves.

Pic related is a comparison between Adobe Camera Raw's curves (which I'm trying to "port" over to Blender's), Photoshop's curves adjustment layer, and finally, Blender's RGB curves. All three are the same image, sRGB with no transforms.
As you can see, they're all different. Even within Photoshop their own curve adjusters are completely different in both CR and the adj. layer. As you can see, the curves are the exact same, and the histograms (between CR and PS) are pretty similar.

What I don't fucking get, is why the fuck Blender is so different. It's a fucking mystery that I have no clue how to solve. It's as if its curves act fundamentally different to how it works in literally everything else. I've found a few posts online of people saying the same thing, but all the replies are essentially "no u".

Anyone know what the fuck is going on here? Like at worse, I had expected it to look like PS's curves, "close but not quite there". Not what I'm seeing now.

>> No.974504

>>974503
And yes, I've tried using the standard and filmic settings on Blender's RGB curves (filmic looks like shit in this case). I also converted PS's curve values by dividing them by 255, to fit into Blender's retarded 0-1 value system for the curves.
If I figure this shit out, it shouldn't be too hard to write some script or something to import the presets into Blender's compositor. I've got a fuckload of VSCO presets that'd be nifty to just drag in via the asset browser.

>> No.974508
File: 470 KB, 2060x744, hdsfsd7f.png [View same] [iqdb] [saucenao] [google]
974508

tried making a curved border around this organic looking shape but can't figure out a proper way of doing it
my method for this was:
>set cursor to point on mesh
>create plane
>merge verts into a single vertice
>turn on face snapping
>extrude vertice all around the object
>convert to curve
>add depth
>convert back to mesh
feels like a pretty retarded and needlessly complex way of doing something so simple

>> No.974509
File: 35 KB, 577x577, 1559205568214.jpg [View same] [iqdb] [saucenao] [google]
974509

Imagine sculpting in Blender before remeshing was added

>> No.974515

Do you think that we will have this https://projects.blender.org/blender/blender/pulls/113270 before the end of the year? it is the last core shit I'm waitting for.
>>974509
Cris, Blender used to have better performance before the 2.8 and the new viewport.

>> No.974517
File: 499 KB, 1411x1025, hashd7.png [View same] [iqdb] [saucenao] [google]
974517

>>974508
never mind, got it
>select edges
>separate by selection
>convert to curve
>add depth
>convert back to mesh
works perfectly. also give me one reason why I shouldn't model everything using subdivision surface and the cage option
the topology is always seemingly perfect and I have no intention to rig, animate or do anything to it outside of blender, and texturing will be a piece of piss since I'll be using a procedural clay shader

>> No.974637
File: 933 KB, 1581x932, blender_7Zh8jFZQHp.gif [View same] [iqdb] [saucenao] [google]
974637

>> No.974645
File: 3.22 MB, 1920x1080, 0001-0110.webm [View same] [iqdb] [saucenao] [google]
974645

I don't know how i fell into this, i saw this stuff being handed around in parties, i thought it was some harmless thing like Xanax or PCP, but it was worse, far worse, this man Andrew gave me a taste, but despite looking like an ordinary torus shaped pastry, it took over my life, i got hooked... started selling feet pics to be able to afford a dose...

>> No.974680
File: 82 KB, 1300x956, 1681513248626.jpg [View same] [iqdb] [saucenao] [google]
974680

Does anyone know how to achieve this effect but with my character's skin instead of a balloon? Basically I want the skin to deform when interacting with another object.

>> No.974687

is it true they're gonna put in something akin to zremesher/dynamesh? not that I mind reupping quad remesh with a fake email every 30 days but would be nice.

>> No.974720
File: 3.23 MB, 1920x1080, Donny.webm [View same] [iqdb] [saucenao] [google]
974720

>>974645
Fixxxx.

Truly the genre defining masterwork of donutvania games.

>> No.974727

>>974687
Where did you hear that? That would be wonderful.
The abysmal quadriflow remesher is still in, I think, but it can't even handle symmetry

>> No.974730

>>974680
soft body simulation/cloth simulation with internal springs
use a lower res mesh for the simulation and mesh deform / surface deform on modifier on the mesh that should have the effect.
shrinkwrap (project) and corrective smooth (bind) also works
Neither are a perfect solution, in theory someone could figure something out with geometry nodes

>> No.974744

>>974730
>use a lower res mesh for the simulation and mesh deform / surface deform on modifier on the mesh that should have the effect.
>shrinkwrap (project) and corrective smooth (bind) also works
thanks for the answer. Could you elaborate on those two methods? Which modifiers do I have to put on my character and which on the object that interacts with it? I'm a noob when it comes to character animation with rigs

>> No.974783

Got a specific simple task Ive gotta do, but no idea how to do it in "the best" way. There is 2 identical objects, but they have different location and rotation, already applied. I need for the second boject to match these of the first object, what do? IVe tried shrinkwrap, but it tends to fuck up internal geo. And it has internal geo. No, its not high poly modeled vagina.

>> No.974787

>>974783
If they are identical, what property of the duplicate object would you lose by just creating a copy of the object which is in the correct orientation already?
If there are some modifiers or animation keys you don't want to lose, you could copy the in-position object, then ctrl+j that copy to the duplicate object, then delete all the vertices except the ones you just joined.

>> No.974788

What's the fucking point when AI can do all of this shit. It's over.

>> No.974791

>>974787
One object is a "combination" of a few different objects, the other has 1 material, with new UVs

>> No.974793

>>974791
I'm sure a blender expert knows some more correct way, but here are my thoughts.

You can get it in the correct position by picking a recognizable vertex, selecting all the verts with that vertex active (by deselecting+reselecting it or w/e) and than enabling snap to vertex with Snap With set to Active.
Then you can snap the 3d cursor to that vertex, set the 3d cursor as the pivot, and rotate it to the correct orientation (almost) with regular and trackball rotation.
Once it's close enough, maybe your shrinkwrap will work okay.

You could also figure it out mathematically/write a script, but that seems unnecessarily complex.
As an example, once you have the translation correct using your first reference vert, you can pick another vertex and figure out how to rotate so that it would be in the correct position. Then you have two fixed points which define a line of rotation. You can pick a third reference point and rotate around that line to get the final exact orientation.

>> No.974794

>>974793
Yeah, thats pretty much what Ive been doing, hope there is some better way. Thanks anyway, gues I just gotta grind

>> No.974796

>>974794
might be able to do something fancy with the data transfer modifier if you just need the uvs

>> No.974797

>>974796
Isnt it usually need objects to be in same-ish position?

>> No.974799

>>974797
I tried this with a couple of suzannes and it worked despite different orientations and locations
https://docs.blender.org/manual/en/2.80/editors/uv/layout_management.html#transferring-uv-maps

>> No.974804

>>974788
Might as well off yourself if you're so scared of AI and can't think of a good reason why people who like making things are making things.
Frozen pizzas made by machines in a factory didn't make pizza places that make it from scratch obsolete.

You're just looking for an excuse to quit because you never actually did shit in the first place. You were always mediocre, but you'll blame AI as an easy scapegoat as to why you could never "make it".

>> No.974816

>>974515
>Blender used to have better performance before the 2.8 and the new viewport.
Objectively untrue.
Blender's performance went to shit for a short while when 2.8 happened but since then they rebounded and it's WAY faster now.

>> No.974831 [DELETED] 
File: 13 KB, 474x266, th-280334064.jpg [View same] [iqdb] [saucenao] [google]
974831

>>974816
Oppenheimer

>> No.975050
File: 1.57 MB, 410x293, 1708556461740119.gif [View same] [iqdb] [saucenao] [google]
975050

>> No.975087

>>975050
that's awesome

>> No.975096

>>975050
security cam footage from the Blender Headquarters

>> No.975102
File: 190 KB, 624x319, 1685075631024795.png [View same] [iqdb] [saucenao] [google]
975102

I fucking hate using curves so much. Cant think of any other ways to achieve this shape though

>> No.975106

>>975050
needs the audio

>> No.975110
File: 123 KB, 526x547, Screenshot 2024-02-17 211925.png [View same] [iqdb] [saucenao] [google]
975110

How do I transfer the weights for only specific parts of my mesh? The hands for my two rigs don't line up but everything else does.

>> No.975111

>>975110
use the vertex group field at the top of the data transfer modifier as a mask, and apply the modifier?

>> No.975119

How useful is boxcutter? should I pick it up while it's on sale?

>> No.975126

>>975119
Isn't that one of the ones you can pick up on github for free? There's quite a few paid addons that just have their source code available if you can manage to find it.

>> No.975356

I like scrolling through the screenshot/art section on blenders steam hub and seeing obviously ai generated images being passed off as stuff made in blender. what's more funny is their comment sections full of people falling for it.

>> No.975724

to make humanoid figures, sculpting or poly modeling for a newbie? not necessarily interested much in which one would be easier, but which one would benefit me more in the long term to git gud at.

>> No.975726

>>975724
>humanoid figures
for humanoids and organic characters in general people usually sculpt and then retopo, if you do low poly or anime characters box modeling could be faster maybe and it could help you with non organic models too.

>> No.975767

>>975724
polymodeling a humanoid is masochistic and impractical but it's a fun learning experience, here's one of the first courses I took when i was a beginner
https://www.youtube.com/watch?v=5eIxHQnLqr0&list=PLok698dKQ_Hj10eKx73qPJRktkB1r8SN5&index=1
Now that i'm more advanced I KNOW I will never do it again because sculpting is a lot more practical, but it still taught me tons of things for when i need polymodeling
It's a big undertaking though so my advice is going straight for sculpting a humanoid and choose a different, simpler project to learn polymodeling. In any case you will have to mess with polys during sculpting when you get to the retopo

>> No.975818

I remember seeing an addon on twitter that lets you select edge loops across triangles but I can't seem to find it anywhere. Does anybody know what it's called?

>> No.975899

>>975818
Could be MESHmachine's LSelect.

>> No.975907

Anons any simple tutorial for humanoid making? I'm starting in blender and I want to try making a human figure to create a Tulpa.

>> No.975921
File: 886 KB, 720x720, 1640960022281.webm [View same] [iqdb] [saucenao] [google]
975921

any addon that let you do webm-related in a plug-n-play manner instead of messing around with geo nodes?

>> No.975922
File: 150 KB, 2070x827, 1640960600490.jpg [View same] [iqdb] [saucenao] [google]
975922

>>975921
the geo-nodes

>> No.975973

So I'm trying to save different layers of a render as different files as a means of compositing them later. Certain layers loop every 500 frames, while others at like 250 or smaller, so I shouldn't need to render the entire scene for the full 500 frames if I can break it up into smaller chunks.
The problem is, when I render out to EXR, and try to composite them back, the result and colors aren't the same as if I had just rendered the full image.

Obviously some color space shit is at play. What should I be saving my EXRs at to make sure it plays nice when I composite later? Just regular sRGB with no transforms? What about when I import them? They default to Linear Rec 709.

>> No.975977

>>975973
Nevermind, I seemed to have solved it somehow. Colors aren't exactly 1:1, but it'll do.
For future reference in case anyone didn't know, EXRs dont get saved with view transforms or any other colorspace info. Kinda obvious in hindsight, but things get pretty fucky when you start messing with shit outside of typical formats.

>> No.976040

I'm great at UV unwrapping and texturing, I have in-depth knowledge and tons of experience manually unwrapping things. Now, that said, I wish Smart UV Project wasn't complete fucking garbage. I don't know what the fuck they're thinking with that. It's almost never usable at all, they think it's cool to make a billion little tiny squares along with shit that's all oblong and un-uniform and present it to the user as " here you go, it's the easy unwrap solution :D "

They need to do something with the algorithm. We need bigger uv islands (none of this 1 face tiny square bullshit), it needs to prioritize keeping them as square as possible (to fill up as much UV Map real estate as possible), and all of the uv islands need to be roughly the same size so the texture is uniform on the model.

I don't think this is impossible or unreasonable. You can go in there, turn your brain off, and just make squares with Mark UV Seam all over a model, Unwrap, and it'll be way better than Smart UV Project. You don't have to be a genius to get better results than this bullshit.

>> No.976068
File: 320 KB, 1080x1080, 1496711181293.jpg [View same] [iqdb] [saucenao] [google]
976068

>>975922
wait.

would it be possible to set geometry nodes up in a way that it could "shrinkwrap" a model around another model but dynamically "bridging" gaps?
easy example:
>shirt
>shirt bridging between boobas instead of tucking in between the breasts like normal shrinkwrap would do

would this be possible with geo nodes?

>> No.976073

>>976068
Geometry Nodes are just a poorly thought, poorly implemented programming language.
When you ask "is X possible in Geometry Nodes" what you're really asking is "is X possible with computers" but with the added complexity of having of having to do with something completely broken such as Geometry Nodes.

>> No.976074

You can mix geometry nodes with regular modifiers. So you could do something clothing / boob related with it and then right underneath that geometry node modifier add a Shrinkwrap modifier.

>> No.976077

>>976074
How exactly? Can you show how you do it with Geometry Nodes instead of just imagining it?

>> No.976078

>>976073
>>976074
well then I wish how to do it then. I have no clue how to make the shrinkwrap not get pulled into the dips

>> No.976080

>>976078
Cloth simulator.

>> No.976081

>>976080
I was looking for a geo node-based solution

>> No.976082

>>976081
It's called Houdini then.

>> No.976087

>>976078
>no clue how to make the shrinkwrap not get pulled into the dips
Just fucking weight paint it bro. Vertex weight the titty bridge area lower than the rest of the shirt so it's influenced by the SW less. It's not rocket science, like 90% of the modifiers' strengths can be adjusted with vertex weights.

>> No.976090

>>976087
that's why I said dynamic.
I do know how you can do it manually but I am specifically interested into what you can do with geo nodes and if this particular thing was possible then I already have several other ideas I want to base on this.

>> No.976095

>>976090
Shoulders guy, look at this:
https://graphics.pixar.com/library/StableElasticity/paper.pdf
it's what you were looking at a year ago when you were sure it wasn't a simulation but a secret complex network of helper bones, constraints and drivers.
Yes, it was a simulation. Yes, it's horribly complicated and no, in the end it doesn't even look that good.

>> No.976097

>>976090
Again, weight paints.
Set up an empty and use its proximity to weight the titty area lower. For that you can use Geo Nodes, or just a vertex weight proximity modifier. Distance to the empty controls the falloff.
Fucking around with cloth as is, even with geo nodes, is just trash. It's a leftover system from like 2004 that hasn't been updated since. Like a lot of Blender "features". Devs are more interested in adding, rather than fixing what's been there for decades.

>> No.976098

>>976097
Ok but can you show an example of your easy method instead of just talking about it?

>> No.976099

>>973621
DO NOT DOWNLOAD AND DEFINETELY DO NOT OPEN THAT ZIP

>> No.976100

>>976097
Also, if a "leftover system from like 2004" (as you say) is more useful the Geometry Nodes, what does that say about Geometry Nodes?