[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/3/ - 3DCG

[Advertise on 4chan]

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • There are 68 posters in this thread.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


Self-serve ads are available again! Check out our new advertising page here.


[Advertise on 4chan]


File: 1597567686720.jpg (103 KB, 1280x720)
103 KB
103 KB JPG
What's your most deep darkest workflow secret?
I'll start, i paint normals.
>>
I use blender only when I need to make curves
>>
>>774684

I just want to do 3D to make stuff that looks 2D
>>
File: 1582863333823.jpg (39 KB, 1008x561)
39 KB
39 KB JPG
i use advanced skeleton 5 to rig everything.
>>
>>774692
Are you actually Double?
>>
>>774692
how do you rig faces then?
>>
>>774684
Painting normals is a valid technique, at least in 2d. Though it isn't widely used.
I saw that one vid where that dude did it, and he was able to change the lighting in a painted scene and it was pretty neat. I'm almost certain the OP pic was from that vid actually.
I wanted to try it out on a 3d model to see if it would be viable, but I think it would just make things look really messy instead of giving it a painted look.

My deepest workflow secret though?
I guess that I steal car models from Forza. I can model most things well enough on my own, but cars are a deep time investment that isn't worth it to me t.b.h. Since at the end of the day (excluding skill) the end result would be the same as a car modeled by anyone else. I still have to spend a good deal of time fixing it up and everything to get it working, so it's not like I'm putting no work into it. But yeah I think modelling cars is a waste of time, for me compared to modelling out an environment and telling a story.
>you wouldn't download a car
:^)
>>
The key to amazing 3D skills is having sex
>>
>>774684
While working at Digital Dimension I got away with using a Poser figure (as base model) on the production of a pretty successful TV show. That was roughly 15 years ago.
>>
>>774709
Pretty sure they used Poser with the first Spider-Man movie, so you're good.
At least, that's what my drawing teacher thought. He said he could tell just by looking at the anatomy of it. Dude knew his shit, so I was inclined to believe him.
>>
>>774709
>>774714
At least it's not Daz.
>>
>>774709
you monster
>>
>>774700
no
>>774701
as5 has a tool to rig faces.
>>
>>774704
i should do this for guns
>>
File: 1516201416866.jpg (140 KB, 500x667)
140 KB
140 KB JPG
>>774684
painting normals is cosy as fuck
it's a much easier way to make normal maps for old games than try and clean up the low poly and make high poly models
>>
I work slow because I have the bad habit of making sure I understand what every parameter is for, even if I'm not going to use it in the task at hand. Constantly reviewing documentation and such.
>>
>>774684
I too manually paint normals and/or enhance rendered normals by painting.

My darkest workflow non-secret is I texture my models seamlessly with standard 2D photoshop by rendering my viewport to a highres texture
paint any number of layers ontop and project it back onto the UV's via a texture from camera material.

Why ~everybody else isn't doing the same is beyond me. I've been doing this since 2006.
More than once a fellow artist have walked past my station and gone all pale in the face when they see how I achieve what I'm doing
>>
>>774738
I mean with guns I think that's where I might draw the line. I think guns are a bit easier to model, since they're not mathematically exact organic shapes. It's all just variations of cubes and cylinders. So it's not that much of a time investment to get something that looks passable, while with a car it has to be pretty exact or it looks off. In either case you'll have turbo autists that'll point out the smallest flaw that's not general knowledge.

Then again, I don't really use guns in my work, so I guess you could just lift the guns from other places.
I don't really have a leg to stand on considering I'm a car thief.
>>
I used the same rig forever, and use Blender automatic weights for flawless results.
>>
I routinely steal parts of models and frankenstein them into my work.
>>
>>774687
Same, theres just something about 2D art styles that appeal to me, none of that anime outlines shit, stuff that looks truly 2D - like Disney's Paperman short
>>
I use uv master for all my uvs
>>
File: 1599408188363.png (366 KB, 528x528)
366 KB
366 KB PNG
>>774684
I paint height maps and generate normals from that
>>
I use Blender mostly one handed.
>>
Opened Maya one time but i cant do anything because im used to use Blender.
>>
>>774775
>I work slow
this, im not that employable because i work really slow, and i don't know why, its not like watch youtube all day, i just take to much time to do each step on the pipeline.
>>
>>774776
>by rendering my viewport to a highres texture
what? explain pls.
>>
>>774777
yeah you're probably right, guns are not that difficult, cars on the other hand are on the other extreme end. the reality is that im not that good with hard surface.
>>
File: 1600189879710.jpg (4 KB, 259x195)
4 KB
4 KB JPG
>>774857
>Blender automatic weights for flawless results.
really? i always end up using Perry Leijten skinning tools because maya is horrible at weight painting.
>>774872
dude, even for client work?, well assuming you have clients.
>>
File: 1588884434100.png (12 KB, 321x321)
12 KB
12 KB PNG
>>774888
>I paint height maps and generate normals from that
nice get, but what's wrong with you man.
>>
I take animations from mixamo
>send unrigged Model into mixamo for automatic rigging
>export Tpose
>import into blender
>redo the weights of the model and add extra bones etc
>get Whatever animation from Mixamo
>transfer animation data from the import to the rig I reworked

All that so I can properly use it in unreal
>>
>>774953
Brainlet spotted
>>
>>774888
Nothing wrong with that if properly done
>>
>>774949
Pretty much you can paint directly onto your model from any angle with any 2D editor by placing a camera in the scene, rendering your character or whatever from the angle you wanna texture, open that image in say photoshop, paint whatever you want onto the character and then go back into your 3D program and use texture baking to project this image onto your character from the camera.
>>
>>774955
>>774704
Based.
>>
>>774965
Why would you do this to yourself? Have you heard of substance painter?
>>
>>775456
I have anon, and before that I had heard of (and used) mari and mudbox too anon. Even deep paint 3D way back when.

What all of these have in-common is that they're not the best 2D editor on the planet while photoshop is.
>>
>>774684
I do that to clean up bakes all the time. When I don't want to go through exploding everything.
>>
>>774776
Sort of an unwrapped projection painting, you're saying?
>>
>>774965
Never mind, I see what crazy shit you're up to lol
>>
>>774888
Checked and do what works my man
>>
>>774707
Can confirm this is false
>>
File: maxresdefault.jpg (44 KB, 960x720)
44 KB
44 KB JPG
>>774684
I GoZ to 3DS max when I need to make a base mesh for something I'm working on in zBrush because zBrush's modeling tools are shallow and pedantic
>>
>>775695
dude, that's not a dark secret, or at least i think its a normal part in the pipeline, i keep jumping from zbrush to maya all the tiem.
>>
All of my textures are procedural because I dont want to pay or torrent for photoshop and gimp sucks ass. So often my nodes are 30+ nodes generated using different mesh data to get a desired result. I usually take 14 hours doing this btw. By now I have a library of uber shaders that I only used once.
>>
>>776176
you should share them if youre not gonna sell em
>>
File: 6289302147_38e8035680_b.jpg (84 KB, 768x1024)
84 KB
84 KB JPG
>>774866
Same here.
>>
>>776254
Ah, the Chinese method I see...
>>
>>776330
The method of lowskills
>>
>>776330
Bruh
https://youtu.be/zd-dqUuvLk4
>>
>>774888
Yeah I mean that's what I do. My clients always ask for normal maps so I just bake the bump :^)
>>
I use boolean on EVERYTHING.
>>
>>776445
you monster!
>>
I box model with a tablet. No keyboard.

[spoiler]I am a hobbyist.[/spoiler]
>>
im the developer of one of the most popular pieces of independent cg software (not blender), and yet i come to this stupid fucking board many times a day for literally no reason other than to laugh at blendlets
>>
>>776503
Post link or piss off, LARPer
>>
>>776503
unfold 3d?
>>
>>776539
cry more, blendlet.
>>
>>776541
Maya + Zbrush user, but okay. Post link or piss off, LARPer.
>>
>>776503
based
>>
>>776553
yeah right blendlet
>>
>>776558
Go fuck yourself you mongrel
>>
Instead of UV mapping I just hit uvmaster in zbrush and accept the results as is.
>>
>>776639
Go back to plebbit, tranny
>>
>>776620
no.
>>
File: 1576681654042.gif (773 KB, 333x358)
773 KB
773 KB GIF
>>776645
why are you so triggered?
>>
>>776503
Leave it to a "rent free" faggot to derail an otherwise good thread.
>>
>>776176
materials are the way to go
>>
File: d5b.png (21 KB, 400x400)
21 KB
21 KB PNG
>>774684

I paint heightmaps and then use the normal map plugin. My shame shall be forever buried here.
>>
>>774684
NOONE does that. Why? There are a billion algorithms for that.
>>
>>777946
Who the fuck is Noone and why does he paint his own normals?
>>
File: 1601097352804.png (470 KB, 750x730)
470 KB
470 KB PNG
>>777946
sometimes is faster to fix normals on photoshop then to rework shitty geo my man.
>>
>>777962
FYI to posters in denial, I have worked in a large game studio doing enviro art and fixing shit in photoshop is indeed faster sometimes especially during LOLCRUNCHTIME16HOURDAYSFUCKYOU.

The smudge tool works wonders. When you gitgud at it you can smudge away most minor normalmap errors in seconds. Freefags can even do this in Gimp. It's more effective than most kids in college would think. If you are in school your teacher will tell you not too because it's "bad habit" but the truth is everyone in the industry has done it once or twice anyone who says otherwise is a lying faggot.
>>
File: mmmm.jpg (26 KB, 350x263)
26 KB
26 KB JPG
>>774684
I stopped giving a single fuck about proper topology and edgeflow and all the goody two shoe modeller practices long ago
my wires are a mess now: ngons everywhere, poles, non-manifold, missing faces, double verts, double edges, everything you can think about - my models have them and by Lord I swear that I will go as far as tris subdivision if I have to
I only start fixing if shit obviously drops into eldritch dimension and it's noticeable, whatever happens outside can burn in faggot hell for all I care
>>
File: 1576356379916.jpg (138 KB, 792x844)
138 KB
138 KB JPG
>>778573
i tend to do this for my personal stuff, you know, fucking zremeshing everything, auto uvs and all, but when it comes to client stuff i go uber autismo cleaning up topology, pristine uvs, animation loops, etc, i think this is the proper way to go about posting shit to social media, fuck it, they dont need to know that my topology is retarded, they just need to see that im posting shit.
>>
I work at 70% of my actual speed and my boss thinks I am fast. I am REALLY fast, faster than anybody else there, but I am not telling or showing how fast I really are. I never do overtime. Sometimes I go earlier.
>>
>>778586
what is your fast retopology secret?
>>
>>778590
cocaine
>>
>>774684
>What's your most deep darkest workflow secret?
I only model parts of objects visible to camera from one specific angle if it's not for an animation
>>
>>774965
would you mind posting a couple of screenshots of this? i wanna try it
>>
>>778575

I strongly suspect most of the shit on artstation is exactly this. Hellscape chaos-topology that has been substance painted and post processed but will never be animated and only serves the purpose of "hire me pl0x".
>>
File: dabbingOnBlendlets.jpg (1.64 MB, 1920x3820)
1.64 MB
1.64 MB JPG
>>778626
Here you go anon
>>
>>778660
I use Mari for that. Gives me way more control.
But the basic principle is the same.
>>
>>778559
freefags should be using krita with the normal brush engine but yes
>>
>>774965
Sounds complicated. I thought I was insane for manually setting up texture painting layers in Blender and using that instead of Substance Painter.
>>
>>778660
this is super interesting, thanks anon!
>>
>>778660
we need more of these.
>>
>>776243
Yeah I might as well. I guess I could make my own website and just churn out textures like a madman.
>>
>>774709
is that not how it's supposed to be done?
t. Brainlet
>>
>>774684
in blender when I apply a texture I rotate the texture to fit it because I don't know any other way
>>
>>778660
if /3/ bros can keep helping eachother out like this then we're all gonna make it
>>
>>778660
This is so weird. Doesn't 3Ds Max have proper texture painting?
>>
I learned 3d so i could simulate shit breaking up into pieces. I'm actually a 2d animator. I then take those, rotoscope it FBF and add effects/other objects/details. Everyone thinks i do all this shit purely by hand, i dont have the heart to tell them its all simulations...
>>
>>778861
No rules, just tools.
>>
File: 1584579370895.jpg (7 KB, 149x165)
7 KB
7 KB JPG
i use mixamo for my animations, some times i download the files, others i just copy the poses and edit to fit the character.
>>
>>779014
>Mixamo
>my animations
pick one.
>>
I use white noise to randomly generate worlds.
>>
>>774684

Krita has a brush set for painting normals it's not as unusual as you may think
>>
>>779069
Why not perlin noise?
>>
i've used roadkill for uv unwrapping and crazybump for bumpmaps since about 2014

i've also only recently updated from max 2016
>>
>>774684
I use photoshop's automatic normal converter to convert my diffuse textures into a normal map

how bad is that? I legit don't know
and what should I be doing instead if its that bad?
>>
>>774684
I've been here for years trying to learn to model anime, and I've only gotten better at editing other people's models. I started putting serious effort into it last week, and I've started studying other artists topo, but it feels like it's not enough
>>
>>779388
You wont be getting the proper height information. Everything is probably all on one plane.
>>
File: dont do this.png (199 KB, 512x256)
199 KB
199 KB PNG
>>779388
It's not optimal...to say the least.
You get a lot of wrong detail into the normal map which does not correctly match what's in the diffuse map.
>>
>>779388
>how bad is that? I legit don't know
Its unironically THE WORST you could do to get a normal map.
>and what should I be doing instead if its that bad?
RTFM and understand what normal maps are and how to get a proper one.
Also don't expect others to deliver you all the information on a silver plate - show some initiative to learn for yourself.
I swear to god - the internet makes people retarded. You have all the available knowledge in the world one google search away, but you are still this uninformed? Shame on you.
Also this is not the question thread.
>>
>>779388
For microbumps on a millimeter scale it's perfectly fine but you need to know what you're doing for clean results. Such converters essentially treat the image as an OG grayscale bumpmap that treat the black/white point of your image as height data, and the angle of the slope is determined by comparing to neighboring pixels.

They can't generate accurate curvature for large geometric features the way a normalmap baked from geometry can.

If you want such features on a texture map you need to paint in the normals yourself like in ops image.

For fine surface features one can get very good results if you play around with the highpass filter and emboss and combine the channels of a vertical and a horizontal pass into a normalmap you then normalize.
>>
>>779406
A normalmap contain no height information what so ever. It stores the angle of the surface at each pixel.
The 'normal' in the name comes from how it stores a 'normal vector' in each pixel which is a direction
given in XYZ coordinates that tells you what direction the surface is oriented in at that point.

The X direction is stored in the red channel with 000/black meaning light is reflected to the left and 255/white means right and 127/gray
means it points neutral towards you in the horizontal plane. Y direction is stored and green channel and does the same but for up/down.
the Z direction is stored in the blue channel and tells you how much the vector points out of the monitor and towards you.
between these 3 values you get the vector/direction of where the light hitting that pixel will reflect.

The reason it has that blueish color if the surface is flat is because it points straight at you with gray in both the red and green channel
and full white in the blue channel RGB(127,127,255) which translates into XYZ(0,0,1).

RGB(255,127,255) = XYZ (1,0,1)
RGB(000,127,000) = XYZ (-1,0,0) - notice how black in the blue channel yields a Z coordinate of '0' and not '-1'
since having a surface pointing away from us would be meaningless for shading purposes.

If you understand the above you will understand how to go about making normal maps and how you may edit them manually to suit your needs.
>>
>>774888
lol at my first job this is what the art director got me to to for scifi paneling
>>
>>778630
heyyy thats what i do
>>
>>779479
That's totally fine for SciFi paneling. I do that too.
>>
>>779446
You are taking technically. I'm talking visually. The most common complaint of the nvidia filter is everything is the same height visually. They would be better off using ndo if they wanted fine control like you are talking about.
>>
File: maxresdefault.jpg (101 KB, 1280x720)
101 KB
101 KB JPG
>>779407
Heightmaps my dude

Convert image to grayscale, use contrast to reduce any details that you don't want appearing in normals, then lighten/darken depending on which direction you want the normals to go.
>>
>>779602
Nah that's a shitty way, doesn't work well for large features. You simply can't automate it well(with machine learning maybe), it will always require manual work.
>>
>>774684
I can't make my own rigs for shit in any software outside of very basic ones.
>>
File: 1573991234992.jpg (6 KB, 230x219)
6 KB
6 KB JPG
When im polypainting in zbrush i send my high poly model to 3d coat to add AO and curvature maps to help me with stylized style painting.
>>
>>779388
You should be baking the normal map from a high poly mesh. If you do convert an image to a normal map, the source texture should be a grayscale heightmap, not just your diffuse texture.
>>
I paint bump maps in Krita, then port them to blender as a bump node, then create an image, link it to a node, then bake a normal map to that image. Works everytime, 10/10, although some people that i've told this to criticize my methods
>>
cool, didnt know i can actually NOT be anonymous here lmao
>>
>>779446
That's not really correct. The Z value is keeps it all normalized, and it preserves the derivative of the slope of the surface.
>>
Nodes and codes for everything. Node groups to automate all my shaders, Python scripts to auutomate all the tedious stuff (mostly rigging). If you design everything to be reusable, you spend almost no time actually doing 3D art! ...... and almost all of your time making and fixing and updating your tools.
>>
>>780898
I left normalization out because it's enough information for someone to digest without that step.

There's however nothing special about the Z value anon, you can paint a value into each channel manually and end up with a vector that is longer than unit length.
but if you now normalize that map all channels will be altered to store the normalized vector preserving the same direction you set, not just the blue channel.

It's not uncommon to throw away the Z value all together and store a normalmap in just 2 channels and treat them as angular values of a hemisphere.
You don't actually need the blue channel to derive the normal, it just saves you some computation to have it prenormalized.
(Everybody tend normalize the normalmap in the shadercode anyways since a vector stored in colorspace doesn't add up to exactly unitlenght regardless.)
>>
>>780906
You're smarter than you let on, no fair!
>>
>>774687
Damn, same. I've toyed with the idea about billboarding with frames instead, but that shit is impractical as fuck given how time consuming all those thousands of frames will take to draw.
>>774704
>forza cars
For commercial purposes? If so, do you change the topology or something to make it hard to detect?
>>
I pay $500/yr for Nuke Indie
>>
I import non UV mapped hi poly models into substance and take my chances with whatever the auto unwrap churns out
>>
>>781547
>For commercial purposes?
Yeah? I'm not selling the models or upload them anywhere. I don't really change the topology though. Odds are, if there is a maker's mark somewhere on the model, it's somewhere hidden from normal view. So that if someone were to distribute the model, the ip holder would be able to look for it and find it to know that it's theirs. Like fake islands on maps.

For shit like renders and animations (which is what I do), that's basically impossible to find if it exists at all. Plus there's the whole shtick that Forza doesn't technically own those models, since they're licensing the cars from whichever company owns them. There's a whole legal grey-zone with it.

Lastly, like I said in the other post. A car is a car. All things being equal in terms of skill, if I modeled it or someone else did, at the end of the day a model will look the same as any other. If Jim modeled a Ferrari and Jack modeled a Ferrari, they're both Ferraris that look the same since they're both modeled after the same car.
>>
I used to be really good at blender. I stopped using it about 5 years ago. Now i'm starting all over and I dont know anything anymore and my progress is so slow I want to kill myself
>>
>>781836
I was in that boat once, and ended up bouncing back even better. You'll find that it all comes back pretty quickly. Just concentrate on the things you want to do, rather than trying to go back over everything. If you get stuck, just look it up.
>>
I promised anon I'd learn python for scripting a few months ago, but I'm learning a ton of 3D things and they keep leading to other things.
>>
>>781850
>but I'm learning a ton of 3D things and they keep leading to other things.
This is the Way.
>>
>>781850
lol, i started learning cinema 4d because i wanted to make my motion graphics more exiting, im now doing characters and animations for games.
good luck with your journey anon, make sure to at least try what catches your interest, you don't know what you may end up loving.
>>
>>774707
Less nudes more nodes
>>
>>781827
Makes sense. Pretty interesting.
>>
>>774687
Do you have any examples of your work?
>>
>>782155
A lot of it is laziness on my part, since I'd rather make a scene than spend 3 weeks modelling a car. Not to mention a lot of my clients are individuals, so making them pay a fuckton of money for me to model a car and animate a scene is just unrealistic.

It's nice to say, "yeah this job will cost $500", and get paid the actual cost of the work, but it's gonna scare off the only people regularly coming for shit like album art. Not all clients are big-fish, and they don't come often enough for me to rely on them.
>>
File: 1602347364017.jpg (33 KB, 960x960)
33 KB
33 KB JPG
Coom powers me and yet I do not find my own work sexy.
>>
>>774684
im so good at 3d modelling in sketchup and rendering with shite softwares that I can charge $500 for a set of 5 renders I did in about a day





Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.