[ 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: 2.86 MB, 3840x2798, Wipkiyo-E.jpg [View same] [iqdb] [saucenao] [google]
901909 No.901909 [Reply] [Original]

/wip/ - Works in Progress
- Moonrunes Edition -

Post your work-in-progress projects, recently finished projects, or things you'd like critiqued here.

Previous thread: >>899902

List of free resources: https://pastebin.com/cZLVnNtB
/3/ Discord for those interested: https://discord.gg/ujt5vtr4DE

>> No.901912
File: 1.30 MB, 1200x1200, happyrobot5.png [View same] [iqdb] [saucenao] [google]
901912

now he can smile digitally yay
next up is figuring out how to shapekey the expressions now that theyre a texture

>> No.901914
File: 909 KB, 800x1080, 1642527581514.png [View same] [iqdb] [saucenao] [google]
901914

>>901909
2nd venture into psx style after the cat train thing. theme is mundane/comfy with an added touch of magical realism. should i add a low poly animu girl drinking coffee to get that mundane part in? I don't have any character sheet in mind so open to that as well.

>> No.901915
File: 35 KB, 450x253, 1644217393918.jpg [View same] [iqdb] [saucenao] [google]
901915

>>901914
le reference style

>> No.901916

>>901914
Really like the sky background

>> No.901917
File: 367 KB, 1024x1024, 1636146148004.png [View same] [iqdb] [saucenao] [google]
901917

>>901916
here you go, used it from the previous one. i think i just downloaded a sky and then just scaled it down to 1024x1024 and used the posterize effect

>> No.901918

>>901907
It could look better if you control them with the AO maybe.

>> No.901923
File: 300 KB, 1380x1880, WireFrame.jpg [View same] [iqdb] [saucenao] [google]
901923

Spent the last couple of weeks on retopology. Was a little harder than I expected. At the end of it I wasn't happy with the result so I started from scratch.
This is what I have now. Next step is to rig this thing and see how baby deforms.

>> No.901933

>>901918
Control what about them?

>> No.901934
File: 660 KB, 891x685, qt314.png [View same] [iqdb] [saucenao] [google]
901934

How do you get yourself out of the infinite WIP loop? I've been sitting on this inkling character for probably 6 months now, basic model was finished months ago and then i kept revising it as i kept noticing more areas where i went completely wrong. Now it SHOULD be done, and i'm on weight painting/making shapekeys/texturing, but i just can't bring myself to pick it up again.
My biggest concern right now is learning how to paint a proper body texture, i have no clue how texture painting really works and i can't find youtube tutorials that explain it well. its like my brain processes the information wrong or somethin'.

>> No.901941

>>901933
I suppose you can have the cracks just in the ambient occlusion and it could look better maybe

>> No.901943

>>901934
>How do you get yourself out of the infinite WIP loop? I've been sitting on this inkling character for probably 6 months now,
not stupid questions thread, idiot.

Release stuff on artstation and track your statistics and likes. Try to get top row.

>> No.901945

>>901941
Oh I confused AO with depth of field for some reason. Yeah I guess it could look cool if it used ambient occlusion, but for that I guess I'd need the cracks to be part of the actual mesh and not just a bump map, and there's no way my laptop could survive with that much geometry

>> No.901947
File: 414 KB, 1416x824, SCP TIMES UP 01.png [View same] [iqdb] [saucenao] [google]
901947

>>901909
TICK-TOCK,TICK-TOCK...

>> No.901950
File: 2.65 MB, 1920x1080, 1642425725819.png [View same] [iqdb] [saucenao] [google]
901950

Does this couch look comfy?

>> No.901959

>>901909
Is this a style transfer/AI thingy, or are you remaking the WIP scene yourself in this style?

>> No.901960
File: 1.43 MB, 1920x1080, room v3 (full res).png [View same] [iqdb] [saucenao] [google]
901960

>>901909
Did some more work on the CAST codebase and optimized its memory consumption to the limit, as well as managed to get it to run on the CPU, and managed to find the option to disable preprocessing. Now the aspect ration is maintained and no scaling is done. Here is the room in full glory. This image's resolution is 1920 x 1080, but I can get it to process 2048 x 2048 without issue now.

My next goal will be to try to denoise the image and see if the noise is causing the patchy look in the style transfer. Apart from that I've read in the AdaIn paper that it is possible to control the stylization strength. Since CAST uses AdaIn at runtime I am going to see if I can get it to work. Right now it preserves the original content a bit too strongly for my taste, I wouldn't mind if it was more stylized.

>> No.901961

>>901950
looks boneless, I'm expecting to fall on the ground as all the air moves away when I sit down
i like the gold studio backdrop though

>> No.901962
File: 113 KB, 1920x1080, test.jpg [View same] [iqdb] [saucenao] [google]
901962

Got banned for a few days so I put some of that time to use on starting something new and put the "bit off more than I can chew" underwater thing on the backburner for a bit .
All of those textures are just placeholders. Same as the characters.

>> No.901964
File: 564 KB, 1421x813, SCP TIMES UP 02.png [View same] [iqdb] [saucenao] [google]
901964

>>901947

>> No.901973
File: 156 KB, 1509x1131, Capture.jpg [View same] [iqdb] [saucenao] [google]
901973

https://www.youtube.com/watch?v=a1__LToamSI

>> No.901974
File: 268 KB, 994x572, unknown.png [View same] [iqdb] [saucenao] [google]
901974

I think I finished the modeling phase of the LADA. Going to work on the texturing later

>> No.901976
File: 313 KB, 1058x599, unknown2.png [View same] [iqdb] [saucenao] [google]
901976

>> No.901977
File: 3.24 MB, 1920x1080, 1623916819215.png [View same] [iqdb] [saucenao] [google]
901977

>>901950
Now I need a carpet and something to go to the right of the couch. A bookshelf maybe? Also I don't know if I want a plain white/fluffy carpet or something more modern.

>>901961
Imagine it's filled with feathers or something.
>i like the gold studio backdrop though
Thanks, I also made a wood variation which I think looks a bit better.

>> No.901982
File: 510 KB, 2880x1800, Screen Shot 2022-06-07 at 3.00.01 PM.jpg [View same] [iqdb] [saucenao] [google]
901982

>>901909
how do I make this look more pro? i need crit and help

>> No.901989
File: 281 KB, 2184x2394, 1636356167257.png [View same] [iqdb] [saucenao] [google]
901989

Haven't done any serious 3D modeling for game dev in maybe 5 years, so I'm trying to get back into a consistent workflow. Modeling this took me maybe 4 hours, and I've spent the last 2 unwrapping everything but the main TV model itself. Trying to decide how to fix up these ngons

>> No.901992
File: 209 KB, 695x811, SCP TIMES UP 03.png [View same] [iqdb] [saucenao] [google]
901992

>>901964

>> No.901995

>>901982
compare it with artstation/sketchfab models and take note of what you're missing

>> No.901999
File: 1002 KB, 5120x1080, Scn 19.jpg [View same] [iqdb] [saucenao] [google]
901999

>>901974
>>901976

This is lovely anon show us the wire man.

>> No.902001

>>901992
I love this actually. Usually alot of artists are boring or samey, so it's nice to see something strange.

>> No.902004
File: 340 KB, 1024x512, mechmodel2.png [View same] [iqdb] [saucenao] [google]
902004

Making a mech to 3D print based on AI-generated concept art.

>> No.902005
File: 253 KB, 512x512, reverse-joint_white_ai.png [View same] [iqdb] [saucenao] [google]
902005

>>902004
And here's the art I'm going off of. I chose the left shoulder, and the right arm since those seemed the most well defined.

>> No.902007

>>901959
I tried a few style transfer stuff, but it all looked like shit so I said fuck it and painted it. It's kind of shit and could be better, there's a bunch of iffy parts with the fills and I'm not too pleased with the colors, but at least (I hope) it's better than a crappy style transfer that'd look like a mushy mess.
I would have done a collage, but I was away from home and didn't have access to the files I usually use for it.

I'm not the best at drawing, but I think it might be fun to try and do something like this a few more times. Painting the wip scene in a few more styles. Takes about the same time as the collages and it's something new.

>> No.902008

>>901982
it comes down to art fundamentals. getting there unfortunately is an arduous journey. The good news is that knowing some fundamentals goes a long way towards improving your art.
Might I suggest you start with these books:
>Figure Drawing - Michael Hampton
>Figure Drawing - Steve Houston
The first is my favourite but check out both. you can find them on:
>b-ok.org
Obviously these books were intended for people learning to draw figures but you can apply the knowledge to sculpting and modeling as well. Good luck.

>> No.902010

>>901747
are you the dude that made that cat character? Are these characters for a game? L

>> No.902013
File: 1.00 MB, 1080x1920, Illustration-2.png [View same] [iqdb] [saucenao] [google]
902013

Playing aroung with xgen and some mixamo.

>> No.902015

>>902004
Cool start, looking foward for the legs

>> No.902021
File: 413 KB, 1572x1168, Screenshot from 2022-06-07 20-02-35.png [View same] [iqdb] [saucenao] [google]
902021

i'm cooming

>> No.902022

>>902021
Minus 8 if he real.
I dig it. Good anatomy and proportions.

>> No.902023

>>902022
now that you mention it he kind of does look like him.
i'm going for an Egyptian pharaon.

>> No.902024
File: 108 KB, 474x669, mech_unbalanced.png [View same] [iqdb] [saucenao] [google]
902024

>>902015
Working on the legs now. Trying to figure out how that sort of staggered design works. So far, I'm getting some awkward looking results. Pic rel. Looks like it's going to tip backwards. I know the stance isn't the same as the ref pic, but something about my take on the legs feels fundamentally off, and I don't know what it is.

>> No.902025
File: 517 KB, 1041x884, mech_unbalanced_2.png.png [View same] [iqdb] [saucenao] [google]
902025

>>902024
another angle. Going to start over on the legs tomorrow. Maybe I've just put too much time into it today.

>> No.902026
File: 189 KB, 1139x746, 1654653376770.jpg [View same] [iqdb] [saucenao] [google]
902026

if i want to animate this should i make the head straight?

>> No.902027

>>902026
Generally speaking, it's best to model something closest to the position it will most frequently be in. If the dragons head will most frequently be up, then that's how I would model him.

>> No.902028

>>902027
ah alright, thank you

>> No.902029
File: 19 KB, 720x630, latest[1].gif [View same] [iqdb] [saucenao] [google]
902029

>>902001
Pretty sure that's a Slogra.

>> No.902031

>>902028
I'm not an expert, though, so just take my advice with a grain of salt. You might want to wait for some other equally unqualified anons to weigh in before you commit to anything too major.

>> No.902032
File: 660 KB, 659x814, blender_ZhNcfmOQFx.png [View same] [iqdb] [saucenao] [google]
902032

Is it okay for me to just leave these hair strands as is for the low poly bake or would it be worth it to remesh it and have things in a connected mesh? I really want this thing done, but I don't wanna have anything look ugly by the end.

>> No.902034
File: 213 KB, 805x501, 1654656658419.jpg [View same] [iqdb] [saucenao] [google]
902034

How do i make the eyeball...

>> No.902039

>>902032
Keeping them separate might be better if you want to pose them.

>> No.902041

>>902034
My guess is that it's done pretty similar to a human eye. You select the edges of the eyehole, then extrude and turn it in about 90 degrees inward, effectively creating a rim for the lids. Then you select the rim, and move along of the eyeball. The snap function can help with that.
Look up a tutorial to get an idea of what I mean.

>> No.902069
File: 866 KB, 2560x1400, 1626014221825.png [View same] [iqdb] [saucenao] [google]
902069

>>901989
UVW unwrapping my behated

>> No.902070
File: 704 KB, 1345x866, 1654667499132.jpg [View same] [iqdb] [saucenao] [google]
902070

>>902041
that take a lot of time

>> No.902072
File: 1.56 MB, 1280x741, drogon eye topology.png [View same] [iqdb] [saucenao] [google]
902072

>>902070
I can faintly see what I think is a rim, so I guess you got it. But I would make the rim more obvious. "Deeper" I'm pretty new to modelling myself, so take what I say with a grain of salt. But you want to look up some references. Find a cool looking reptile eye. And copy that. Pull some of the edges out, and then the rim comes back to the eye.

>> No.902074
File: 342 KB, 865x858, file.png [View same] [iqdb] [saucenao] [google]
902074

here's an Amy I might never finish.

>>902010
Legit surprised you made that connection.
The cat was for something else, it's just trash now but this and the bunny I posted earlier is for a game. I do a bunch of other stuff idk.

>>902032
good stuff.

>> No.902075

>>902072
noted, at least i get some tips, better than nothing

>> No.902077

>>902013
Cute.

>> No.902078
File: 647 KB, 900x836, Slim B 07.png [View same] [iqdb] [saucenao] [google]
902078

Feeling like an idiot, because for the last few days, I've been working on this shoulder. Trying to make it so the chest muscle connects to the shoulder properly. I see it in all the references and it looks like a good way to ensure good arm articulation when it comes time to pose/animate. But in doing so, I completely broke the topology I had there already, and it was rough trying to put it back together in clean quads.

I finally got the shoulder area stitched back up. I'm not sure if it's the optimal topology. But it will do for now. I really need to learn armature to put the topology to the test. Right now, it's all a guessing game.

>> No.902083

>>902074
Ken Penders/10

>> No.902086

>>902007
With the deep dream generator you need to lower the style weight in the settings to get it to look better. CAST on the other hand preserves the original look very well, but you have to clone the repo, download the model, and apply the changes in the resolution issue by hand before you can use it. You'd probably need to install PyTorch as well...as well as Python and VS Code, and apply the 'requirements.txt'. It is easy if you know some programming, but not so much otherwise.

I guess being able to do these kinds of things is going to be my strength as an artist.

>> No.902089

>>902069
this seems very high poly for a game asset

>> No.902091
File: 132 KB, 640x480, SCP TIMES UP 04.png [View same] [iqdb] [saucenao] [google]
902091

>>901992

>> No.902093
File: 737 KB, 2060x1051, 1642821634776.png [View same] [iqdb] [saucenao] [google]
902093

>>902089
This is the high poly mesh I will use to bake normals for the lower poly ingame model

>> No.902095

>>902093
you texture the high poly before baking?

>> No.902096

>>902095
not normally, but I'd like to have a few different LOD models this time just to keep for my portfolio

>> No.902097

>>902096
cool

>> No.902098

huh, guess all this time i ignore the scale when modeling

>> No.902102
File: 1.77 MB, 856x480, SCP TIMESUP.webm [View same] [iqdb] [saucenao] [google]
902102

>>902091

>> No.902105

>>902025
Maybe try and gather more refereneces if you feel stuck with this one. Also play with the leg inclination an size, plus add the feet, even if its just a tapered box it might look more balanced.

>> No.902109
File: 596 KB, 800x1080, 1653107859494.png [View same] [iqdb] [saucenao] [google]
902109

>>901909
i think this is my first project i genuinely enjoyed doing, gonna animate the whales and be done with it. shame that these kinda side projects can be put in portfolio's but i hope you guys get a comfy feeling from this.

>> No.902120
File: 1.37 MB, 1920x1080, room v4 denoised (great wave off kanagawa).png [View same] [iqdb] [saucenao] [google]
902120

>>901960
Did a 15m render and denoised it. I won't post the noisy version as it it is not that different from this. I thought the effect would be more pronounced. What I'll investigate next is if it possible to go more in the stylization direction with CAST.

Style: https://uploads3.wikiart.org/00129/images/katsushika-hokusai/the-great-wave-off-kanagawa.jpg!HD.jpg

>> No.902121
File: 976 KB, 887x782, unknown (1).png [View same] [iqdb] [saucenao] [google]
902121

>>901909
After consulting with some friends that actually have a job in digital art, i got bullied enough to finally replace the mannequin head with a face.
Its a weird blend from a very old Mesh that originated somewhere in DAZ and gto proted to blender. After mangling it a whole bunch and redoing half the horrid topology and stenciling a rather low res texture together out of AI generated Faces and using Hair Textures from open game art it looks like a face.

Gimme your harsh critique on that Visage.
And this marks a break where i can not claim 100% my work anymore. But CC0 asstets don't count right?

Also Changed the Glass dome to be a fishbowl to get more retro vibes + some internal lighting.

>> No.902130
File: 1.42 MB, 1920x1080, room v4 denoised (great wave off kanagawa, 1.5 style weight).png [View same] [iqdb] [saucenao] [google]
902130

>>902120
Investigated it, and it is possible, but not like with DDG. It feels like the net is basically a single agent that is skilled at imitation, but still has a single style underneath it all. There is not too much variety in the nets approach to the problem. This is not too bad, as humans in general have only a single style. I am not disappointed in it the least.

It is just that I am curious what it would be capable if I had 10-1000x better hardware. I hope the AI chip wave won't take too long before it arrives.

>> No.902135

>>901914
I'm a software dev trying to get into 3d modeling. I have been following your refence for a while and cannot understand how he creates his textures.

Sorry, im retarded when it comes to this stuff

>> No.902137
File: 550 KB, 1440x874, 1654695913606.jpg [View same] [iqdb] [saucenao] [google]
902137

>Have to use different polygon to make the wing
hope it won't cause a problem when rigging it

>> No.902139
File: 802 KB, 1920x1080, 1637764776745.png [View same] [iqdb] [saucenao] [google]
902139

>>902135
even i got inspired by picrel and and had some anons explain it to me,it was made by anon who occasionally uses this thread and props to him. i am not exactly sure how the gamedev guy gets his textures but for ps1 there were 2 ways to create texture one is index colors and vertex colors. index colors is the one i mainly use which can be done in photoshop. go to image->mode-> indexed. indexed colors essestionally reduces the number of colors in your texture and introduces the noise kinda thing. vertex colors is used to fake lighting and i haven't really used it yet. hopefully my explanation isn't super ass

>> No.902143
File: 44 KB, 898x697, 1635539680765.png [View same] [iqdb] [saucenao] [google]
902143

>>902139
left is indexed and the right is the original texture. also forgot to mention scale your texture down to 128x128 or 256x256

>> No.902145

>>902135
>>902139
The term you're looking for is dithering. I'd recommend using an ordered dithering method (also called bayer dithering) to get an even more old-school style.
Here's a nice article about it if you're interested in the implementation details.
https://surma.dev/things/ditherpunk/

>> No.902149

>>902145
reminded me to play obra dinn but goddamn does it make my head hurt

>> No.902150

>>902139
And to think this shot would be worth weeks of work (if not months depending on reqs) if it had to be done in a photorealistic style.

>> No.902151

>>901962
you should be banned for promoting communism and judaism also.

>> No.902154
File: 2.27 MB, 1920x1080, 1647459864759.png [View same] [iqdb] [saucenao] [google]
902154

>>902150
not exactly, i threw this project together in less than 2 days, because easy to make models and getting a photorealistic look is a matter of plugging in textures and using an hdri. its kinda insanely simply and i am 100% sure a professional could have made the same exact scene in less than 10 mins and have more details

>> No.902162

>>902154
Would you say that the 2 biggest elements are scaled down textures and indexed colors? Am I missing more techniques that go into something like this?

>> No.902171

>>902162
https://www.youtube.com/watch?v=Gf5KP5ugLrA&t=34s
the bigggest thing would be pixelating and changing the filter size. this practically takes care of 75% of the work. rest is textures and your imagination.

>> No.902173
File: 307 KB, 928x491, 1630286655540.png [View same] [iqdb] [saucenao] [google]
902173

>>901909
can you guys help me with how in the fuck i can model the stormveil tower looking platform thing?

>> No.902174

>>902171
Thanks for all the info. Do people usually get pre-existing base textures online/photography or are allot of people hand painting them?

>> No.902176

>>902174
https://www.textures.com/library
this is where 98% of my textures come from.

>> No.902180
File: 433 KB, 1102x1065, file.png [View same] [iqdb] [saucenao] [google]
902180

so much cool stuff in here!
making a simple anthro wolf model to use in a game prototype i want to make
>>901912
so fucking cute

>> No.902181

>>902078
Why are you doing major shaping at the same time as topology? Usually, I'll construct, sculpt, join, then reshape, and only when I'm happy with all the shapes, proportions, and anatomy, then I'll throw down my topology.

>> No.902182

>>902151
I'm not promoting neither you silly sausage, it's a human sacrifice. Ironically I normally do Nazi related scenes.
The idea is to have the scene look the way it would while it was in use, and then redo it after liberation where it's suffered war damage and the like.

>> No.902184

>>901962
What did you get banned for?

>> No.902188

>>902184
"Racism" but it wasn't on this board. And it wasn't even racism. I was just pointing out the flaws in another anon's grandstanding by changing X race to Y race. For some reason they didn't get banned, but I did. Likely because they were the ones who reported it.

>> No.902193

>>902121

... is this the 50/60's like (pulp) spacesuit you posted some time ago? when yes, please post more of it, the whole suit, etc, it looks realy great - when you want critiqe on the face of the chick, post just her face/head, but from what i can see, it looks rly cool!

keep on!

>> No.902227
File: 384 KB, 1065x621, HOTWHEELSBEATTHAT.png [View same] [iqdb] [saucenao] [google]
902227

I'm very proud of this tire texture I made considering that I did it in Paint.Net

>> No.902230
File: 361 KB, 2023x1416, SomeFurtherDevelopments.jpg [View same] [iqdb] [saucenao] [google]
902230

>>902193
My Apologies for just posting subsections.
And the last one, with the mostly finished /k/ inspired bag was lost between threads.
If anyone has better patch suggestions let me know.

To the face >>902121
There is not much too it as i did not bother with anything that is covered by other parts, so its just a face plate with 2 clipped hair strands.

Overall i'm currently less sure about the overall design. Swapped out some buckles to look less modern. But May do some more drastic changes shortly. Everything looks to modern / not consistent.
Any suggestions let me know.
Any critique welcome.

>> No.902231
File: 697 KB, 1738x740, mechmodel3.png [View same] [iqdb] [saucenao] [google]
902231

I redid my mech legs. This design feels much more balanced to me. I think I'm happy with these proportions.
>>902227
Hand-made textures are soul. I remember seeing so many tutorials on the internet for it in the early/mid 2000s.
>>902105
Adding feet was very good advice. Just made things work. Thank you. I also made the angle of the upper and lower legs different, and I think that was the other key.

>> No.902234
File: 187 KB, 523x378, serva_pan.jpg [View same] [iqdb] [saucenao] [google]
902234

>>902231
Thank you, I love the low poly style you see in PS2 games, so that's what I'm trying to recreate

>> No.902235

>>902230
>Overall i'm currently less sure about the overall design

...having an idea, and developing it into a production ready asset are ..two different things.

dont spend to much time on perfect mesh, and all the other shit, develop the idea, crude, but readable for yourself and other who are not <100, so when you show it someone that he gets the idea, and once you found something good, invest time!

idea > perfecte mesh, perfect anatomy!

>> No.902240

>>902235
"Perfect mesh" I got some optimized ones in, some kit bashing bits recycled form my older projects. But there is also some real topology gore in there.
Currently sitting at 720k is vertices. And about 3x that at render when all the subdivisions go to their render settings.
Goals are just a simple rig in the end to throw a quick pose for a turntable and some detail shots for a render portfolio.
My computer can handle about 2x my current detail with manageable lag so i'm still good on that side.
But i try my best to keep small parts as sub projects that could stand on their own so i have in the worst case something i could salvage for the next project. To better work with my schedule of random productivity bursts every other day of the week when i got some free time.
Getting done is technically priority number 1, but at this point looking at way too much references as procrastination i feel like i need some significant design changes.

>> No.902249

>>902086
Eh, even then it usually ends up a mushy mess of pixels that doesn't really match the reference all that well. I've seen some impressive things to be sure, but I don't think AI is quite there yet.
Besides, I already did the AI thing already with some threads, might have seemed like a cop out if I did it again.

>> No.902255
File: 1.48 MB, 1920x1080, untitled.png [View same] [iqdb] [saucenao] [google]
902255

Wasting my time..

>> No.902257

>>902181
My concern isn't necessarily if I can sculpt a thing or not(I'm fairly confident I can), but how efficient is the topology, and how well it will eventually deform. So my thinking is that by learning to box model, I will have greater control over such things. I shape as I go for funsies, and because it helps be decide where vertices should go.
Also, my first tutorial was how to create a base human mesh using box modelling. So I was introduced to blender through topology autism.

Also, also, I have to trick my brain to do work that it doesn't want to do. If I just sculpt the character first, then I'll sit there and stare at her boobs, too satisfied with the results. Too daunted by the idea of retopologizing it all. By doing the topology first, I get that work done, and shaping feels like a little break or reward. Little by little, it will come together.

>> No.902270
File: 245 KB, 500x500, ntfbottest.webm [View same] [iqdb] [saucenao] [google]
902270

>>902255
Same!

>> No.902272
File: 3.60 MB, 1920x1080, 1633799761195.png [View same] [iqdb] [saucenao] [google]
902272

>>901977
Made a carpet. The material would benefit from having a bit of noise and less uniform strands but that's good enough from now.

>> No.902277

>>902272
Looks very comfy. The couch seems a bit too smooth. Maybe some worn in dent where someone always sits or something.

>> No.902278

>>902272
The couch just isn't convincing as a real couch. If you're going for some abstract artistic thing, then cool. But if you're trying to pass that off as an actual couch people can sit on, then No. It needs more structure.

>> No.902285

>>902277
>>902278
The goal was to make it look really soft, the type of couch you really sink into and can't get out of. I also didn't want any part of it to look rigid, like it's just a bunch of bean bags pressed together instead of an external rigid frame that holds cushions like a normal couch. I agree that it's too smooth, especially the backrest. But since it's made with cloth sims I'm going to deform it using a character as a collider, I think it'll look better once I'm done with that.

>> No.902286

>>902285
I'm trying to imagine it as a beanbag thing, or stuffed with feathers as you mentioned previously, but I just can't. It reminds me of a badly inflated bed, that if I attempted to sit on, my ass would smack to the floor, and the thing would fold in on me.

I don't know what to do to fix it. But something needs to be done to give it the appearance of mass. Perhaps if you created lumps that vaguely form the couch shape, and then wrapped the exterior around those lumps. With a bit of physics, it will cling to the lumps in some places, and billow in other places. Add some seams and wrinkles to sell the look.

>> No.902287
File: 2.42 MB, 1536x2048, stardust.png [View same] [iqdb] [saucenao] [google]
902287

Am I GMI?

>> No.902289

>>902287
only if you keep practicing

>> No.902290

>>902287
>stardust
he's going to come out of the computer and punish you with some form of eternal torment

>> No.902292
File: 1.20 MB, 1075x900, amongification37.png [View same] [iqdb] [saucenao] [google]
902292

amongification

>> No.902298

>>902287
I have never been more excited for a model in my life.
I don't care how it looks at the end, you NEED to finish Stardust!

>> No.902308
File: 512 KB, 684x939, file.png [View same] [iqdb] [saucenao] [google]
902308

It's babby's first clothes for a game. So happy with the progress right now.

>> No.902310
File: 120 KB, 645x362, image_2022-06-08_220242474.png [View same] [iqdb] [saucenao] [google]
902310

>>902139
To continue talking about implementing this type of aesthetic, how would you then go about lighting it properly.

If you have an actual scene set up with lights that are baked would it work well if you use textures like these or do most people fake the lighting completely?

I work in Unity and have been blocking up environments and have been going insane trying to figure out how I want to handle my lighting in conjunction with the type of textures I plan to use.

Here is another game with a similar style, I fucking hate being an artlet

>> No.902311
File: 376 KB, 1034x293, image_2022-06-08_221229548.png [View same] [iqdb] [saucenao] [google]
902311

>>902310
Here is another example from the same game I'm trying to mimic. I have scenes set up with modeled environments that have their lighting baked, similar to the left photo.

In my head I think that I can just bake the lighting over the textures and it will look as expected with normal textures.

>> No.902326

>>902310
Haven't used unity but in blender i just put a bg which i scaled down to 1024x1024 and change it to indexed mode. Then add a sun light and adjust it to my liking to get a comfy feel.
>>902311
Game name ? I don't know exactly if they used vertex coloring for their textures because i can only seem to see lightning on the ground black matte thing which looks like light from an hdri or a skybox. Hopefully other qualified anons can help you as well.

>> No.902327

>>902032
Damn that is really good.

Do you have a social media account?

>> No.902330

>>902230
>>902121
Looks amazing, anon.

>> No.902333

>>902311
the ground texture looks like a seamless texture with environment lighting on it and roughness set to low

>> No.902343

>>902139
Make texture, scale down, then scale up... voila u have a blurry mess u people call art.

>> No.902345
File: 12 KB, 376x116, 1627395056688.png [View same] [iqdb] [saucenao] [google]
902345

>>902343
yup that's pretty much it without the post processing and limiting the color palette.

>> No.902346
File: 267 KB, 2161x1287, 1643581268071.jpg [View same] [iqdb] [saucenao] [google]
902346

>>902093
somehow the files for this texturing project were just completely missing when i woke up this morning, so i spent another few hours redoing them lol

gonna photoshop in the details later, but i'm kinda liking how it looks so far

>> No.902347
File: 2.85 MB, 1280x720, 1645517238133.webm [View same] [iqdb] [saucenao] [google]
902347

>>902346
UE5 render

>> No.902356
File: 157 KB, 640x362, ElG3-XHWMAUhYaI-1.png [View same] [iqdb] [saucenao] [google]
902356

>>902326
It's a project from Sukeban Games: https://log.sukeban.moe/category/project-d/

I love this style as it is a bit more detailed than ps1/psp.

Any more details on the elements that go into this type of work, specifically textures, is appreciated. I'm completely ignorant to anything outside of modeling and dynamic/baked lighting.

>> No.902358

>>902130
Yesterday I did some research on 3d inverse rendering. There was that video being doom posted on NNs taking images and spitting out 3d models along with the textures and lighting, but I've asked the author and they tend to use >50 shots from various angles when doing inference. With just 1 he said I'd see bad geometry as soon as I moved the camera. It is too bad, but it seems inverse rendering is only a game changer for people doing photogrammetry. Not me, in other words.

NNs are good at speeding up simulations, so in the not so distant future, we will get orders of magnitude faster renderers and simulators, probably when AI chips start getting en-volume. But for now I'll be satisfied with just style transfer. It will be a huge time saver all on its own.

>> No.902360
File: 651 KB, 1244x868, 1654763417467.jpg [View same] [iqdb] [saucenao] [google]
902360

nice rig

>> No.902372

>>902347
Looks very nice. Only thing missing is some VHS static on the screen!
Seems a bit low poly in some places like the hinge on the top but depending on the Level of detail you need it should be fine.

>> No.902374

>>902347
What can i say... UE5 is the king of plastic thats for sure.

>> No.902378

>>902074
You've got some nice skills anon, keep up the good work. looking forward to seeing more from you.

>> No.902385

>>902347
Roughness feels off
>>902356
textures is the easy part really, 128x128 or 256x256. Use a limited color palette with the help of indexed mode instead of RGB. Maybe add a postotize effect for a good measure and then the rest is uv unwrapping (god help us all) and modelling

>> No.902388
File: 461 KB, 598x632, 125157547543.webm [View same] [iqdb] [saucenao] [google]
902388

adding a pretty basic implementation of deceleration to prevent overshoots results in some pretty neat "surrounding" behaviour.

>> No.902403

>>902358
Dude honestly in all the time you have been posting ai tests you could have rendered a 1 minute hd no noise walktrought of that room. If you don't have the machine don't sacrifice the look for time limitations, sacrifice the resolution. This posting of yours is getting on schizo levels. Chill man.

>> No.902405

>>902403
lol, i think he started 3d so he could use it to do a paintover and learn drawing. i think he just has crippling autism.

>> No.902407
File: 417 KB, 1200x640, matrix-cat.png [View same] [iqdb] [saucenao] [google]
902407

>>902405
Yeah but he even modelled the window mechanism so I do not understand why destroy all that time and work with a stupid ai filter because hurr dur my machine. I feel like I am in this scene.

>> No.902409

>>902407
>why destroy all that time and work
like i said anon, crippling autism.
i'm guessing he tried drawing, found it hard and gave up. same thing when he tried to do basic polymodelling. i've lol'ed when he's like i modelled this cardboard box in moi.
just leave him be, he'll get bored and move on.

>> No.902412
File: 21 KB, 384x373, 88888888.png [View same] [iqdb] [saucenao] [google]
902412

>>902374
You can adjust roughness and shading models in ue just like any other renderer.

>> No.902415

>>902409
Yeah I guess. I suppose when you turn left all the time you'll eventually get right.

>> No.902417

I'm doing an axe :3

>> No.902418
File: 52 KB, 412x750, Axe.png [View same] [iqdb] [saucenao] [google]
902418

>>902417
Forgot the pic

>> No.902429
File: 213 KB, 504x634, obabo.png [View same] [iqdb] [saucenao] [google]
902429

trying to learn sculpting, and I just made this.
what went wrong? I know it's bad but I can't exactly figure out what's bad about it. I used obama as reference.

>> No.902432

>>902407
It is not like I knew I would end up using NN style transfer when I started this. This room is just my second mastery project where I get a handle on modeling props. It that I ended up wasting a lot of time modeling what won't appear in the scene, but who cares about a single scene. The lessons from this are intended to serve as a basis for many more in the future.

As if scraping hard work is a big deal.

>>902403
You misunderstand the point of those tests. The basic 3d rendered look appears as a bunch of plastic and it would take a huge effort and heaping of details to get it to look decent in 3d. Substituting that with a painterly look is a large improvement, and I can do it with no effort at all. Wouldn't you be elated in my place? Before discovering NN style transfer I thought I might have to repaint it by hand.

>> No.902433

>>902432
Ok anon you do you. I do find value in experimentation.

As if scraping hard work is a big deal.
>For me yes but only because 3d is what puts food on my platter. As for myself I do find value in the Pareto principle.


>nded up wasting a lot of time modeling what won't appear in the scene...might have to repaint it by hand.

If you find value in modelling props that will not appear then you should consider painting it by hand if learning is your goal. Why waste time when you can waste some more.

>> No.902434
File: 56 KB, 600x800, wollsmoth.jpg [View same] [iqdb] [saucenao] [google]
902434

>>902429

The face traits seem not to be proportionate to the size of the head. Look at picrel I tink its the same problem.

As for sculpting I think you should learn to draw a face by hand. Drawing lets you identify the most important traits and that synthetic process translates well to scultping.

>> No.902438

>>902433
>If you find value in modelling props that will not appear then you should consider painting it by hand if learning is your goal. Why waste time when you can waste some more.
Because when painting or drawing, I'd have to deal with perspective and shading myself, while in 3d, the program does that all on its own. It is hard and requires talent to do well in addition to requiring large amounts of practice that I honestly don't feel like doing. Throughout my life there was no indication that I had any talent in 2d art, unlike in programming for example. 3d offers a lot over 2d to those who bother to master the tools. I got into it because of sculpting which I found easier than drawing because I could refine endlessly and just focus on form rather than numerous other aspects that drawing requires, but 3d software happens to perfectly automate.

As a matter of principle, I intend to position myself so that as much as possible the improvement in the quality of my art come from better computer hardware and algorithms rather than my own practice because in addition to 3d, I'll have to do writing, music, programming and who knows what else to meet my goals. If I had a machine that could read my brain waves and turn that into digital paintings, you bet I would not be bothering with 3d either. I might have time to do a few projects from start to finish in 3d, even if it takes me months, and in fact that has happened, but I do not have the time to approach it like a dedicated artist would. I'll do the covers for my VN and then move on to music next as well finish that Zbrush course.

I think I am good now as far as my 3d skills go now, but I am getting hung up on ML again. When it comes to it I am like a crack addict.

>> No.902441

>>902438
https://ali-design.github.io/deepcreativity/

Instead of me suffering over it alone, why not join me? Right now I am dying to know if diffusion GANs have controllable generation (think DALLE) like vanilla diffusion models, but I know that if I start doing research that would be months of time going down the drain. It doesn't really matter, because even if it can't, somebody in the future is going to come up with a model that can.

>> No.902451

>>902438
Yep. Crippling autism.

>> No.902460

>>902102
Looks good! Nice work , but i think it deserves some better eyes.

>> No.902462

>>902441
Ew an academic.

>> No.902464

>>902438
>but 3d software happens to perfectly automate.
We are absolutely getting AI'd then when it was already happening to 2d artists last year (not just now with the DAL 2 hype) because we're not even artists in truest sense of the word. We're 3d printers that aspire to semi-trace from reference someone else's vision at a studio. 3d people don't even have their own gesture practice ffs.

>> No.902471

>>902464
That's why you come up with your own vision the same way someone else did whose reference you're tracing.
AI isn't capable of doing that. A vision is something that humans have because we have, for lack of a better word, "soul". AI doesn't. AI can only take a bunch of references and make something out of them. For every one that's a hit there are 6m others that aren't. They're all based on what human imagination gives them to work with.

>> No.902473

>>902438
>>902432
>>902471
>>902464

Get the fuck out to >>>>/g/ We do 3d modelling here.

>> No.902478

>>902298
Thanks fren

>> No.902479

>>902471
When I say vision here in terms of industry I don't mean some once in a decade avant-garde cult classic. I mean generic carbon copies that both 3d and 2d are full of these days that the marketing AI see people eat up like pig shit.. Art Station already is full of work that looks like it was done by an AI. You really think another elf or goblin is beyond AI? You think it can't come up with necromorphs just by mashing concepts and references together? /ic/ is already pissing it's pants and they should have less reason to be worried than software monkeys who are getting carried hard by /g/ compared to traditional sculpters.
>soul
Oh ffs you're going to underestimate the potential of AI to do even derived work with that magic word? It doesn't need souls it just needs to pass the Turing Test. You're making a gamble here that you know you'd eventually lose money on. They're just assumptions what make you think AI go on to engage in emergent behavior?

>AI can only take a bunch of references and make something out of them.
Yeah try billions

>> No.902481
File: 9 KB, 307x310, firefox_HpmDp47kDG.png [View same] [iqdb] [saucenao] [google]
902481

>>902412
U didn't get the joke it seems.

>> No.902495
File: 1.21 MB, 1282x702, ShoulderSolar.webm [View same] [iqdb] [saucenao] [google]
902495

>>902230
So by drastic design changes i apparently meant slapping on more stuff. More Features, more better.
I was thinking way to much about how to integrate some solar panels into the suite. And some beefed up pauldron to introduce some asymmetry like seen in late medieval armor where the weak arm is reinforced. Not truly sold on this concept.

>>902235 throwing around some more ideas. Instead of finishing up the legs....


Lemme know what you guys thing / why its retarded or something.

>> No.902508

>>902495
Solar panels on a suit are a stupid idea. The uncertain alignment towards the sun, the extra weight and the penalty to mobility are not worth it.

As a concept its cool but that solar panel area could charge an iphone in about 6 hours on direct earth sunlight.

I've already critiqued the usability of some of your suit's aspects so I wont do it more. You do you.

>> No.902514

>>901947
>>901964
reminds me of a nightmare I had as a kid
trapped in a clock room with some skeletal version of a creature from some old RTS settlers type game
trapped in a dark room with the clock running down as more and more of the room becomes instant death zone

>> No.902525
File: 95 KB, 720x340, marsianer.jpg [View same] [iqdb] [saucenao] [google]
902525

>>902495

... it would make more sense when the suit itself has some solar panels integrated, or, a device, that gathters energy when you move, like this swiss watches who have this mechanism which charges the clock ...

...or, why not a use a poncho, like in the django movies, but made out of solarpannels?? the person who wears it can put it off and lie it on the ground to charge something, and when on the run, wear it as a poncho.

>> No.902526

i hate paint weight so much

>> No.902527

>>902508
You're basing that idea off current technology. For all we know those solar panels could be many more times efficient than what we have now. They can be much lighter and stronger and act as a secondary armor too.
What I mean is that, for example, Subnautica has technology we aren't even close to possessing. But we don't let that spoil things because it's science fiction and a fun game.

>> No.902528
File: 580 KB, 799x868, solarpanelponcho.jpg [View same] [iqdb] [saucenao] [google]
902528

>>902525
Nice idea. picrel

>>902527
It's not based on technology it's based on physics. In the equator the sun produces about 15/h watts per square foot. That hould charge a cellphone in 1 hours at 100% efficiency including perfect heliostatic alignment.

Anyways you do you.

>> No.902532

>>902528
This would mean that a perfect 1sq foot solar panel at 100% efficiency would charge at 7watts on a surface like mars's. A smartphone consumes about 12 watt/hours so it would have to be off to charge.

>> No.902540

>>902429
>I know it's bad but can't figure it out on my own I need someone to do the thinking for me.

You are on your own pal.

>> No.902547

>watch one random paint weight videos and a few paint weight ref
>understand what should i do
amazing, can't believe i love paint weight now, still, fuck paint weight

>> No.902560
File: 83 KB, 453x403, 1618843249490.jpg [View same] [iqdb] [saucenao] [google]
902560

> Maya
> Zbrush
> Marvelous Designer
> Marmoset Toolbag
> Substance Painter
> Blender addons
The blender addons are okay but how the fuck am I supposed to pay for all this as an honest hobbyist?
I mean I did pirate all this but sometimes, I'd love to just enjoy the automatic updates and shit.

>> No.902561
File: 1.39 MB, 1411x890, Screenshot_1.png [View same] [iqdb] [saucenao] [google]
902561

>>901909
Doing some wrapping up

>> No.902567
File: 1.49 MB, 540x387, 1641797159500.gif [View same] [iqdb] [saucenao] [google]
902567

>>902560
Wrong board, wrong thread, wrong everything, disregard this post

>> No.902573
File: 836 KB, 1584x855, Slim B 08.png [View same] [iqdb] [saucenao] [google]
902573

Back to hands, I finally got the hand topology to match the wrist. So I can finally join them! But the funny thing is, that I want to increase the amount of edges around the arm, so I think I'll have to reconfigure the wrist area. Right now, the arm is 12 edges around. I'm thinking about doubling that to 24. But this model is getting very expensive already. I'm already up to 27 triangles without subdivision. And I haven't done feet or hair yet.

>> No.902574
File: 702 KB, 127x118, skM2.gif [View same] [iqdb] [saucenao] [google]
902574

>> No.902576
File: 124 KB, 1538x1204, 45g45g45g45g54g.jpg [View same] [iqdb] [saucenao] [google]
902576

>>902173
just make faces that match it like I drew. when your done making it low res you can add holding edges or creases or whatever so you can smooth it, or just leave it low res. But that's the key to modeling. reduce complex things to the lowest poly count possible and keep it simple. only increase the complexity when necessary to get to more detail.

I model everything low, and then I add holding edges and turn it into a subdivision surface.

>> No.902596

>>902576
that looks so complex to do, though i will do some booleans. how do i start modelling complex shit like that?

>> No.902597

>>902596
>how do i start modelling complex shit like that?

Start by modelling not so complex shit. Then when you feel comfortable model some medium complexity shit and then you can venture to complex shit?

Do I need to explain basic logic to every tard that ventures into /3/?

>> No.902598
File: 219 KB, 623x737, Y25t2bQ5wo.png [View same] [iqdb] [saucenao] [google]
902598

This looks hideous and the topology is pretty trash but it does seem to match the tutorial I'm following so I guess I'll keep going

>> No.902599
File: 144 KB, 1607x1113, whaleModeling_v2.jpg [View same] [iqdb] [saucenao] [google]
902599

making a blue whale for a personal project

>> No.902601
File: 77 KB, 782x1050, whaleModeling_v3.jpg [View same] [iqdb] [saucenao] [google]
902601

>>902599
added dimension to the eye area

>> No.902604

>>902528
>in 1 hours at 100% efficiency
Yes and if you wear it on your suit at all time on a planet with potentially thinner atmosphere and more sun exposure that means you are continuously charging equipment. What's your point?
>>902532
>dude i'm not basing it off current technology

>> No.902605

>>902598
cute ayy

>> No.902606

>>902495
Bro... there is a liquid that boils at the temperature of ur body... just make a pseudo generator that uses that as basis... u will generate more power than walking with solar panels inside a dungeon.

>> No.902607
File: 669 KB, 1132x770, 1644653895464.png [View same] [iqdb] [saucenao] [google]
902607

>>902599
made a low poly version a few days ago with the same reference lol

>> No.902610
File: 2.86 MB, 1280x720, 1642733962063.webm [View same] [iqdb] [saucenao] [google]
902610

>>902372
i appreciate it anon, here's a render with a tv static material i made a lil while ago

>>902374
>>902385
agreed lol, something does feel really off and i've been fucking around with the roughness map but nothing ever feels completely right. i'm sure my lighting doesn't help, but oh well

>> No.902636
File: 223 KB, 883x575, 1623782316698.png [View same] [iqdb] [saucenao] [google]
902636

>>902597
what about something like this?

>> No.902637
File: 64 KB, 292x388, 1630892217543.png [View same] [iqdb] [saucenao] [google]
902637

>>902636
tried doing a basic boolean but

>> No.902641
File: 224 KB, 611x533, 1632283024411.png [View same] [iqdb] [saucenao] [google]
902641

>>902637
huh, surprised me it didn't look like total shit

>> No.902647

>>902180

are you the furry war crimes guy?

>> No.902648 [DELETED] 

>>902180

are you the furry warcrimes guy?

>> No.902650

>>902636
>>902637
>>902641
stuff like this makes me glad i learned to model on maya where doing everything is painful so the easiest option is to do it right and not spend your life being a booleanlet

>> No.902657

>>902650
thanks for such encouraging words anon

>> No.902670
File: 32 KB, 339x509, istockphoto-145832388-170667a.jpg [View same] [iqdb] [saucenao] [google]
902670

>>902573
Your thum isn't supposed to look like that, anon...

>> No.902672

>>902657
i hope you take it in the spirit it was intended:
bullying you into submission

>> No.902677

>>902650
Uhhh based and true 3d artist pilled

>> No.902682

>>902641
Problems with booleans only start once you begin beveling those edges. I am honestly tempted to just mark them as freestyle edges.

>> No.902686
File: 796 KB, 1072x964, image_2022-06-10_110519261.png [View same] [iqdb] [saucenao] [google]
902686

How do I prevent this?

>> No.902687
File: 517 KB, 1920x1080, Earth.png [View same] [iqdb] [saucenao] [google]
902687

>>902130
Did some ice breaking work for what is going to be the VN cover.

One cool thing that I've figured out is that using a combination of a Data Transfer node, and Copy Location and Track To constraints it is possible to set up the scene so that the night lights area always follows away from the light. Using a holder object to hold the vertex colors and data transferring them to the earth one allows me to rotate the earth freely while keeping the mask steady in the unlighted area, while the other two constraints ensure that the holder object is always ontop of the Earth and pointing towards the sun.

>> No.902689

>>902686
I could be wrong but I'm sure you just need to mark a seam and then unwrap it

>> No.902692

>>902686
>How do I prevent this?
you need to use a 3d painting program like substance painter... or photoshop. you can make a tileable texture like this.

https://makeitcg.com/create-tileable-textures-in-photoshop/236/

>> No.902696

>>902607
lol thats funny I thought I chose a rare whale to model because I didn't see much reference.
>>902637
>>902641
boolean is typically bad to use because the geo moves out messy and cant be uvd easily. and its a crutch to not learning how to model properly
>>902650
>stuff like this makes me glad i learned to model on maya where doing everything is painful so the easiest option is to do it right and not spend your life being a booleanlet
I agree learning to model with proper topology is best in the end
>>902657
>thanks for such encouraging words anon
go learn some topology and break froo of being a booleanlet

>> No.902700

>>902689
>>902692

I did make the texture tileable, but won't it do this no matter where I mark the seam to unwrap?

>> No.902701

>>902700
You need to unwrap it completely in the 0,1 uv space.
If you want to skip that step you can project the texture with triplanar mapping which automatically blends the seams it creates but in this case I think the regular approach should be very easy for you.

>> No.902702

>>902700
>I did make the texture tileable, but won't it do this no matter where I mark the seam to unwrap?
hmmm looks like there's something wrong. if it was truly seamless that seam would go away... it could also be a problem with a normal map or maps you have. id have to look at the scene itself to understand it. make sure the object is properly unwrapped, apply one texture to it and see if the seam goes away, check if its truely tileable ect...

>> No.902703

>>902702
>looks like there's something wrong. if it was truly seamless that seam would go away.
No it wouldn't. You have to actually unwrap the model properly for a tiling map to tile.

>> No.902704

>>902700
i think i figured it out... try scaling the UVs to make your texture smaller on the tree. I think whats happening is yes you made a tileable texture, but the tree is too short and doesn't get a chance to tile it. so scale it up so you can see the tiles and check it that way

>> No.902705
File: 168 KB, 458x584, Slim B 09.png [View same] [iqdb] [saucenao] [google]
902705

>>902670
I haven't completed shaping of the hand, so I realize it's a little off. But I bent the thumb a little so as to emphasize the joint. You can make your thumb look like that too, if you spread it out more, then give it a slight bend. It's not a completely relaxed posture. But a stretched posture. I stretched the thumb out a little, because I was referencing my own hand, and trying to get that bit of webbing between thumb and index finger to taut. Hopefully, when it comes time to rig the hand, and I make the thumb more relaxed, the webbing will squish into realistic padding. I already changed some of the polygons in that region to allow for pliable deformation.

>> No.902707
File: 330 KB, 497x603, mechmodel4.png [View same] [iqdb] [saucenao] [google]
902707

>>902005
finished my mech model. Already successfully 3D printed 2 of them. Just need to rig and make a few poses now. Forgive the bad lighting. In a rush.

>> No.902717

>>902707
Very cool Anon.
Filament or Resin Printing?
Also show prints pls. Really interested how the model detail transfers.

>> No.902723
File: 1.15 MB, 923x1118, image_2022-06-10_144113521.png [View same] [iqdb] [saucenao] [google]
902723

>>902689
>>902692
>>902701
>>902704

Thanks for the tips, you got me thinking.
I did a quick test to check it on a cylinder, and the seam dissappears when I do the same mark and unwrap. So that rules the texture being the problem out. I think I just need to figure out how to unwrap this properly, the other tree model is much more complex.

Sorry for the newbie questions, I literally just started my pursuit of 3D/game dev.

>> No.902724

>>902723
As I said, unwrap it properly. The default primitives are unwrapped properly by default which is why it tiles perfectly.

>> No.902731
File: 1.97 MB, 728x624, %pn_2022-06-10_15-34-29.webm [View same] [iqdb] [saucenao] [google]
902731

I've got a lever here with a little sticky pull animation, when it's done it activates the gears with the counterweight. When it's done with its animation it activates the portcullis.
I can link them together to make apparent rue goldberg devices. The counter weight animation can let you know a particular switch has already been activated.

>> No.902741
File: 300 KB, 2982x851, drag.jpg [View same] [iqdb] [saucenao] [google]
902741

>>902026
for reference, here's DAZ dragon 3's neutral pose.

>> No.902747
File: 1.19 MB, 1536x2048, toyman.png [View same] [iqdb] [saucenao] [google]
902747

>> No.902751

>>902741
thanks for the reference

>> No.902753
File: 135 KB, 1920x1080, 1642280289316.jpg [View same] [iqdb] [saucenao] [google]
902753

I didn't realize how big those candle flames were until I rendered it.

>> No.902767
File: 336 KB, 1736x900, 1654914251277.jpg [View same] [iqdb] [saucenao] [google]
902767

finally, took me a while to find the right how to rig the leg tutorial

>> No.902770

>>902717
>Very cool Anon.
>Filament or Resin Printing?
Thanks! I love making my own stuff. I'm doing resin printing. I think I need to adjust my supports, as 2/4 print attempts have failed in some way or another. Of course, it might be easier if I broke the model up into a few separate pieces. Just going to keep experimenting.
>Also show prints pls. Really interested how the model detail transfers.
Sure thing. Either tomorrow or the day after, when I get to my days off.

>> No.902779
File: 47 KB, 374x500, tenor.png [View same] [iqdb] [saucenao] [google]
902779

>>902753
it doesnt appear like any light is coming from them at all. Candles have lazy solid flames anyways. Not like a torch
Just use a texture like this

>> No.902792

>>902779
i'm sure he meant torch.

>> No.902793
File: 572 KB, 976x1312, prote20.png [View same] [iqdb] [saucenao] [google]
902793

crushed the polycount to 976 from 1.5k and added in hair plus a much more organic headshape.

>> No.902828

>>902779
There is, it's just not a lot. It's easier to see on the ones at the back because there's nine of them right next to each other.
I did do it the way you suggest though (I think). I just don't have the emission turned up very high.

>>902792
That too now that I look at it. They were supposed to be kinda big anyway. I fucked around with the focal length and I don't know if that makes any difference. Like how the seats look huge in comparison to the character, even though they're to scale, and the character is closer to the steps than it seems.
I'll just keep fucking around and see if I can get something to work.

>> No.902836
File: 327 KB, 853x778, file.png [View same] [iqdb] [saucenao] [google]
902836

>>902378
arigato.

>> No.902846
File: 512 KB, 1139x1055, file.png [View same] [iqdb] [saucenao] [google]
902846

>>902647
lmao i'm not the furry war crimes guy, but he sounds like an interesting one

still working on wolfy, here she is with a singular boob becuse im too tired to work it into the topology today. i'm happy with how the hands came out, got claws and pawpads. end is in sight, after boobs its just the hair and clothing and then rigging time wahoo

>> No.902850
File: 209 KB, 1120x840, 11-06-2022.jpg [View same] [iqdb] [saucenao] [google]
902850

bros I don't know how to post this motherfucker without it looking retarded, an advice on a cool pose would be much appreciated

>> No.902851

>>902850
any pose from the manga should work
also, give him a smug face

>> No.902853
File: 212 KB, 1920x1920, efea90496f92bf6962ee7f2756ae2549f01cfd6b.jpg [View same] [iqdb] [saucenao] [google]
902853

>>902850
like this maybe?

>> No.902855
File: 3.66 MB, 1920x1080, earth in space.png [View same] [iqdb] [saucenao] [google]
902855

>>902687

>> No.902858

>>902850
nice cell bro, just copy something from the anime or manga dood

>> No.902883
File: 603 KB, 1194x766, blender_MHrKguST89.png [View same] [iqdb] [saucenao] [google]
902883

>>902327
Yes.
https://twitter.com/ApplefatGov

I don't do that much 3D work, I'm mostly a 2D artist. but I've been doing 3D about as long as 2D so I can't really get away from it, haha.

Something I'm deciding to finally finish, a submarine for a Subnautica mod.

>> No.902884
File: 64 KB, 720x720, contain_800x720.jpg [View same] [iqdb] [saucenao] [google]
902884

>>902308
nice. i like it.

>>902561
congrats. nice character. particle hair? I think a slightly lighter more vibrant blue would look nicer. nice volume also always looks good too, but that's just a suggestion.

>> No.902891

>>902883
>>902032
i like your style, very clean and pleasant on the eyes

>> No.902894
File: 2.40 MB, 362x487, ezgif.com-gif-maker(31).gif [View same] [iqdb] [saucenao] [google]
902894

he died

>> No.902926
File: 2.00 MB, 1920x1080, 0001-0250.webm [View same] [iqdb] [saucenao] [google]
902926

Yeah... whatever.

>> No.902933
File: 1.57 MB, 1920x1080, earth in space (pencil work).png [View same] [iqdb] [saucenao] [google]
902933

>>902855
This net is as usual quite good at drawing. I bet if I tried the CSP filter I'd get a way worse results. Mangakas who spend a lot of time converting photos could benefit from CAST.

>> No.902935
File: 1.22 MB, 1920x1080, earth in space (great wave off kanagawa).png [View same] [iqdb] [saucenao] [google]
902935

>>902933
This came out better than I thought it would. When it comes to drawing styles, I feel settled, while when it come to painting I know I need to do a lot of research. After I finish the cover, I am going to have to take some time to adjust the CAST scripts to make that easier.

>> No.902937
File: 26 KB, 846x437, mahk.jpg [View same] [iqdb] [saucenao] [google]
902937

>>902731
>motion blur

>> No.902946
File: 414 KB, 1024x1024, Sphere01-b.jpg [View same] [iqdb] [saucenao] [google]
902946

>> No.902947
File: 252 KB, 910x910, Sphere01.jpg [View same] [iqdb] [saucenao] [google]
902947

>> No.902952
File: 679 KB, 381x307, tumblr_45fbda0e37552a076feddfc34e4b10e2_687cf136_400.gif [View same] [iqdb] [saucenao] [google]
902952

>>902894
I would recommend making the floating part more slowly

>> No.902953

>>902952
DIESOFCRINGE.gif

>> No.902957
File: 133 KB, 467x425, 1649697566577.png [View same] [iqdb] [saucenao] [google]
902957

>>902696
>>902682
>>902672
i tried but i am confused about the corner piece

>> No.902958
File: 23 KB, 356x500, 1643732528884.jpg [View same] [iqdb] [saucenao] [google]
902958

>>902957
le refernece

>> No.902960

>>902957
mein negro, make a cylinder with 6 sides and look at it from the top down.
note how the the edges of each face push outwards from the center along an angle.
you got str8 lines.

>> No.902970
File: 405 KB, 1288x597, 1638083356333.png [View same] [iqdb] [saucenao] [google]
902970

>>902960
aaaaaaaaaaaaaaaaaaaaa i am just a fucking dumbass monkey, my base is 8 faces tho

>> No.902971

>>902970
Use an octagon then :P

>> No.902972

>>902970
>>902957
Mate, just model one side and radially symmetrify it.

>> No.902997
File: 340 KB, 969x621, LADAfront.png [View same] [iqdb] [saucenao] [google]
902997

I think I "finished" my LADA model for now

>> No.902999
File: 124 KB, 695x397, LADArear.png [View same] [iqdb] [saucenao] [google]
902999

>> No.903002

>>902999
Checked and nice shitu desu.

>> No.903008
File: 71 KB, 1920x1080, rIG.jpg [View same] [iqdb] [saucenao] [google]
903008

Well shoot! Looks like this rig is going to need some clean up. :)
It is an awesome feeling seeing your character come to life a little even is a little messed up.
I wonder if I shouldn't just take a couple of weeks to properly learn rigging at this stage so I can make my own rigs from scratch.

>> No.903013
File: 1.01 MB, 2085x1529, BKn_S024.png [View same] [iqdb] [saucenao] [google]
903013

a rough body blockout i have been working on

>> No.903014

>>902957
Model the hexagon and then model the cutter for it. Your drops straight (horizontally) towards the base, but if you look at the ref carefully you'll see that it narrows as it drops. The vertical sides should not be parallel to each other.

You do not need to model all the sides by hand, once you do a single one, you can just do a radial array on the cutters. Or if you've applied them, on the piece itself after erasing the others.

Modeling this sort of thing would be easier in Moi than Blender.

>> No.903016
File: 194 KB, 758x845, 18-05-2022.png [View same] [iqdb] [saucenao] [google]
903016

>>902851
the og version of this kinda had a smug face but I don't think it really worked so I gave him a neutral one in the final version
>>902858
thanks bro, I was getting so frustrated with this project that you wouldn't believe me, it warms my heart to hear someone appreciates the work

>> No.903017
File: 770 KB, 1024x1024, 12-06-2022-2.png [View same] [iqdb] [saucenao] [google]
903017

>>903016
this is where current version is at, first time using Substance and I find it quite clunky, wanted to make him clench his fists but it's too hard to do in zbrush

>> No.903020

>>903008
>BOOBA!

>> No.903021

>>903013
looks good. i'd check that top buttock crease so that it comes down a little lower over the hip and maybe reduce the size of those knees a little (width) but it's just a blockout so time to take it to the next level.

>> No.903024

>>902926
It's come together nicely. Are you done with it? It looks complete to my eyes.

>> No.903025

>>902850
There's a face in his chest.
His pecs are two large eyes shut closed. The sternum and upper rib area creating a nose complete with nasolabial folds and cheeks. His abs create big meaty lips and a long chin.

>> No.903026
File: 797 KB, 1920x1080, vn cover (djeeta dark).png [View same] [iqdb] [saucenao] [google]
903026

>>902935
Here is a dark version to contrast the previous one. It has a murky, suppressed look, like the world is shrouded in fog. The net really understands lighting well, you wouldn't get this just by toning down the values globally.

I did some work on the scripts and now they are easier to use. I have a bunch of images that look great, but I'd get people calling me schizo autist if I started dumping so I'll hold back.

>> No.903027
File: 66 KB, 500x284, Sfondamento-dei-cieli-Gurren-Lagann-2.png [View same] [iqdb] [saucenao] [google]
903027

>>903025

>> No.903028
File: 1.29 MB, 1920x1080, vn cover (jigsaw).png [View same] [iqdb] [saucenao] [google]
903028

>>903026
This one has good contrast and a vibrant palette. It is well balanced. I am not sure if I should strive for balance or if I'll go with an intense look.

I am not really happy with how drawings comes out on this image, the one I posted was the only one worth showcasing.

I think that I'll grab the 25gb wiki art dataset tomorrow and try a much wider selection of styles. I'll also get started on the next scene. I'll try out Houdini's urban modeling capabilities for the first time.

>> No.903033

>>902850
A little thing that might make him look better, make his antennas longer and separate them a little bit more, they're not this perpendicular to his head in the anime. Also his waiting pose before the tournament might be cool.

>> No.903034

>>903017
the base is too big for that model, it could look ok if you add a couple of characters more

>> No.903036
File: 325 KB, 1238x732, 1634779212043.png [View same] [iqdb] [saucenao] [google]
903036

poly modelling

>>902767
post with local rotation axis visible

>>903017
my man has a bad case of anterior pelvic tilt

>> No.903039

>>903034
Too late to fix that, but maybe after this deadline I have i'll model an Arale poking at some shit to add to the base

>> No.903049

>>902508
tell that to my backpack which has a solar panel hanging on the top back in a plastic slot, shits cash money charges my phone while im backpacking through the mountains.

>> No.903051
File: 639 KB, 1634x893, 1655083605932.jpg [View same] [iqdb] [saucenao] [google]
903051

>>903036
Hope it's alright since this is my first time rigging my own model

>> No.903052
File: 483 KB, 1464x726, 1655083847566.jpg [View same] [iqdb] [saucenao] [google]
903052

>>903051
oh god oh fuck the mirrored joint

>> No.903057
File: 203 KB, 1246x905, 1655086753480.jpg [View same] [iqdb] [saucenao] [google]
903057

>>903052
already fix it on wings and feet, but somehow the tail and the head still fuck up even thought i orient the joint

>> No.903066
File: 327 KB, 631x725, 1634810469630.png [View same] [iqdb] [saucenao] [google]
903066

>>902972
yeah i just modelled 1 side and arrayed it
>>903014
both like pretty ass, should i just go with the upper one?

>> No.903069
File: 1.66 MB, 1447x810, Marnie.png [View same] [iqdb] [saucenao] [google]
903069

I'm taking a break from making characters and making something that's not a character. For once.
Blocked out the stuff, starting to make the shapes.

>> No.903074
File: 765 KB, 900x1080, 1635582639564.png [View same] [iqdb] [saucenao] [google]
903074

>>903066
tis good enough?

>> No.903091

>>903074

No. You failed at achieving 8 sides and it looks nothing like the reference image. Sorry.

>> No.903103
File: 1.61 MB, 1920x2700, 1637004905166.jpg [View same] [iqdb] [saucenao] [google]
903103

I got bored so I tried using an AI (https://huemint.com/brand-intersection/)) to generate color palettes for my scene. Some of the results are definitely too funky for a comfy living room but some others are pretty interesting. I'm tempted to write an addon to automate the process.

>> No.903121

>>903074
Bro.... check the math behind hexagon and octagon... heaxon for example rotates by 60 degrees, octagon rotates by 45 degrees... just use ur brain. Model one side correctly and just rotate it by 45 degrees and then merge the vertices by distance together.

>> No.903123

>>903074
Not sure why you made the balconies rounded like that. They should be flat to emphasize the octogonal shape like the rest of the tower. Should be a few seconds to fix.

>> No.903124

>>903123
All he has to do is make an octagonal circle for reference and just snap the vertices.

>> No.903125

>>903124
You don't need a reference for an octagon. You just employ radial symmetry at the top of the stack and you can model one side while seeing the completed result.

>> No.903127

>>903125
didnt know about that.

>> No.903130
File: 90 KB, 608x355, 1627361780503.png [View same] [iqdb] [saucenao] [google]
903130

>>903121
anon please understand, i couldn't learn to multiply until I was in 4th grade, and turning the empty 45 degrees gives this result.
>>903123
>>903124
>>903125
>>903127
thanks for all the replies, i will try to make it as beautiful as possible

>> No.903137
File: 1.09 MB, 900x1080, 1636837917007.png [View same] [iqdb] [saucenao] [google]
903137

>>903130
is this a texture problem or a model problem?

>> No.903138
File: 150 KB, 671x800, raoul-dufy_seated-nude.jpg [View same] [iqdb] [saucenao] [google]
903138

>>903028
>I think that I'll grab the 25gb wiki art dataset tomorrow and try a much wider selection of styles.
I've downloaded and checked it out.

Most of the stuff in the dataset looks like crappy children's scribbles, and even the good stuff has a somewhat washed out palette. I haven't tried out style transfer with it though, maybe I'll get something good once I do that.

As I suspected, the standards of modern and old art are on completely different. It really is necessary to make full use of modern tools to get an edge in this field.

>> No.903141

>>902731
the legend of zelda the ocarina of time from 24 years ago looks better than this. What are you even doing?

>> No.903152

>paste left wing movement to the right wing
>ruin everything
is it worth it to buy rig...

>> No.903154

>>903152
come back to correct side, comrade

>> No.903155

>>903154
i don't what to fix, the orient or the joint, i don't even know the right keyword to find the tutorial...

>> No.903157

>>903155
i was making a shitty joke about political ideologies, anon

>> No.903159

>>903157
that's explain why i'm retard...

>> No.903168
File: 190 KB, 837x807, frogmanwip.png [View same] [iqdb] [saucenao] [google]
903168

>> No.903171

>>903057
yeee most people rig t-pose so the rotation axis are automatically correct right angles. You can freeze transforms in Maya but game engines will still see the rotations if the joints aren't aligned to world

>> No.903174

>>903171
with that kind rotation axis is it normal for the rotation to be different between left and right wing? i'm kinda seething because i though i make some mistake again and want to stop but want to try again for a few days

>> No.903179

>>903174
Did you swap positive rotation and transforms to negative transforms and vice versa? (except z and y transforms)

>> No.903183

>>903179
i use orient joint to the world option

>> No.903188
File: 32 KB, 370x304, Screenshot 2021-02-18 232033.jpg [View same] [iqdb] [saucenao] [google]
903188

>>902946
That's one weird looking cabage head, anon.

>> No.903201

>>903183
ye but did you swap positive rotation and transforms to negative transforms and vice versa (except z and y transforms) when you copied the keys across

>> No.903212
File: 1.35 MB, 4013x1984, UNS-Converso.jpg [View same] [iqdb] [saucenao] [google]
903212

>>901909

>> No.903252
File: 1.54 MB, 1471x1769, sylv.png [View same] [iqdb] [saucenao] [google]
903252

>> No.903254

>>903201
i don't since i don't know how to do that

>> No.903265
File: 558 KB, 1441x831, Marnie&#039;s bedroom.png [View same] [iqdb] [saucenao] [google]
903265

>>903069
Fleshing it out bit by bit. Most of those items can probably be baked into a cube, but I'll focus on optimization at the end.

>> No.903311

>>903252
is this a sculpting program?
what's it called?

>> No.903312

>>903311
i think it's nomad sculpt, anon

>> No.903315

>>903311
>>903312
Yup, Nomad for ipad

>> No.903355
File: 2.16 MB, 900x720, 1641922801730.webm [View same] [iqdb] [saucenao] [google]
903355

>>901909
god i missed using after effects, the cam shake feels off but eh i just wanna wrap up this project

>> No.903358

>>903355
aahhh the ol "have to make it B&W, add a shitton of artifacts and camerashake"

>> No.903359
File: 232 KB, 642x401, jhygfd.png [View same] [iqdb] [saucenao] [google]
903359

Does anyone know how to animate the cloak with bones?

>> No.903360
File: 1.10 MB, 900x1080, 1652604332790.png [View same] [iqdb] [saucenao] [google]
903360

>>903358
yeah pretty much, the colored one looked absolutely shit because i couldn't make any good procedural textures and i just rolled with it and tried to make ww2 footage shit

>> No.903361
File: 771 KB, 900x1080, 1644009027059.png [View same] [iqdb] [saucenao] [google]
903361

>>903360
wrong pic, also forgot to mention the thing I learnt from this project is that, I really need to take the effort to see the smaller details and not rush shit

>> No.903362

>>903355
Brainlet who's never used after effects before and knows nothing about animation here, how do you do something like that? Do you have to render the animation in blender first and then open it in after effects to add the camera shake and what have you in after effects?

>> No.903366

>>903362
i rendered the the building and planes in blender but didn't like it so i imported it into after effects. Added an adjustment layer and added noise, changed the saturation to b/w and made a few key frames for contrast and brightness to emulate the exposure. Made the footage shakey with the wiggle expression and added a generic scratches green seen to tie it all together. After effects is wayy easier to learn than blender in my eyes. Also if you're going to make something WW2 shit then collect a lot of reference and look at the small details instead of trying to emulate what you see on a surface level.

>> No.903368
File: 79 KB, 1920x1080, Testing.jpg [View same] [iqdb] [saucenao] [google]
903368

Deformation mostly okay. Might have to revisit topology in future.
Been thinking of subdividing the mesh once before applying the rig. Surely that will give better deformation and more control with weight paint. Do normal maps help to make deformation appear smoother?
Anybody know what is acceptable vert count for character models these days?

>> No.903369

>>903368
you can keep the subdivision unapplied as a modifier so it's faster to work with while animating/posing, and just keep it enabled for renders.
>Do normal maps help to make deformation appear smoother?
No, they don't change the silhouette/shape at all. What you can do is make corrective shapekeys in sculpt mode while the mesh is posed to make the deformation smoother.

>> No.903371

>>903369
>What you can do is make corrective shapekeys in sculpt mode while the mesh is posed to make the deformation smoother.
what you SHOULD do is use a muscle simulation system to avoid 90% of this nonsense, like they do in the industry for the past 15 years at least

>> No.903372

>>903369
Thanks for the reply
>corrective shape keys
gonna have to look into that.

>> No.903373

>>903371
prove that you actually use muscle sims and you're not just a bucket crab then

>> No.903374

>>903366
Based. Thanks. My brother has used after effects for years, but I never see him so I can't really get him to help me out.

>> No.903375

>>903373
I have thousands of lines of ziva scripting under my belt as well as have written maya custom nodes to work with ziva with the maya api. Thats all I can prove

>> No.903385

>>903371
>>903375
>He decided to bring his bullshit into the /wip/ thread

>> No.903390
File: 183 KB, 974x683, MyoMyBeloved.png [View same] [iqdb] [saucenao] [google]
903390

>>903359
I have moved onto the next one. I'll probably use wiggle bones or research some more to find out what to use for the cape.

>> No.903391

>>903359
it would probably be easier for you to use cloth physics, are you animating with these inside of blender?

>> No.903395

>>903391
I only have blender so I'm planning to do that. I'll try to see how cloth physics will work though I forgot about them.

>> No.903397

>>903385
someone tells someone else to use corrective shapes on a character, i have to "corrective shape" their way of thinking. If you think muscle sims are bs you are the one with the problem. Hell, its past the midway point of 2022

>> No.903401

>>903397
>current year

>> No.903403

>>903390
>>903395
>>903359
https://www.youtube.com/watch?v=eggOO9CIyp4

>> No.903406

>>903397
look, not everybody has a team of artists and programmers at their disposal. It may be the industry standard but it certainly isn't the way for a 1MA to go.gpngy

>> No.903409

>>903359
https://youtu.be/55bJB5Wmd_U

>> No.903410

>>903403
>>903409
fuck blender is such a cool program bros. you can make magic with it.

>> No.903477

New pastr-- I mean, new bread?

>> No.903506

>>903477
>>903487