/3/ - 3DCG

>>899902
Wip anon here.
Well, that works too. Thanks for looking out and posting on my behalf.
Looks like I dropped the ball on that one. I went to bed thinking I had time to work before the thread hit page 5. Hitting the file limit was something I never really considered. It almost never happens.
Thanks again. I'll make it up to you all next time.

>>899922
Yes, somebody asked for a new /wip/ in the previous thread and i opened an emergency one, no collage this time sadly, it is one of the threads that I like to follow here.

Oh boy! I can't believe this is happening! A new /wip/ thread and I get to post first, everyone will get to see my masterpiece for the next two weeks! When I last made my own thread for my sculpt, I will admit that I wasn't entirely satisfied with what I had done. I did some thinking, and I figured out that my sculpt didn't have enough sex appeal, so I game my owl some tits. I will let you guys know when I start a patreon, because I think I finally have what it takes to GMI.

>>899962
>I game my owl some tits.
Oopsie, I meant to say "gave". Silly me, I must have gotten too excited.

Doing daz shit!

Working on a few dnd mini commissions. This is what I got for one of them so far. Anyone got any good resources for sculpting hair? It's a weak point of mine despite loving stylized hair so much.

And it is done, idk if I'll do the other 9 zombies, I should do at least a couple more, or I'll do some other character next.

>>899969
https://youtu.be/G3SVAN9TPbo?t=940

https://youtu.be/x3ZzXDcuPWw

2 approaches depending on your desired level of stylization.

Alright! New day, new thread! Time to return slaving over my base mesh, and squashing the feeling that I'm wasting my time!

Doing some raytraces of an apartment complex studio, I'm a beginner

>>900100

>>900101
slow and low quality renders courtesy of my dying water cooling

>>899790
anon I'm not literally saying to use a 16 segment loop
>>899965
horrible

>>900103
Well the boob I created has 16 segments on the outer loops. That's just the reality of the situation. How do you suggest I go about deleting the patch, while also reducing the number of segments down from 16? I don't see it.

I just downloaded blender.

>>900100
I find it weird how shiny the ceiling edges are.

>>900101
As well as the floor side here. Just what is going on in these images?

>>900120
It's a known glitch in which light seems to seep through wall connections with ceilings and floors, unfortunately the latest cracked version of this program is v10 and the latest version is v14, in which this is fixed IIRC
I made one more render

>>900102
I like this one in particular.

>>900122
Thanks, I enjoy making floor plans for tiny apartments.

Blue thing for scale

Get your new and improved Space Boots NOW! for 3 easy payments of your soul, first born, and feet pain.

Damn i suck at making clothing but the point of the project is to push my abilities to the limits to learn something.
Lemme know what you guys think.

>>900140
Blue thing isn't working so great for scale because we don't know how big the blue thing is because it's just a blue thing.
Whenever you put something in for scale it should be something identifiable. Like something we know the rough scale of from the real world if that makes sense.

>>900149
It's a person (1.7m tall).

>>900150
Yeah but what I'm saying is we don't know it's supposed to represent a person so we can't tell the scale just from it. To you you know because you're modeling it so you know it's supposed to be a human. To us it's just a blue thing that could be 1" tall or 100' tall.
If it was a model of a human then we'd be able to have a rougher idea of the scale because we know the rough average height of a human.
I'm not knocking you btw, I'm just trying to help you out.

>>900149
Now that I think of it, a blue thing may work for me when scaling a scene since it uses a color that stands out (relatively) and is roughly the dimensions of a person, but It would be more recognizable as a person to other people if I gave it some simple arms, legs, and a head.

>>900151
It looks like I posted >>900152 too late. But yeah, I get you.

>>900104
just dissolve every other edge?

>>900159
Doing that will come at the cost of the titty shape. The titties are too angular already. It's not noticeable when I turn on subdivision. But when that's off, parts of the boob are very jagged. I was actually thinking of creating more edges to round them off.

>>900168
>i want less polygons
>but i also want more polygons
dude just make up your mind

Look at ur hand


>you are now conscious.

>>900100
>>900101
>>900102
>>900121
>>900130

Archviz pro and architect here. That is a nice start anon is this your design?

On the architectural design:

Depending on the country and zoning laws this design is extremely inefficient for various reasons.

1.Kitchen and bathroom zones should be located on the same service to lower the quantity of slab water connections and number of ventilation ducts. W.C. is located on a corner that does not permit efficient in-wall ventilation and waterway (in case the building is several stories high with a typical floor-plan). Fridge and all appliances should be located on the same service wall in order to reduce the number of connections.

2.Too much space of the private area is wasted on wall divisions and useless areas (some countries and zoning laws do make obligatory isolated service areas of the kitchen and bathroom). At this scale of apartment the line between architectural design and interior design blurs so many of the divisions between spaces and uses should be flexible. Consider using light wall panels and projected awnings for bedroom division. Consider integrating the kitchen with the social area using a single piece of furniture that services cooking, table and multimedia uses. Consider a wall bed. If the entry is required the door should let an entry furniture to occupy the wall.

3.Goes with the former but try to use the full width of the facade to catch light. This can be achieved with projecting awnings or panels for the bedroom and a wall bed. A good proportionate tiny apartment uses 10% of its total area in circulations.

>>900188

> le me checked

4.Understand the user. For example most student residences do not design bathroom with bathtubs in order to save space. Living areas are not generally obligatory for single user apartments. Washing machine, dishwasher, fridge and othe appliances should be scaled to standard according the the usege and number of residents in the apartment.Example: Is 12 dish set dishwasher necessary for an apartment that can only fit 2. Is study furniture applicable in the bedroom when it is only a 2 person apartment?

On the renderings:

0.What's the rendering plugin or software. I can help you set the most optimal parameters.

1.Tiny apartments do enjoy on one class of materials and those are light materials. Dark wood on the flooring would eat up the little light that would enter from the kitchen.

2.Good renderings do enjoy from harmony in materials my best advice is to look up tons of reference images from good sourced design magazines or websites. Harmony is the key no color or material combination is correct on its own.

3.If you need any asset I can give it to you. I have literally terabytes of architectural 3d models.

>>899902
Better Flight Render

>>900192
Cycles

>>899902
no buly

>>900195
Is that an A10? If yes then nice.

>>900197
Even in silhouette form that Ugly fucker is still visible.

>>900177
Poop

>>900174
I want as many polygons as necessary to make a good shape at the base level and nothing more.

>>900192
>>900193
Looks like a 90s video game render. If that's what you were going for, then congrats. If not, then I'm sorry.

>>899902
recently came across naohisa inoue's paintings and i am completely captivated by his works, i thought i will try to recreate this painting to some capacity in 3d, i thought of just painting different types of bushes and plants and insert a multitude of particle systems on a cylinder to achieve this style but then i thought of geo nodes to get something like this. any thoughts or suggestions you guys have to help me with this project?

>>900233
a quick test with a particle system and it looks like it can get the job done, will geo nodes give me any advantage tho?

>>900225
that's the goal.

Can anyone explain (or point me in the direction of a video if it will take too long to explain) how to do that "high poly mesh bake to low poly mesh" thing? I've tried googling it and got nothing that I'm looking for.
I watched it on a video once and haven't been able to find it since. In the one I watched it was like

>Models something super high poly, something like a million triangles
>Does some wizardry
>Bakes it
>Now it has like 6k triangles

It was as if the whole model should have had millions of triangles, but when it was finished it had about 50k and was indistinguishable from the one with millions.
Does anyone know what I'm talking about? I thought I found it on Markom3D last night, but it was a totally different method and seemed to have a lot more steps to the one I watched.

>>900195
why is it triangulated?

>>900239
You bake details from a high poly model into a texture than you then put on a low poly model. The reason you'd want to do a high poly model to begin with is because it might be easier to sculpt the details on a high poly model and then texture it.

>>900258
Thanks fren that makes sense. I'll give it a try tonight.

>>900257
Why would it not be?
Modifiers are great because i can toggle them without changing the topology until its ready for export into the game engine.

>>900262
further clarification, outside shell is only visible from outside, the inside parts are all high detail because they are what will be seen up close in the first person view. the models are created together and seperated for their different uses in runtime. any details on the outside that are visible on the inside are recreated in high detail for the inside mesh view.

I do this for fun.

>>900263
Nice shit anon.

>>900239
>>900259

It is literally called low poly baking/retopo, some call it normal baking etc etc. Maybe this can help? If you tell me the software you use I surely can find a resource video. Here's an example in blender.
https://www.youtube.com/watch?v=o568T79egTs

>>900130
>>900188
>>900189

As a followup personal exercise I designed and modeled a 30sqmt2 apartment similar to your general dimensions.I hope this helps you to get the general idea of my design process of a small residence. Did it remotely on Anydesk with my home pc using only max so I had no access to an architecture suite but the general idea is the same. It is not the best architectural design but I did try to condense everything in a single piece of furniture on one side of the apartment. Software used: (3ds max 2022 / Corona renderer )2100usd/year on a professional licence, about 500usd/year on an indie licence or 60 usd/year on a student licence. Hardware: 1700 ryzen 7 / Geforce 1060.

Here's the whole 3 hour process: https://youtu.be/dnZoJNoMv_0

Cheers anon glad to help improve your process.

>>899902
Done with Pablo's Zbrush for Illustrators course. The course itself is not bad and gave me some ideas, but it is not what I am looking for my own style. What I have to to get my own style is paint masks and use gradient maps or posteriazation + a bit of blurring by hand to get that cell shaded feel. Freestyle edges for at least the outlines. I'll find ways of speeding the process of creating masks. One idea is the export freestyle edges as svg. If the lines are a vector, CSP will able to put the mask right in the middle without trouble. Putting flat colors on each object and exporting the raw diffuse map is also another option, having flats will make it easy for the magic wand to select them just as with vector freestyle edges. It won't be too much trouble, I just need to make up my mind to do it.

Before I do that what I want to do is try a particular idea.

My view of painting is that it is just about using the reference values as a probability distribution of canvas strokes. From that view the difference of drawing vs painting is just the shapes of those strokes and the degree of overlap. Assuming I have a render like the one of my room, I know that I should be able to the brightness values as density for scattering strokes as if they were objects. In order to determine their direction what I'd need are screen space normals. I'll find a way to extract those as AOVs and export them through the compositor. I can also use depth to determine the stroke scale. I'll test out this idea using Substance Designer. It has nodes like the Shape Splatterer that are well suited for this kind of trial. I didn't think I'd ever find an use for that program. If this idea works it should give me a stylistic advantage so it is worth trying out.

>>900267
Nice one anon I remember you had a healtry amount of detail on the assets, window and other details. That is why you should plan on the final result so don't waste time modeling thing that would never be evident on the final render. Anyways good shit anon.

>>900236
Geo nodes are more flexible. I've read that they are intended to replace the particle system in its entirety at some point in the future.

>>900266
>If you tell me the software you use I surely can find a resource video

It should be cool, that video you gave me should be enough for me to understand it because I'm using blender anyway.
Thanks dude.

>>900293
flat shading mixes badly with textures

>>899902
Started working on this absolute Chad today

I had a dream recently with a cool looking chair I wanted to recreate, in the same vein as the one in pic related.
But does anyone have an idea how to recreate these ridges? My first thought was to use curves, but I don't know how I'd make them "stick" to the cushion part of the model

>>900299
Whenever I've done those I've always

>Selected the edge
>Copied it
>Converted the copied edge to curve
>Increased the thickness

I'm by no means an expert though so you might want advice from someone more qualified.

>>900193
>>900192
I'm an animation nerd anon, are you hoping to bring your boi to life?

>>900312
To be honest, I'm not sure, especially when it comes to the timing and the framerates of each motions.

>>900299
Like this anon: https://youtu.be/XQmObEhafuU

started working on the body

>>899973
good shit

at this rate ill finish it next year

Still working on that hand. I got distracted with a bunch of other things, like working on the face, eyes, failing to learn hair particles, and creating a cheap skin texture. But then I returned to the hand. I haven't done the nails yet. And I can't attach it to the wrist yet either, because the hand has nearly 3x more edges along the wrist than the wrist itself. So I'm going to have to figure a way to reduce the edges, while retaining the shape. Which means deviating from the reference I was using. I think I know how to do it. But I'm too tired to do it now.

thought i would move away from photorealism and try toon stuff and i am kinda depressed with the result. can anyone helpp

>>900432
the reference

>>900350
Holy crap thanks for making a whole video from scratch. I'll try this topology and post results later

Working on a OSRS theme project. There is a ton of minor to medium size stuff I want to fix. The models are made in blender and the scene is created in three.js and I'm quite new to both of them. It feels like there is something major I should fix or add, but I'm not sure what it is. Most of the characters have basic animations to them. Would like to make the scene more alive, but not sure how.

>>900359
Glad you like it, I think I'll make some coom bait next :^)

>>900388
Looks a little weird, if I'm being honest. I'm not quite sure what style you're going for. When you initially posted the head, I thought she was going to be full sized. But her body is very squat. So you're going for a small cartoon style?
With her thick proportions and brown skin, the closest comparison that comes to mind is Nani from Lilo and Stitch. But that's not really a match. I'm just reaching to make sense of what I'm seeing. Her body is kind of blobby now. But might look better with clothes on. Her face however, just seems off somehow.

What I would do is lower her hair a fair amount, so her head isn't so tall. Push her hair line back slightly, lips her lips slightly, and bring out the plane of her face that goes around her eyes. I made an illustration of what I mean. Just a paint-over, so I can't say for certain it will look like this in 3D. But hopefully it gives you an idea of what I mean. After looking at her face a little longer, she strikes me as Indian. I brought her brow out a light bit more too, made her lips a wee bit larger, and painted the lower one a slight shade pinker. And gave her the suggestion of eyelids, so her eyes look like they're inside of a socket.
I hope you find this helpful.

>>900350
>>900434
Hell yea, now that's a good looking cushion. Thanks a lot Anon, modeling all the details as part of the model itself is way better than what I was trying to do before.
I'll post more progress later

>>900439
Very nice. Did you paint those shadows on the frills or is that done with AO?

>>900267
> I should be able to the brightness values as density for scattering strokes as if they were objects.
This was such a stupid idea, it ended up wasting my whole day today. What you get when you sample an image is a nosy image. Exchanging pixels for directional strokes does not mean the result will be in any way appealing or legible. Crunching the midtones is a lot more appealing than this technique. Today sucked, it felt like I was rolling nat 1s all day today. Despite the idea being simple, it was really difficult to try this out and I kept running into all sorts of weird problems.

Next up, I am going to do some studying on mattes. I am not sure where I read it, but they might be a succinct way to extract object ids from the image, so I want to investigate what they are.

>>900456
https://www.youtube.com/watch?v=-ghSlQkdxww

>>900459
I think he's talking about cryptomattes.

>>900451

All hand painted

>>900459
>>900462
https://youtu.be/onYNLY8B62I

Everything in the Matte tab of the Blender compositor is either about creating masks or doing color correction from blue/green screen spills. Cryptomattes are definitely what I want to create masks from object and material ids. In the manual glossary, mattes are literally under Matte Mask so that tells very explicitly what mattes are.

>>899902
>>900432
Think i am going to wrap it up.
I hate how depressed i get when my project looks complete ass. How do you guys cope? Any critiques or improvements also welcome

>>900478
I have no clue how to cope. Let me know if you figure out how to do that. Your weird tower looks like fruity pebbles. So that cool. Who doesn't lie fruity pebbles. A tier cereal. But for real, I think I saw you in another blender thread talking about trying to match some painter's style. Seemed rather ambitious.
Can you somehow get your fruity pebbles to mix with human architecture?

>>900432
Get rid of the fruity tower and do more detailing on the environment. In the Blender thread I saw an anon recommending that you bevel edges, but unless they are large round shapes, the line art is a substitute for that. Bevels are nothing but trouble and will significantly increase them modeling difficulty. With a boolean workflow you only run into trouble and have to clean up once you start beveling so a benefit of a cell shaded style is that you don't need to do it so much.

Looking at the ref, the main reason why your own feels so dull is due to lack of detail. That is easy to deal with using effort in 3d.

>>900479
>Seemed rather ambitious.
>Can you somehow get your fruity pebbles to mix with human architecture?
Yeah ended up merging that test into this because i need to learn oil painting and essentially just draw a bunch of different colored grass and make a fruity pebble tower.
>>900481
>That is easy to deal with using effort in 3d.
How so? I thought of doing majority of the work in blender but ended up doing 85% of the work in photoshop. i assume i can turn the patches of moss grass with help of geo nodes but what about the details/ scribbles on the walls ?

I made this in blender from scratch and i don't know how many times i have redone the topology because i couldn't get it to look right when animated. Now it finally starts to look like i'm getting the hang of it. Still some room for improvements, but now it looks like the loops are in the right places. I can't stress how important topology is for animated characters, things will just not bend at the right places when topology is not good. Good topology also prevents problems with shadow artifacts and such in subdivided renders.
Hair is actually done in about 20 minutes or so, while the topology i have worked with for months, so it kind of sets it in perspective.
I should probably start dressing her up with some different outfits, but a naked version is all i have atm as all the previous clothes i made for her dos not fit anymore.
What do you think?

The topology

>>900478
You cope by starting a new project and by focusing on the ways you've improved instead of agonizing over stuff you haven't learned/mastered yet. The fact that you're not 100% satisfied with what you're doing is a good thing, it means you can think about your own stuff critically and that's the first step to getting good.
One thing you could try is a vertical composition with lots of empty sky above your render.

>>900482
>but what about the details/ scribbles on the walls ?
Here's something that should get you started.
https://www.youtube.com/watch?v=8Odon-JrQ7o
Procedural textures and geometry are extremely powerful but kind of a pain to learn especially if you don't like math.

>>900478

What you generally lack anon is a sense of scale and perspective. What makes the reference image interesting is the planes,composition and blocking elements.

If you are trying to replicate an image that obviously has a vertical composition and don't start with your image proportions you'll never get the same result.Even if the toon shader is spot on your fron wall occupies 70% of the image so you don't leave space for the scale and backgorng elements to come in.

his is why in arcviz, environment concept art and other disciplines you HAVE to study your composition. Maybe only by rescaling and repositioning some elements you can get a far better composition.

As for coping It's just like weight loss, exercise and being less conformist is the only solution.

>>900472
My headcannon is that this is a game that comes pre-installed on a hacked DS-Videopix hybrid - like one of those minigames that came installed in the gameboy camera printer

>>900450
Cheers anon I still see the buttons a little too small.

Also try and relax a little the overall mesh. Remember that it's a cushion meaning leather wrapped around foam, not a piece of marble.

>>900482
>Yeah ended up merging that test into this because i need to learn oil painting and essentially just draw a bunch of different colored grass and make a fruity pebble tower.
I meant, is there a way to get the oil paint to blend with architecture. Like, say Skyward Sword for example. When you look at objects at a distance in SS, objects become more blobby like dabs of paint. Maybe there is a way to deform objects at a distance, so they take on the texture of the pebbles you created.

>>900397
That's a nice hand. I like the tip shape. I know it's a WIP and you probably will, but If you could narrow the hand a little by moving the pinky edge of it a little more inward it'll look nicer.

>>900483
>>900484
That is really nice topology! I'm not an expert myself but I have also recently been looking into topology more. Do you have anything specific in mind for this character or did you simply make her for fun?

>>900497
Thank you.
I found that the best way to go about when retopologizing is to figure out where you should put your loops. Then add strings of faces to all the creased or ridged curves of the anatomy. Then fill in as smooth as you can. All this while trying to use the least amount of faces.
This character is intended to become a cute and innocent girl that happens to be a crazy sociopath who likes to brutally murder everything that gets in her way. Eventually i intend to make a fighter platformer/adventure game with her as the main character.

>>900478
>. How do you guys cope?
You just have to power through. It feels like you're working really hard, putting in a lot of effort, you get fatigued and feel lost and at the end of all your hardship you feel like you haven't done shit.
That's just the process; that pain you feel is growth. You just need to persevere.

>>900502
Allow me to suggest a few changes to the face to make her a little cuter. These are just suggestions, it depends totally on your vision and of course the uniqueness of characters creates it's own appeal. These are just tips I've found to make characters as attractive as possible.
Bring the brow ridge, the nose and the mouth way forward.
Square the Jaw more (give the jaw more volume)
Lose the duck lip.
Nose. Could be a little narrower. Nasal folds smaller and the tip of the nose a little bigger.
Rotate the eyes so the outer corner is higher than the inner corner. Eyes could also afford be a little smaller.
Bigger cheek bones.
Lower hair line.

https://www.youtube.com/watch?v=wP5pwrRStbo

My first attempt at animating something.

>>900522
If it's supposed to be a ball jumping instead of bouncing that's really good. Although the last jump looks really off in the air.

>>900439
Best one in this thread you will make it unlike Cris.

>>900485
thank anon
>>900491
>his is why in arcviz, environment concept art and other disciplines you HAVE to study your composition. Maybe only by rescaling and repositioning some elements you can get a far better composition.
do you have any good resources you can recommend? I have noticed that the compositions i use are the basic ass rule of thirds and its interesting points.
>>900495
looking around other examples the effect relies on the camera's depth of field i think. the best way would be to figure out how skyward sword did it,
found a leddit thread discussing something similar
https://forums.tigsource.com/index.php?topic=22866.msg650721#msg650721
https://www.reddit.com/r/Unity3D/comments/8ub1d2/started_messing_around_with_zelda_skyward_sword/
hope this will be of use anon.
>>900513
guess its a new day, new me. thanks for the encouraging words anon
>>900522
>>900524
any particular reason you made the last platform jump very exaggerated?

>>900524
>>900522
I just wanted to respond to my post and say it's a very good first attempt even if you did mean to make it look like a stylised bouncing ball. And the specific thing really off with the last jump is it accelerating halfway through.

>>900380
i know how it feels.
dont give up, you'll feel like shit. atleast make it a worthy waste of time

>>900553
don't worry, im just lazy. I never drop a project

>>900437
You gotta replicate the text the right way anon. That pixely yellow font is crucial.

>>899974
Thanks for these they're really useful. I didn't get to apply too much of it since my deadline for this is getting really close. Also for those that make minis how far do you apply detail? I find that not much detail for my prints show through in the final product.

>>900560

Currently working on a project and decided to make Lyla for Library of Ruina.

>>900441
It is a little bit like these. I haven't hit the mark yet though.

>>900433
You can model all of the scene in 3d, but getting the wobbly hand drawn look would be difficult. Which is not say you can't get line work via freestyle edges or posterizing the image. You can get detail on the walls via sculpting. You can get the greens via texture painting or vertex painting or grease pencil. You can make the lines wobbly by coming in with the sculpt move brush and putting in some light touches here and there.

In the ref here, the stairs look pretty good, but in yours they look so flat. And indeed, they a lot less in number and just a bunch of horizontal straight lines so the scene feels a lot smaller. If you want to make the fruity tower work, you need to put some shading on it and make it feel a lot bigger. This might mean going beyond the reference and making the whole scene bigger in order to make the tower loom over the city.

>>900594
>You can get detail on the walls via sculpting.
but they wont have freestyle on them yeah?
>>900594
>You can get the greens via texture painting or vertex painting or grease pencil.
any way i can make it procedural?
>>900594
>You can make the lines wobbly by coming in with the sculpt move brush and putting in some light touches here and there.
didn't really understand this part
>>900594
>n the ref here, the stairs look pretty good, but in yours they look so flat. And indeed, they a lot less in number and just a bunch of horizontal straight lines so the scene feels a lot smaller. If you want to make the fruity tower work, you need to put some shading on it and make it feel a lot bigger. This might mean going beyond the reference and making the whole scene bigger in order to make the tower loom over the city.
lemme see if i can make it work somehow

>>900601
> but they wont have freestyle on them yeah?
It depends on the angle, you can always put in the missing lines yourself either in PS, or by marking freestyle edges.

>any way i can make it procedural?
It wouldn't be worth it for something this simple. Once you start getting into procedural art, you start getting into programming. And once you start programming a simple task that would take you 10m starts taking hours. It is only worth it if you see repetitive patterns that you need to automate.

Actually, if it is just something like painting masks and scattering leaves in a particular area that would be easy. Just split the geometry in order to get the vertices which to paint and pass them as inputs into the geo nodes before plugging into the density.

> didn't really understand this part
Unless you bevel the geometry, which you shouldn't when using a cell shading style, you'll end up with angular geometry. So in order to match the original style more closely, it might be good to introduce some deformations with sculpting.

>>900557
Thanks for the tip! Got the correct font working now

>>900485
>Here's something that should get you started.
hmm not sure it works in favour of the toonstyle but maybe adding the outlines will help
>>900605
looks like i will be painting the grass in the ps because it just looks like the faster and easier option.

>>900622
>hmm not sure it works in favour of the toonstyle
Probably not but you have the basic building blocks to make it work with your shader. Instead of plugging the cracks into the bump you can add them to your base color for example.

>>899902
working on an alleyway scene after a loooong break

>>900623
then it will just be blobs with no depth
>>900627
ian hubert way?

>>900516
Thanks for your feedback. I'm not sure i agree with all your suggestions, but i will try to make some small changes before i rig her.

>>900478
that colorful pole of mold in the middle looks very out of place imo.
I suggest you remove that and it will look alright.
May i suggest adding a wooden gate in the front of the woman. It would give the image a new meaning, but i think it could look good.

>>900516
>Rotate the eyes so the outer corner is higher than the inner corner. Eyes could also afford be a little smaller.
but this is only found in asian genes and additionally the eyes are already incredibly low on the face

>>900146
You're a pro, anon.

>>900478
I get how you feel. I’m probably projecting my own experience, but sometimes spending less time on this board helps. When I’m making something, sometimes I’m too aware of how bad it looks that I try to rush myself to make it look presentable so I can post it without getting shit on, even though I should be slowing down and working without worrying about the outside world. It can be discouraging because some people have a really bad idea of what it means to critique, but fortunately the criticism you got is useful, and if you work on this without being worried I’m sure it could look great.

>>900456
I've decided, I am going to try cultivating the NN style transfer approach.

https://deepdreamgenerator.com/

The effort to use this is so negligible and the quality is so high that I'd need to train for years just to take a month to produce by hand something a NN could in a couple of minutes. I have a lot of ML experience too, so if I had the resources I could train my own net to do this. I know it is a cheat, but I'd have killed for an advantage like this in a different domain. I am not sure how good the net would be at this at the moment, but I am sure that eventually NNs will be capable of producing anime style content in full res. Being able to just focus on modeling will free up time for me to pursue music composition for my future VN. ML based methods will only get better as time goes by.

I've really been stressed out about this for the last few days. I have a tendency to want to do things with my own two hands, but NNs were always an option. It is fitting since I'll be writing a sci-fi story about the Singularity.

My plan now will be to finish modeling the room in its full glory and then move on to the next scene. I have to get 200 likes in order to get access to full res images so that gives me something to aim for in the near term. After that I'll move on to music.

Ok, so I'll do this next, first the scan.

>>900266
Good shit, thanks for the advice.
I've been using Chief Architect Premier X10 for now.

>>900580
oh I see. Now that you post the girl's face on the left, I see the influence. You're already part way there. you just need to regard the planes of the face. specifically around the eyes and upper cheek area. And lower her hair. I think those are the two key areas that will make the most difference.

>>900628
Yes
Ian Hubert is as close to blender jesus as we'll get

>>900655
No prob anon. I truly reccomend you try archicad. It has a larger userbase and also larger forum support. Also it has support to a wide array of 3rd party render engines. If you are a student you can get one for free but you cand find cracked versions very easily on the int. Chief architect is obscureware and maybe I't ll hinder your devellopment. If it is only a hobby then I reccomend you try sketchup. It is a very potent software with TONS of libray objects and resourses. It was the main architectural modelling tool of a previous serious archviz company I had with a partner so it shouldnt be taken lightly. Independent of the software the design process cand be improved. Cheers anon.

>>900130
What program do you use for making the floorplans like in that pic?

>>900681
He literally just told he used Chief Architect Premier .

You can do the same with various software like: Revid, Archicad, VectorWorks, Autocad, Rhino, Skp and even Illustrator.

I've had architectural firms as clients that use all those software to create their floorplans. Hech I even once recieved floorplans made in excel with squares painted as walls... yes I know.

Some small advice: If you're like me, and you're modelling a figure with the X mirror modifier enabled. And then you wish to add hair, so you add a particle modifier and you notice that hair is only coming out of one side. There's an easy fix for that.
Look for an option in the particle panel that says "Use Modifier stack" and check it on. That way, the particles will actually realize the mirror is on, and act how it's supposed to act.
I suffered for days trying to figure out what's wrong. Until I finally found the option from a random commenter a from a couple years ago.

>>900450
Something really bugs me about this design. Maybe the arms aren't ergonomic enough, or the fact that the backrest isn't being supported by anything.
Designing chairs is harder than I thought

>>900736
it looks like you glued cushions to a plastic lawnchair; you're right about both of your observations though

>>900736
looks like it would break if I farted on it too hard

Will sculpting real clay help my digital sculpts? I feel like I'm wrestling with the software, the fundamentals of sculpting, and a lack of knowledge of human anatomy, and it's all just working against me. If I learn anatomy and sculpting on real clay and practice every day, will that make my digital sculpts better?

>>900634
went back to the drawing board to rework on my theme and composition. the tower thing was supposed to be a bit more saturated with my theme being life in a stalemate situation.
>>900647
>When I’m making something, sometimes I’m too aware of how bad it looks that I try to rush myself to make it look presentable so I can post it without getting shit on, even though I should be slowing down and working without worrying about the outside world. It can be discouraging because some people have a really bad idea of what it means to critique, but fortunately the criticism you got is useful, and if you work on this without being worried I’m sure it could look great.
well i think i just found the problem, i wll slow down and work on my projects for longer but i keep getting conflicted with other advice that you shouldn't have any emotional connections to your works etc. wish it was a bit more clear

>>900741
>I feel like I'm wrestling with the software
are you using a mouse or a pen tablet?

>>900743
pen tablet. It doesn't have a built-in screen. It's fairly basic.

>>900744
then its probably a practice issue, it will only get easier as time goes on

>>900736
Topology anon here. Also I am also chair nerd anon. What you lack is a general sense of ergonomy and a good study of chair construction. Don't feel bad, a good chair is the holy grail of every industrial designer, the simplest of chair designs take years of human study and testing. For example the barcelona chair is a system originally designed by mies van der rohe and perfected by knoll to combine the flexibility of stainless steel laminated plates with leather straps that support leather bound cushions at a ln angle designed for lounging. Not all chairs are designed with the same angles, and dimensions some are for work, some are for relaxation and some are even designed for eating. My overall advice anon is try to study references not pull designs out of thin air. I've seen this symptom with many artist that don' t take the time to learn and think designs can flourish without putting the work. Trust me I've been trying to design my own lounging chair for years and even after modelling tons and tons of chairs as part of my work I still have much to learn.

>>900742
Perspective anon sorry I did not respond on yourcrequest for resourcrs I've been having a full plate at work. I see you tried and fixed the image proportions. It is a good start. Keep it up.

>>900742
It’s all about filtering the criticism you receive, instead of treating it like commands. Improving the composition of your scene is good advice because it might make it more appealing to more people, but it ultimately depends on whether or not you want it to look that way. The way I filter the emotional vs. non-emotional approach is that both sides have their strengths, because attachment and emotional connection are a great source for creativity, but it can also prevent you from improving because criticism will feel more personal.

On one hand, you have Chris-chan (not our cris) type people who love their ideas so much that it blinds them to just how bad they are, and you also have others who are aware of how bad something they made is, but are in denial and react to criticism the same way a drug addict reacts to being told they’re an addict. On the other hand, you have people who avoid their emotions completely, and can create some impressive stuff, but their work is not typically mentally stimulating. For me, if I had to completely remove my emotions from the equation, there wouldn’t be a point in me making anything. It’s not about what you feel, just how you process and act on what you feel.

>>900741
Yes but you don't really need to. It's normal to feel a bit overwhelmed at the beginning.

Modeling tunes:
https://youtu.be/K2ayHWR1aFY

It's been a while. Staind. Instrumental.

Model on.

>>900649
face done, I've blocked the body too, I should have splitted the chest and the abs and maybe to add the shoulders too.

Maybe il include some text on it, maybe not... who knows..

>>900742
It does feel better like this. The buildings need some windows. How big you make them will influence the feeling of scale for the overall scene.

i made a desktop background.

started to worrkin on this big mann

>>899902
Oh my fucking God, I'm making a walk cycle, and it's good EXCEPT for the last fucking frame, which I want to be EXACTLY the same as the first frame but it won't fucking do that! I shift + D and move the keyframe to the last frame but the pose is not the same, it's as if it partially inherits some of the transforms of the frame immidiately prior to it. I swear this shit is driving me up a fucking wall.

>tl;dr: How to make exact duplicates of keyframes, without said duplicated keyframe being altered by the frame before it?

>>900793
Ok ok I fixed it, like so:
1) went to the keyframe I wanted to make an exact duplicate of
2) selected all bones > pressed G > left click to make sure ALL bones were properly keyframed
3) copy + paste the keyframe where I wanted it
It was really that simple. Thank fuck. I'm at peace with myself once more.

>>900793
copy the pose itself instead of the keyframe and dup it on the frame you want that might fix it.

is the single pose the problem or the animation? if your end frame is 24 ending it on 23 will make it loop better

>>900756
thanks anon
>>900757
thanks for the advice anon
>>900779
true true but i wish there were some holy bible for making good compositions or something, rule of thirds doesn't really help

>>900648
I think Moi might have ruined me as I am no longer capable of doing modeling in Blender. I tried modeling these railings in Blender and it I got stuck after an hour, switched to Moi and did it without a hitch. The tools in Moi are so well designed and work so well together. Sigh, maybe I have the same problem when it comes to drawing because of 3d. But it is fine, being CAD-pilled is a-ok.

These railings won't actually be visible in the room directly, the reason why I am doing it for is because I want the light to bounce off them and through the grills into the room. This happens in real life and it will add a lot of flair to the scene itself. I'll bring it into the scene and try it out. After that I don't think I'll be doing any more modeling. I'll just sculpt some messy blankets on the bed as well as bed creases and be on my way.

>>900755

>try to study references not pull designs out of thin air

Yeah I totally get that. I managed to come up with the general "vibe" I want from the chair - a living-room piece, a little less sloped than the original Barcelona Chair, and with armrests to make the it seem a little more boxed in (thus the need to make it narrower so you can actually rest your arms). But it doesn't really matter if the design clearly isn't plausible in real life, in regards to comfort and its irl production. Can't forget that form needs to work with function. I feel that the Barcelona Chair looks great because none of its design choices where done randomly, they each have a purpose even if you can't see them.
I'm not gonna stop making prototypes, but I am going to check out some industrial design literature and furniture studies. Thanks a lot for the advice anon

>>900812
Gg thats the spirit

>>900478
It doesn't look like ass, anon. You just made a misstep in your composition. 85% of it is a blank, flat wall while the star of the show and as I understand it the entire point of the art piece is obscured in the background. Just rearrange it and it'll look fantastic. The cel-shaded look already looks great.

>>900295
I actually finished this guy

>>900847
That's good. You're gonna make it.

>>900855
Thanks senpai. Means a lot. Here's another render.

>>900857
looks good, needs more samples or a denoiser

>>900871
More samples, yeh. I just wanted to do a couple quick renders to post. Upping the samples in Arnold will probably make each render take more than 30mins.

>>900876
Wow a whole 30 mins.

>>900876
Yep arnold is designed for ilm's server and nothing else.

>>900857
Very nice.

>>900661
I think I may have seen the issue. Her facial expression was not recognizable as an actual emotion. This captures her intended personality better.

got my splatmap shader working, now working on geometry nodes roots. not super happy with them, I draw them on the mesh surface and in render adaptive subdivision + displacement eats like 80% of them. not sure what to do about that beyond manually moving them.

>>900100
Looks like a 90's render
SOVL

>>900441
would

>>900811
Doing these blankets was way harder than it should have been. Here are some tips for how to do it right:

* For folded blankets, use a cloth sim to put it into place and do the higher freq details using the cloth brush.
* Quad remesh the colliders and the blankets to a small quad size before running the sim. Try to go under 1k quads next time so the self collision distance can be set higher.
* Remesh without adaptive quad sizes. Having vertices be too close will cause the cloth sim to go haywire.
* The snake hook cloth brush is great, but has low power even at 1. Try it with brush strength above 1 next time.

Had I followed this advice from the start of the day, I'd have long been done with both the blankets and the curtains. As it is, I'll have to leave curtains for tomorrow.

Also I have to report a negative on those balcony railings. It seems Luxcore is not smart enough to bounce intense indirect light from them into the room through the grills. It is a pity. I am not sure if the reason for that are bad shader settings, but it shouldn't be. Since then I've worked around it using point lights arrayed outside the door. I think I should be done with the scene with a day or two of work more.

>>900914
That's slightly better. It kind of looks like she has eyelids here, so that helps.
I still think my advice applies. But if you're happy with what you got, then I can't argue.

>>900771
today i worked a bit more on this.

I made this for a friend. Unfortunately it's exactly what he wanted.

>>901072
when's the next ratboy genius video coming out, Ryan?

>>901072
KNOCK KNOCK
I T ' S M E

Rendering this is a massive pain in the ass

>>900560
You can push the details pretty far on standard 32mm minis. I keep a little ruler which I hold up to screen then zoom out to see the model at actual scale.
I'd suggest making the things on his belt a little larger and just generally exaggerating any intersections and creases so they hold paint. In this case make the belts and chest plate thicker.

>>901016
I'd want to render this for longer, but it keeps hogging and crashing my system. At any rate, I think I've done enough modeling on the room. I am going to put in the flats and then clear my hands of the project. During the night I'll let it run for a few hours. I think that even if the display freezes, it still keep running in the background, but I do not want to wait an hour to find that out.

(1/3)

My first time doing retopo and rigging using bug dude. I guess it went ok, but it got a bit out of control in the chest area when I tried to close some loops cleanly.
I'm getting a bunch of creasing around the underarms and inner thighs when I deform it a bunch. Is that a sign of bad topology or bad weight painting or something else?

>>901171
I do not have the energy allotment to really play with deep dream. The subscription costs are such a scam. If I could get better resources I'd make it a priority to train my own net for this.

>>901173
I had to lower the style weight for these two as it was coming through too strongly. I'll try looking for a style with crisper lines next time.

>>901172
Your topo looks fine to me. Probably something to do with the weight painting around your shoulder/clavicle controllers. Try keeping it in that pose and painting the weight until there are no more creases.

>>901171
You could do many shorter renders with a different seed each time and add them together.

I'm just curious, does anyone with a shitty computer use a render farm and is it worth it? I would like to upgrade obviously, but I have a severe lack of funds. I'm just not sure if a render farm is a false economy or not.

>>901191
It's not a fake economy, but depending on how much patience you have it's better to use your own hardware. Unless you're using such high sampling that each frame takes two hours to render and you're rendering a 30 minute short film or something, you should avoid spending the cash.

>>901192
I am impatient, but the problem isn't even so much with the render time (although it does take a while even at low samples). It's after the render, even with low samples, I need to either restart my computer or have it move at a snail's pass for about 30-60 minutes. It's really annoying.
I didn't think it would really be worth investing in unless I was already making money from modeling though. I'll just wait and keep saving to upgrade my shit instead.
Thanks senpai.

>>901191
I've always had a slave computer to delegate rendering, even small tests to another machine. Even if its a shit machine, not wasting time on test and keeping the work rythm has done wonders for my time managment. Today I have my own render cluster but nothing compares to that first leap when I started to offload my rendering to another machine. It was an fx 6300 machine so any modern pc will do better.

>>901201
Fuck, I was looking for something similar to what you just posted. It was before I knew anything about rendering but was something I was interested in pursuing and someone mentioned something like a slave computer.
It had been reduced from like €1200 to €300 and I bookmarked the page, lost the bookmark, couldn't find the link again, couldn't remember what it was called so I didn't know what to search for, and ended up asking someone who insisted there was no such thing and I was talking shit.
Something like you have there would suit me down to the ground. How much would it cost to build something like that?

>>901205
Not cheap the whole thing costed me about 12k but this is a rig I bought in about 2 years.

Trust me you do not need professional hardware to make a decent cluster. My first one was about 4 amd fx 6300 machines in a cardboard box and I got to render 10 minute videos with that.

>>901216
Thanks a lot I appreciate your help. I'll look into making a cheap one instead and see how I can get on with that. It seems like the more sensible and cheaper option. It doesn't need to be anything fantastic for me, just an improvement on what I have.

>>901201
>>901216
How much noise does this generate?
And heat?
And how much power does it draw when rendering? Do you power them up specifically to render or do you leave them on always?
Does it ultimately outweigh the costs of just renting some time on a render farm?

Trying to into retro-futurism and make some more props like this one. Also fan is done, I think

Poop

>>901220
No prob anon. Just offloading to one machine will do wonders to your workflow.

>Picrel is my first render cluster back in 2013

>>901225
>How much noise does this generate?
>And heat?
A lot if you are on the same room but I have them on another room so I do not notice it.

>And how much power does it draw when rendering?
On a full workload it draws form 3k to 4k watt/hr. Or the same as a large hair dryer.

>Do you power them up specifically to render or do you leave them on always?
I only turn them on when needed.(For tests I just power one) They do not hold any info only the Synology, that i leave on all year round.

>Does it ultimately outweigh the costs of just renting some time on a render farm?
Definitely. For example a full hd 3 min architectural video fully animated takes me about 120 render hours on this cluster or 40 hours divided into 3 slaves. This will cost me about 60usd of electricity (I live in a third world country). The same render job on Rebus wich is my farm of choice can const me about 500-600 euros. So you have to have a healthy amount of work to outweigh the costs. This rig payed for itself in about 1 year.

>>901235
Also better than a render farm because you can do small tests without worrying to upload on a shitty internet.

Also I use them for not very demanding but mechanical tasks like video processing, fotogrammetry, image processing.

The lan parties are crazy at my office tho.

>>901172
Is the geo collapsing inwards sorta around his shoulder and hips? It's unavoidable losing some volume, that's what corrective blend shapes are for

>>900397
Took a break from the hand to play with some other stuff. Remade the ear. This is probably my 4th ear attempt. Though, only my 2nd finished one. I'm not sure if the topology is good. I didn't use a topology reference for this one. The shading is messy in the base form, but it subdivides nicely. I 'm thinking I'll need to add more polygons to help deepen some of the contours.

I actually did manage to reduce a lot of the polygons in the hand, on the palm side. I need to reduce more on the back side now.

>>901174
Here is v3. I am done with this scene. A style transferred version of it will show up in a VN at some point.

>>901257
(Writeup for the Deep Dream post.)

My room modeled in Blender as viewed near the entrace. By itself adding flats to the objects in the scene was not a difficult task, but I am reaching the limit of my hardware and Luxcore is often crashing during rendering. After many tries, I barely managed to squeeze out 10m out of it this time. Adding darker colors to the objects made the room a lot darker so I had to ramp up the light power as well as increase the brightness of the colors. Well, at least now that is done with. I think the colors do make the scene a bit nicer, don't you agree?

The 3d scene as is rendered directly looks fairly plastic, but looking at the style transferred result you'd never guess it was 3d. NNs are no doubt a game changer for artistic expression.

The style transfer acts both as a denoiser and detail remover. A lot of the details I spent a lot of time of working on simply aren't visible in the resulting image and it is not due to the 0.6m resolution. I am still a novice at art and the lesson I'll take from this is to anticipate this and not work on what won't show up in the final result.

Since physical based rendering is so demanding, for my next piece I might want to try rendering with Eevee or Malt. Luxcore really killed me today.

I made this almost a decade ago and just found it on an old hard drive.

>>901191
I just embraced stylized as a cope for my crappy computer

>>901265
In addition to lowering the computational burden, stylization also allows the human to work less. It is no wonder that developing a style is a priority for an artist.

>>901188
>>901173
>You could do many shorter renders with a different seed each time and add them together.
This. It's what I used to do back in the day. I'd render about 5 frames with different seeds and layer them. 1st at 100% opacity, 2nd at 50%, 3rd at 33%, 4th at 25%, and 5th at 20%. You get diminishing returns after 5 layers.
The end result actually looks pretty nice and smooth, especially compared to actual denoising.
Whether or not it takes less time is up for debate, but I've found that with some scenes throwing more samples at it just doesn't equate to a smooth image. You could render all night and have a sample count in the tens of thousands and still get noise, while you could render like 5 frames at 200 or so samples and get a cleaner result when you layer them.

>>901191
I only found them useful for really long animations that would, realistically, take me a month (or at least a few weeks) to render with my computer, if I rendered every night. It was like a 4 minute animation at 4k. I did my best to keep every frame around 30s to render on my machine. I put that shit up on SheepIt and had a bunch of other people render it in a night. I had to go in and look at every frame, since I found the Nvidia 20xx series cards always fucked up denoising, but apart from that it was pretty worth it.
For single renders though I'd say it's not. Just set it and forget it and render overnight.

>>901280
>1st at 100% opacity, 2nd at 50%, 3rd at 33%, 4th at 25%, and 5th at 20%. You get diminishing returns after 5 layers.
That's stupid, you're supposed to add every pixel and then average them.
50 renders of 100 samples is the equivalent of a single 5000 samples render.

>>901281
>That's stupid, you're supposed to add every pixel and then average them.
Not that guy, but what he is doing is a rolling moving average, so what he is doing is in fact equivalent to averaging. It is not a bad idea if I want to squeeze a bunch of 5m renders together. I'd expect the render itself to do this kind of thing though.

I haven't really looked if Blender has some kind of option to save a render every minute or so. Having this would have saved me time due to crashes.

>>901072
POTATO KNISHES

>>901067
Well, I know I have to add minor details and to finish the hair but it is mostly done.

>>901288
ass sphere looks hideous

>>901292

>>901292
That's a block out, retard. You would know that if you weren't a nomodeler tourist. Please go learn how this field works if you're going to critique people.

>>901294
>>901288
>That's a block out, retard
>I know I have to add minor details and to finish the hair but it is mostly done.
post your work

>>901295
The block out would be mostly done, retard.

>>901295
these are 2 different people, you can see my work here >>901288 and here >>899973, i post a lot in >>>/tg/ too, mostly shit like this to print.

>>901295
Yes, in sculpt that's a common workflow "draw a circle draw the rest of the owl" after merge all that you retopo and texture if you want to do animation with that, yes to do that it should be in a T/A pose but that's a model to print so that's not needed, you can sculpt that way.

made a funny face after looking in the "VR sculpting" thread
:)

>>901288
render, yes I know the plastic shader sucks.

>>901312
plastic shader is good if you're using a really tight ao.

>>900273
The sculpting phase is done. time to retopologize.

Blendlet Shenron here. No doubt this is probably way beyond my skill level but could someone just briefly describe how they'd go about making this? It doesn't look as nice with the glass material on it or at least not in anyway I tried. I wish /3/ had some kind of monthly or weekly prompt for us to all contribute towards with a theme maybe?

>>901338
With what you have there my next move would be to spend time sculpting the wave and setting up a water material for proper light transmission.
I'd add the outer layer last since it's just a basic glass material.

>>901338
Also I don't know what those blue spheres are but bubbles would be the last thing I added.

>>900818
thanks, anon but composition as a whole is really kicking my ass. is it wrong to have shapes like these in your composition? I removed the buildings from my scene because it really isn't necessary but that uneven shape is really turning me off, is being a symmetry fag bad?

>>901338
https://www.youtube.com/watch?v=n9ZNGVvMOSQ
https://www.youtube.com/watch?v=5RQLWFxAN2Y&t=191s
try these tutorials for the edges anon

>>901345

>>901345
> is it wrong to have shapes like these in your composition?
It depends on the rest of the scene. Given the ref you've showed, these stairs are give a small and shallow look to the walls and the arches. There needs to be more of them to emphasize the depth. If you'd model the original scene exactly and compare it to this one, you'd see what i mean.

>want to make wyvern
>check sketchfab
>can't find the right topology
i'm ngmi

>>901360
at that point why not just download it?

>>901360
>>901360
With that mindset your never gonna make anything.
Just start, and if its ending up like dog poop try again, or a different model. Its never wasted effort because you will still learn someting.

>>901358
are you talking about the shading?

>>901363
i want to know how to make one
>>901364
there's a lot of dragon with different topology, luckily i found what with the easiest to copy

>>900433
>>901369
Check out how many steps there are in the original image and compare that how many there are in your own model. You can make the scale of the arcs seem bigger by making the staircase steps smaller while keeping the staircase overall size the same as now. Try halving their size and doubling their number.

>>901380
Actually, I was explaining to what you had in the previous post. Here you are on the right path.

>>901259
https://youtu.be/1-26FDZM-Zg?t=678

The host is annoying, but the examples here are good. The rest is worth watching, but I wanted to highlight how good it is at drawing in particular here. Considering how many mangaka use photobashing to do backgrounds this could be quite valuable to them.

Unlike DALL-E 2 for which we are going to have to wait who knows how long to get good enough hardware to run, a 12gb GPU would be enough for this. Too bad I have an 4gb one. Nvidia RTX 3080 is only 10gb, so you'd need a 2.5k dollar RTX 3090 (24gb) to give it a try.

I do not know why it requires so much memory. Right now I am reading the paper for NNST, maybe I'll get a hint how to adapt it so it uses only 3gb.

>>901380
>>901381
any other improvements or critiques? does it feel comfy any chance? even tho the reference has shakey lines and scratches it still looks pro while mine looks very amateurish.

>>901390
there's a lot of 'texture' on the reference and most of it's coming from subtle colour variations.
take the image into an editor and run your color picker over it and see how often it changes.

>>901390
I think it is fine. It would be interesting to see what the deep dream generator comes up if you passed in the original reference. If it does the job properly it might just add those subtle changes >>901391 mentioned.

>>901391
>how often it changes.
not a lot, guess i will introduce more colors
>>901397
which style? is it art if i go about such a roundabout way of collecting reference?

>>901399
one style

>>901399
I meant that you should put your own image as the base, and your reference as the style image.

>>901370
make a proxy of it before working on the topology, start by only doing the shapes you want

>>901401
mhmmm, is it my composition then? doesn't look comfy

headsculpt I am working on the eyebrows are just placeholder, because faces without them always look strange.

>>901412
the nose needs a bit work where it connects, and the hole on the mouth's too much
but aside for that looks very good anon

>>901404
Hmmm, it seems not. Is the original image bleeding through? I don't understand why it's distorted in places like this.

I'd also be interested if you could run it with a lower style weight. I've found that for my images 30% works better than the default 50%, but you could go even lower to something like 10%. You are aiming to just add a bit of variety, not to significantly change the style.

Once I am done getting neural neighbor style transfer to run and implementing checkpointing for it, I'll try running it on your image just to compare the result. I think it should work better than the deep dream generator, but we'll see.

>>901399
>not a lot
poor child

>>901425
dosent seem to work
>>901425
>Once I am done getting neural neighbor style transfer to run and implementing checkpointing for it, I'll try running it on your image just to compare the result. I think it should work better than the deep dream generator, but we'll see.
no clue any of this means
>>901441
we all start somewhere

>>901402
manage to understand the body, is there any closest animal head with the same topology as wyvern?

>>901338
>>901342
>>901343
>>901347
The videos were great, thanks, I really should just youtube more things. Trying to make the toon glass doesn't work out too well because it needed more verts/the reflection looks awful on flat surfaces. I tried snake hooking the waves in Dyntopo but then my gradients got fucked and I wasn't sure how to fix that. I'll keep trying to figure out how to get the toon glass to work though.

>>901412
Looking good. I'm trying to be like you.

Did some small changes to nose and mouth and made her some simple casual clothes. Plan to make her some better outfits, but this will do for now. I havent textured it yet, only some vertex paint. It UV unwrapped nicely almost in one piece, but that left me very little detail on certain face areas, so i probably should unwrap the face/head on its own to make better use of the texture space.

With a little bit of AO.

>>901259
I can't run NNST on my machine yet, but here is the high res output from the web demo. II think it is a good deal better at imitating the original style than the deep dream generator.

>>901444

uguu

>>901532
looks good, you should try box modeling

i'm gonna make it, fuck the hand muscle

feet

>>901523
The algorithm is a lot more complex than vanilla style transfer, just adding checkpointing is not going to work. I am going to have to rent a cloud GPU with >12gb memory. Any recommendations?

>>901412
Really good.

You are gonna make it.

>>901561
a good start, but you should refine them further

>>901534
how many times you gonna post this crap

>>901561
based footposter

What are some quality free texture resources?

>>901644
What is the sticky?

>>901661
links from 2017

>>901662
The what is google and your fucking front lobe. Fuck you

>>901526
hmm so its color variations and a small amount of noise that gives some kinda texture to the whole image. just need to learn a more advanced toon shader and how to do chicken scratch outlines and i can pretty much do everything in blender itself but is this a style worth pursuing?

>>900273
>>901334
Is that Zeiram?

fuck the head

>>901678
Make a hole in it and i will fuck it, but i won't let that thing finger me.

>>901644
Yandex.com

>>901662
Wow!
What the fuck is that in the OP?!
>List of free resources: https://pastebin.com/cZLVnNtB
I swear that's never been there before!

>>901523
Here are some more style transfer results. This one is from the 'Domain Enhanced Arbitrary Image Style Transfer via Contrastive Learning (CAST)' paper. Unlike 'Neural Neighbor Style Transfer (NNST)' I had no problem getting this to generate 1408 x 1408 images even on my 4gb potato GPU. I'll look into getting it to run on the CPU and getting it to preserve the aspect ratio later, but so far the results are promising. Unlike the other models, it does lighter stylization.

>>901697
I have hunch that you can tell how a good a model is by how well it draws so I'll try that next. So far the claims from the paper do seem true. This is really good style transfer and it does not have the expensive optimization steps of the vanilla. It just passes a content and a style image through the generator and extracts the result in a feedforward fashion.

>>901671
You can take it as a hint as how your style deviates from the original. Here is the style transfer using CAST. It isn't perfect, if you zoom in you'll see some discolored pixels and across the top of the arc there some distortion, but I can tell what the net is intending to do. Some artist have a style where they'll paint over the lines, and the net captures this. It also does brightness variation without spilling weird large chunks of black like NNST.

So, which seems better to you?

>>901698
Here is a butterfly image in the starry sky style of Van Gogh.

I'm some sort of fucking weirdo that's been doing hardsurface and subdivision modelling so long that I can't get myself around the idea of sculpting, so I feel as though I'm kinda behind on my work. Even then, I've grown... FAIRLY adept at abusing subdivision oddities to make things pretty quickly.

>>901700
Before I post the drawing of the room, here is the style image.

>>901702
Here is the room. I find this result astonishing. I am into 3d because I like the workflow rather than the look so this is very interesting to me.

>>901701
My favorite thing is always showing the cloth seam and folding because it's a really ugly trick, but it takes all of 3 minutes to do and will look great on a lowpoly bake.

I guarantee someone will get angry at this, haha.GJ0MK

>>901704
...why is 4chan adding text into messages exactly? it didn't do that before.

>>901703
Here is the style drawing.

>>901709

>>901710
I couldn't decide which page of OPM to put in as the style, so I settled for the very first.

>>901711
Indeed, it has a lot of contrasting areas compared to the previous one. This aspect of Murata's style is easily overlooked, but it is easier to see now that the net has pointed it out to me. I'll try anime next.

>>901712
Fuzichoco is a great illustrator. He did the art for the Kenja no Deshi LN.

>>901709
>Right foot drawn in shape of left foot. Classic coomer mistake.

>>901714
I am really impressed how on the left wall it projects green shadows around the light. It is certainly an interesting aesthetic choice and Fuzichoco likes to play with coloring shadows.

>>901716
The generator for this only has 3.5m parameters and was trained for 18h on a RTX 3090. With a larger net and longer training times, I bet the results could be significantly better. DALLE 2 for example has 5b parameters.

But I am happy with this. This is the power of modern nets in action. I'll leave playing with this aside for now and study the code for it in depth for as long as needed. I'll be using the net, so it only makes sense that I grasp it enough to be able to build it myself scratch.

>>901716
>>901712
mate, this isn't even 3d anymore.

>>901716
>>901718
Yeah, I know. Let me do one more. Here is one of the backgrounds from Dohna Dohna as the style image.

>>901719
It feels similar to the Fuzichoco one. The net is not good at immitating the cell shaded look, probably because it has been trained on a lot paintings as oppose to anime illustrations.

>>901716
>>901719
>>901720
stop posting this garbage, schizo. No one cares about this shit

>>901721

>>901699
not even sure anymore. this project started as an R&D for a style test for a short film I need to make for my final year/next year. i think i will leave this as it is and move onto other styles.

>>901706
bro...
that was the captcha

>>901730
I think that style isn't something you specifically imitate. Rather you have something you want to do in mind, and you optimize your skills for what you want to do. Then you arrive at what is called a style.

>>901722
I don't think he's crabbing on you, I think he's just frustrated that you spam posted a bunch of images. If you want to post a large amount of images again, it would be better to compile them into one large image, and maybe exclude the parts that aren't as relevant.

I shall now end this thread.

>>901747
looks great, but the last webm is mine!! hahahaha!!1!

No.

>>901746
Ah come on, he totally is, after all he called me a schizo. That kind of post is not worth any reply other than a crab image. You telling me obvious crabbin' is not crabbin' already borders on crabbing. If he had been reasonable he could have told me politely to tone it down, but even so what would be the point of that? It is not like this is going to be in any way frequent, how often do you get people posting NPR content here? Also it is not like I am lowering the quality of the thread with my posts. Spreading images across multiple posts isn't going to break the board.

If he wanted somebody to get mad at, how about the guy posting the 3d dicks?

>>901741
That makes sense, guess I will try to keep on finding it

>>901588
Fuck you're right, thanks.
I've already started texturing, but it's not too late to fix this.

>>901747
>>901749
>>901751
What's the file limit on this board? Is it not 150? We're only at 137 currently.

>>901756
Yes.

>>901752
I suppose that is true, and a better way to word what I said was that he was crabbing because of the spamming. People with bully complexes tend to do that kind of thing, someone will do something they don't like, so they will insult that person in a way that has no connection to the situation because they can't always articulate what makes them mad.

>>901752
>>901758
i made this post >>901718 but i didn't call anyone a schizo

>>901759
We were talking about this guy: >>901721

same low poly char ive been working on for a while, finally got around to fixing up the loopcuts so it subdivides well


>>900353
i like him

>>900380
youre doing good anon

>>900437
a runescape scene using the ingame models but with really detailed raytracted lighting and some subtle material changes could look really neat, like a toy diorama

>>900472
love, maybe add a lil overshoot

>>900566
i like them

>>900847
love the horns

>>901412
really good stylized headsculpt

>>901747
love them, you've improved so much at modeling heads senpai good job

>>901774
>low poly

>>901775
Relax Chris

Does anyone know why render region doesn't work for me? When I press ctrl+b the box select comes up, but as soon as I drag out a selection and let go it instantly disappears. This isn't like something that used to work and now it doesn't, it's genuinely never worked for me once.
Yes I'm in cycles not eevee,

now I'm motivated

>>901776
He's not being a hater. He's pointing out that the model isn't low poly. You don't get rounded edges like that with low poly modelling.

/wip/ OP here.
I'll be putting the next thread up tomorrow.
I haven't had a lot of time lately, so it probably won't be a collage.
I'm just hoping we don't reach image limits before I get to posting it, like last time.
If anything, I'll just whip something together and keep an eye on the thread.

>>901813
>>901775

it is a low poly model i posted previously which i have worked on further and modified to be properly subdivided.

the final render --> not low poly

the model as previously shown in past (see attached) in /wip/ --> low poly.

hope this helps

Started trying to make a race of animal people for a game I'm working on. I had Dota 2's Hoodwink in mind, but I need them to be beefier, look good in plate armor and hold axes and shit like that. Otters seemed like the best fit. Did this in about an hour, will decide if I want to tweak it or start over tomorrow.

>>901840
And here's what it looked like a couple minutes in. I have no idea how people get good looking bases together fast.

>>901307
I rather love that shader though, how'd you do it?

i'm gonna make it

>>899902
going back to psx style after some toon and photorealism shit. there is something so comfy about making these

Took me longer than it should have.
But this my best attempt at a "tactical" bag. Inspired by the stuff you see on /k/!

If you got better patch ideas let me know.

Also some critique would be very welcome!

>>901842
Just solid view with a matcap. I think this is one of the default options, but you can import new ones easily. Then you can use View -> Viewport Render Image to render it.

>>901889
oh, and I set the outline colour to a grey instead of the default black

>>901774

tried to make the face an led screen
i suck at uv mapping so aaa

How do I import a specific keyframe to a different project?
I've made a simulation that I only need one frame of for a different project. I've baked the keyframe, but will I be able to append that single keyframe into the other project and then resize it as need be?
I've done a similar thing with characters before, but I've never tried it with smoke, which is what this is, so I'm not really sure how to go about it without fucking it up. When I try to append it I can only append the whole collection rather than just that keyframe that I want.

>>900736
Honestly think orange is a godawful color for a leather chair, but I'll try to make it not terrible somehow

>>901903
You can try an orange version of this
https://www.youtube.com/watch?v=a8s66wolONI

>>901905
I did actually, just think it looks weird when the cracks are too big

NEW THREAD
>>901909
>>901909
>>901909
>>901909
I'm gonna be out all day, so I figured I'd post it early just in case.

>>901853
Looking promising.

>>900380
>no cylinder guard
this is how cylinder guards were invented

Name my cowboy