[ 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: 444 KB, 1024x1024, ropponginew.png [View same] [iqdb] [saucenao] [google]
520159 No.520159 [Reply] [Original]

I'm pretty new to 3D and this is my first "big project" in blender, a small part of a city inspired by Roppongi Hills in Tokyo.

I'm still learning how to render scenes better, so I would like some tips on it, I'm currently planning to have street with red-ish lights, and tall buildings with blue ones to have some contrast.

Also, I'd like some opinions on it.

>> No.520160

Maybe a light night sky to cut out the skyscraper.

>> No.520168
File: 716 KB, 1024x1024, nightlightskypng.png [View same] [iqdb] [saucenao] [google]
520168

>>520160
You mean something like this? It looks nice

>> No.520171

this looks like blender internal

why would you use blender internal

>> No.520173
File: 739 KB, 1024x1024, render100.png [View same] [iqdb] [saucenao] [google]
520173

>>520171
As I said, i'm pretty new and I didn't know about Blender Cycles until now.

I rendered using Cycles, and it was the worst experience I ever had. The image was completely noisy, so I changed it to render 100 times each pixel. It took me a century to render it all in a I5 with 6GB of RAM, and it still came out noisy.

>> No.520180

>>520173
thats because you are doing it wrong

you got to use luxrender

>> No.520185

>>520180
dont listen to this guy.

I swear, in every fucking thread there's one person that says to use that shit.

Even on topics that don't even coincide with rendering.

OP, just learn cycles for now.

>> No.520186

>>520173
You really need to learn cycles

If you set it to GPU rendering it's gonna be even faster than internal AND look way better

>> No.520187

>>520180
New guy here

Why should I use Luxrender over cycles?

>> No.520188

the reason you should use a modern renderer like Cycles is that most global illumination problems will be taken care of by the renderer. the lighting set up is more natural and makes sense.

with blender internal, you need so many hacks to simulate bounced lights it's not worth the effort. learn Cycles, it's pretty easy anyway and you won't regret it when you see the results.

>> No.520190

>>520186
is this actually true?

>> No.520191

>>520186
I'm using an on-board graphics card, so I don't think GPU rendering will be actually better.

>> No.520192

>>520191
Get a decent GPU m8 ...invest or get left behind....

>> No.520193

>>520192
I got a 2007 computer, if I wanted a new GPU I would need to buy a new motherboard too. I'm saving money to buy a completely new computer, but that's going to take months

>> No.520194

>>520193
get a job you idiot

>> No.520195

>>520194
I'm not a wagecuck

Plus, i'm in college

>> No.520196

>>520193
Can get in the game for under a g / do ur homework and start looking for deals

>> No.520204
File: 121 KB, 240x249, 1226472285662.png [View same] [iqdb] [saucenao] [google]
520204

>>520191
>doing 3d with on-board graphics card
well, this settles it

>> No.520206

>>520195
I had a job in college plus an internship. At the same time. Now I'm happily employed in aerospace.

>> No.520207

>>520204
I used to use a great off-board graphics card, but it burned a few months ago and i'm poor.

>>520206
I don't even know how I study, I can't even imagine doing anything else at the same time

>> No.520208

>>520185
>>520187
theres a reason we say use lux, because its superior.
its much easier to use and you have WAAAY more control than cycles.

also having a part time job is a good thing. its not that hard to manage really.

>> No.520209

>>520208
The problem isn't it being hard to manage. It's that I will lose my precious free time

>> No.520211

>>520209
like i said, i had a job and an internship

>> No.520214

>>520211
You are you and I am me

>> No.520215

>>520209
>free time
so you can jack off?
priorities man

>> No.520216

>>520168
more light, the roughness its the same in all the materials, if you can use vray for blender

>> No.520276

>>520159
>you need Luxrender it's better
>No, Vray is the best
>No Renderman is far superior.
>No, the best is MegaMeme render??

I get get sick to death of this in every thread.

Cycles is perfectly good for the job but you will need to turn down the default settings. Learn how to use it, you will get far better results for a still than BI.

>> No.520290

>>520276
Andrew Price please leave.

>> No.520306

>>520276
Cycles cant even touch those renderers

>> No.520317

>>520290
>>520306
I can bet you - that you haven't a clue what you are talking about!

My name is listed on more box office movie credits, more times than you have sold . . .well anything!

Please tell me more about your knowledge and experience?

>> No.520318

>>520317
>implying we'd fall for such obvious bait.

>> No.520320

>>520318
Up to you.
But if you think for one minute that the highest Grossing films involving CGI is because of Autodesk or a few low level rendering suits, then bigger fool you.

>> No.520322

use more directional light

it reminds me alot of unreal 2 maps, i like it
it looks pretty clean because there is no sense of weather
if you want the floor to be watery you would need bump mapping to highlight those areas, in general bump mapping would help alot
so far the composition is not bad, keep going

>> No.520327
File: 698 KB, 1024x2048, roppongifinaledited.png [View same] [iqdb] [saucenao] [google]
520327

What about now?

>> No.520331

>>520327
for what purpose this render?
i understand the shadowy and ambient light, but what's the selling point of it? if its a game environment do it in real time so you could actually rotate the model and see how the lightning affects the environment
but since its a render, you would want to get more out of your image which is proper lightning composition, displacement and post effects that will make the render appealing.
i suggest you watch this video
https://www.youtube.com/watch?v=O8i7OKbWmRM
https://www.youtube.com/watch?v=m-N149FMlWk

>> No.520333

>>520327
that's a penis

>> No.520412

>muh cycles can't even touch muh renderman

Cycles is far superior for hobbyists on a single computer

People actually suggesting full on professional CPU renderers to a guy learning blender are fucking retarded

>> No.520413 [DELETED] 

>>520412
cycles is slower than gpu renderers - arnold comes to mind. Its definetely slower than PRMan, too.

cycles is a hobbyist project by wishful hobbyist tier coders

>> No.520415

>>520412
>renderman
>>520412
cycles is slower than cpu renderers - arnold comes to mind. Its definetely slower than PRMan, too.

>> No.520417

>>520415
What the fuck are you even talking about, cycles on a GPU is 5 to 10 times faster than all "professional" CPU renderers with the same featureset enabled

>> No.520418
File: 46 KB, 625x390, 79dda10fcac2f5a38521b657f824173f08a68bb11d800021990bab390b8f35a0.jpg [View same] [iqdb] [saucenao] [google]
520418

>>520417

>> No.520420
File: 561 KB, 960x540, cycles.png [View same] [iqdb] [saucenao] [google]
520420

>>520418
Show me a CPU renderer not running on an average hobbyist computer that's able to deliver a better result in less than 45 seconds, including caustics and all

>> No.520421

>>520420
>not

Why did I put that there

>> No.520422
File: 480 KB, 493x342, 1451857612647.gif [View same] [iqdb] [saucenao] [google]
520422

>>520420

>> No.520437 [DELETED] 

DING DING DING! SPEEDTEST TIME!

>>520420 claims that Cycles can beat any renderer on a hobbyist toaster with CPU rendering. Let's test that. Reminder that this is for speed and quality. but mostly speed. So not all parameters are going to be the same. Like go-karting, you race for speed not time. Pic related is my specs. I will post the finished results as the finish. The finished samples per pixel will be 512SPP.

>> No.520438
File: 49 KB, 673x452, Capture.jpg [View same] [iqdb] [saucenao] [google]
520438

DING DING DING! SPEEDTEST TIME!

>>520420 claims that Cycles can beat any renderer on a hobbyist toaster with CPU rendering. Let's test that. Reminder that this is for speed and quality. but mostly speed. So not all parameters are going to be the same. Like go-karting, you race for speed not time. Pic related is my specs. I will post the finished results as the finish. The finished samples per pixel will be 512SPP.

>> No.520439

>>520438
Also forgot to mention that the renderers against Cycles will be
>Mitsuba
>Luxrender
>Yafaray
>Renderman

>> No.520441
File: 867 KB, 810x960, Mitsuba_cornellbox.png [View same] [iqdb] [saucenao] [google]
520441

First up is Mitsuba.

Also I am a lazy bastard and don't feel like waiting around for 512samples so I lowered it to 256SPP.

>> No.520442

>>520441
>not smoothing the teapot

>> No.520447
File: 1.29 MB, 810x960, Luxrender_cornell_box.png [View same] [iqdb] [saucenao] [google]
520447

Up next is Luxrender. Sorry this will take a while I might be playing video games between each image.

So far Mitsuba gives cleaner results (asside from those blotches) but that may be because I used Path Space MLT. I'm not sure what Lux uses. Probably Primary Sample Space MLT.

Also quick side note I might not be able to add renderman in here if I can't figure out how to stamp the render stats in the final image.


>>520442
I left it faceted to try and showcase reflected caustics.

>> No.520449

All those renderers just wish they were Arnold.

>> No.520450

>>520441
>>520447

The cycles scene that was posted is infinitely harder to render than that scene though. 3/4th of the cycles ground-plane can only be reached by tracing a path through a refractive glass-y material

>> No.520451
File: 1.62 MB, 810x960, yafaray_cornell_box.png [View same] [iqdb] [saucenao] [google]
520451

>>520447
Yafaray, 256SPP.

Yafaray looks pretty clean but is lacking in caustics and noisier. However is a bit faster than Lux and Mitsuba.

>> No.520452

>>520441
>>520447
>>520451

upload the blend and I'll run it through cycles on a 780

>> No.520453

>>520450
I am aware of that and I was initially going to do a heavily distorted plane to show caustics but like I said I'm a lazy, impatient bastard. And that is only because this test don't prove shit or change any opinions. Especially on this "literally who" of a 3DCG imageboard.

>> No.520454

>>520452
That guy said CPU. Not GPU.

>> No.520455

>>520454
It'd still be interesting to compare it to GPU rendering

>> No.520456

>>520455
Okay then. What is a site that let's me upload .blend filed? besides blendswap.

>> No.520458
File: 1.39 MB, 810x960, Cycles_cornell_box.png [View same] [iqdb] [saucenao] [google]
520458

Here is the Cycles test. Fastest time with the cleanest output. But lowest quality. No no reflective caustics, filter glossy isn't accurate looking. If you want to get into the nitpicky details you could say Luxrender has the cleanest output if you don't count all the clumps of unsampled caustics.

Sometimes I'm honestly fooled into thinking that Cycles is actually a biased renderer.

>> No.520460
File: 24 KB, 343x326, Screen Shot 2016-04-23 at 8.58.13 PM.png [View same] [iqdb] [saucenao] [google]
520460

>>520447
if you left it at default it would be bidirectional

lux has a (as pic shows) a lot of diffrent types

>> No.520461

>>520456
mega should do

>> No.520462

>>520460
I know but I did want to keep it unidirectional.

Also I used the non-OpenCL Luxcore path.

>> No.520463

>>520461
http://www.mediafire.com/download/t82lwqo44nq8r57/untitled.blend

How fucking lazy are you to not make a fucking cornell box?

>> No.520486

>>520458
>User\Nicholas
is this nicholas fedorov?

>> No.520730

hdrsky.hdr get, make texture, world texture set horizon and what it was and real both but ambient

>> No.520763

>>520159
Try to up the render samples in the render tab.