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



Thread archived.
You cannot reply anymore.



File: sssDragon1_noDither.jpg (435 KB, 1257x709)
435 KB
435 KB JPG
Just rendered in Blender but couldnt dither it in GIMP. It looks horrible on the phones I tested.

Please help.
>>
Bump
>>
>>607584
I know you say you just rendered it but is this a 3d print after an acetone bath? Looks like a photo.
>>
>>607635
its just regular ss. I figure you dont render much? Why are you even on this board then?
>>
>>607584
Looks like some half digested thing barfed out
The texture I mean
>>
File: Capture.jpg (77 KB, 1251x583)
77 KB
77 KB JPG
Its a render guys. Heres different settings for scale. This is the old old old shader that came out in 2001. You can notice how its leagues above realtime ss. Just years ahead of unreal shit.

This is also free, in Blender. I dont know how cycles subsurface stacks up, but I'm testing now.
>>
>>607584
Instead of Gimp, try to convert with Photoshop (automatic dithering, IIRC) if you can, or go with Krita.

My approach would be to save the render as OpenEXR, full float bit-depth, lossless compression, and then, in Krita, lower the image to 8-bit color depth with manual dithering, and export as PNG with color profile sRGB, applied with perceptual rendering intent.

Manual dithering:
>Krita doesn't have built-in dithering currently, which means that 16 to 8bit conversions can come out a bit banded. But you can simulate it by adding a fill layer with a pattern, set this fill layer to overlay, and to 5% opacity. Then flatten the whole image and convert it to 8bit. The pattern will function as dithering giving a smoother look to gradients.
(https://docs.krita.org/Color_Managed_Workflow)

This method is appropriate:
http://polycount.com/discussion/comment/2369458/#Comment_2369458
>>
>>607648
>Just years ahead of unreal shit.
You are comparing apples to oranges.

A raytracer such as Cycles is not intended for real-time graphics. The sort of computations required for the quality you see can’t be done quickly enough for real-time, so in engines like Unreal you’ll have only approximations that look as good as technically feasible.

That being said, UE tech is state of the art in real-time, and I don't think it's at all appropriate to qualify it as “shit”.
>>
>>607653
>The sort of computations required for the quality you see can’t be done quickly enough for real-time
sure they can, just depends on how the software is written and what the hardware is
>>
>>607640
>its just regular ss
Yes, and colors, and a model, and lighting. All which make up an image. Don't try to oversimplify things. Doesn't make you look smart.
>I figure you dont render much?
You figured wrong
>>
>>607653

State of the art? It looks outdated as fuck.
>>
>>607664
>Yes, and colors, and a model, and lighting. All which make up an image. Don't try to oversimplify things. Doesn't make you look smart.
KEK

>>607665
this. It literally cant even achieve 2001 level graphics. We're talking 17 years behind guys.
>>
>>607667

Stop acting like an idiot. UE4 can't even render good looking forests, foliage always look like plastic.

Just look at Cryengine

https://www.artstation.com/artwork/6Be0O

And find something that looks as good in UE4.
>>
>>607668
This looks better than UE but still not that good
>>
>>607659
Post paper.
>>
>>607672
what?
>>
>>607648
Alright I give up on this board.

"Years ahead of unreal shit" by some idiotic anon who doesn't realize the difference between a raytracer and a game engine.

But you still need to, for some reason, call other software "shit" and outdated.

Go fuck yourself and take your ugly, half digested dragon thing with you.
>>
OP, don't ever "bump" your thread. Always ignore the first handful of posts, those are ALWAYS by children.

>>607676
>Alright I give up on this board.
Grow up Anon, or lurk moar. You aren't even talking to OP, assuming you aren't a canadian ruseman same-fagging.
>>
>>607676
stop defending ueshit.

>who doesn't realize the difference between a raytracer and a game engine.
games are never played with no delay time. There are different games. Some have more delay between moves, others less. Every engine is a potential game engine regardless of time delay between starting and stopping the render job.

It's already widely accepted knowledge that NVIDIAs AI denoising has paved the way for raytracing to be used in realtime. We are still waiting for a good implementation in blender but work is underway. This is cutting edge research anon - search 2017 ai denoise nvidia
>>
>>607682
Alright please send me one publication or article where it is anywhere near "widely accepted" that denoised raytracer will replace realtime engines within the next 5 years.

Just won't happen.

I'm also not defending anything, I'm saying you shouldn't point fingers at other software if your own "art" literally looks like shit
>>
>>607684
Anon the paper just came out in the past 3 months. Stop paper chasing - this is NVIDIA
>>
>>607674
Post a technical whitepaper, an academic paper, or something of the sort to back your claims.
>>
>>607584
What the fuck is that piece of shit.
>>
>>607690
Again, why are you caught up in academia?
>>
File: 1284033794581.jpg (20 KB, 251x251)
20 KB
20 KB JPG
>>607584

Why does subsurface scattering in Blender get fucking GREY in thinner places?
I've seen this in tons of Blender renders.
>>
File: comp_secondSET2.jpg (732 KB, 3253x1500)
732 KB
732 KB JPG
>>607696
Its from 2001.

The new renderman ss is much better, its insane.
>>
File: wtfsss.png (204 KB, 960x960)
204 KB
204 KB PNG
>>607696
You’ve picked my curiosity and now I wonder the same. Maybe it’s just that the material is so thin that the shader doesn’t “find” any light to scatter, and so we see the diffuse color, or a shadow? (The default diffuse color is gray, btw.)

Pic related, using the Principled shader.
>>
>>607704
gosh that looks horrible compared to ops ss
>>
>>607705
Yeah, I didn’t quite get the dragon shape right.
>>
>>607706
even if you did, i dont think cycles is capable
>>
File: came_buckets.jpg (30 KB, 709x492)
30 KB
30 KB JPG
>>607696
>>
>>607584
Make render with new random walk sss.
>>
>>607713
is that supposed to be sss? Thats the difference...op used a different renderer within blender as a plugin, not cycles...that was the difference. Cycles cant compete
>>
>>607714
Shit nigger, that looks good.
>>
>>607716
take it to /b/
>>
File: okaysss.png (259 KB, 960x960)
259 KB
259 KB PNG
>>607696
>>607704
Okay, so the commit notes regarding the new SSS method (>>607714) explain why:

>The main difference with actual
volume scattering is that we assume the boundaries are diffuse and that
all lighting is coming through this boundary from outside the volume.

>This [the new shader] gives much more accurate results for thin features and low density.

(https://developer.blender.org/rB0df9b2c)

I took my previous crap and re-rendered it in 2.80.4. This is much better indeed.
>>
>>607718

so in the end it wasn't me fucking up and it was actually the renderer being incapable? holy shit, and I wondered why I didn't fucking get the SSS right.
getting the newest version RIGHT FUCKING NOW.
>>
>>607704
>>607718
This is a fucking night and day difference, damn.
This is baffling like the Filmic color management change.
>>
>>607728
>This is a fucking night and day difference, damn.
looks exactly the same. Post proof, or stfu
>>
>>607729
>looks exactly the same.
Are you handicapped?
>>
>>607730
post proof.
>>
File: Capture.jpg (217 KB, 2808x1124)
217 KB
217 KB JPG
>>607732
found proof that its objectively worse.
>>
>>607729
>>607732
Not that guy, but can't you see the darker top in the first image?
>>
File: source.webm (17 KB, 480x480)
17 KB
17 KB WEBM
>>607729
>>
>>607747
looks like shit
>>
>>607748
How is it supposed to look like?
>>
>>607751
like an actual model
>>
>>607736

What the FUCK, both are horrible each in its own way.
Left one lost all detail in a huge SSS MESS while the one on the right has tons of disgusting grey in it.
>>
File: nomoresadness.jpg (168 KB, 1744x928)
168 KB
168 KB JPG
>>607752
There is no need to build a complete model for showing the behavior of a shader in just pointy shapes.
>>
>>607763
use suzanne
>>
Blender SSS isn't plug and play, you need to adjust it quite a bit; don't really feel like going into it though.
>>
>>607769
you are just lazy, cant even do ss on a stock model
>>
>>607769
I agree. Moving sliders in Principled shader is very hard.
>>
>>607764
suzanne is not even non-manifold, so it's unsuitable for SSS.




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

[Enable Mobile View / Use Mobile Site]

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