Nebulae

Started by Denis Sirenko, July 26, 2017, 07:40:59 AM

Previous topic - Next topic

Denis Sirenko

#345
Hi guys! New stuff!

This time the backgrounds do not have large nebulae, we need several images with a lower nebula content.

bg-045
[attachimg=1]

Denis Sirenko

bg-046
[attachimg=1]

Denis Sirenko

#347
bg-047
[attachimg=1]

(If you clicked on the "Last post" button, do not forget to scroll the page up, I shared three pictures in separate posts.)

sjefen

It's so damn beautiful  :-*


- Terje
ArtStation: https://www.artstation.com/royalt

AMD Ryzen Threadripper 1950X
128 GB RAM
GeForce RTX 3060 12GB

luvsmuzik

Recent shares are terrific! I tried some path tracing on your file share, but gave up after 15 hours of render...project for a long weekend, haha!

mhaze

Beautiful, incredible work.

Oshyan

Path tracing is not going to do anything different here. It basically only affects surfaces like terrain and objects, not atmosphere or clouds.

- Oshyan

Denis Sirenko

#352
Thanks, guys!

Quote from: luvsmuzik on October 28, 2018, 09:47:27 AM
Recent shares are terrific! I tried some path tracing on your file share, but gave up after 15 hours of render...project for a long weekend, haha!

Yes, it also seems to me that this is not a matter of path tracing. There is just a lot of clouds of a rather complex shape. You can try to just reduce their content. Before the render was much faster, because I rendered a flat cylindrical layer of clouds, looking up at them. And now the nebula is a big sphere that we look at from the side. This is better of course, but longer.

WAS

Perhaps some day the warp shader/redirect shader won't be as heavy as it is and it would really speed up rendering. They both super slow down displacement and clouds. A V3 with a nice setup will render decently but as soon as you warp for shapes you're looking at double the render time +. It's a little unfortunate.

Denis Sirenko

#354
Quote from: WASasquatch on October 30, 2018, 02:26:34 PM
Perhaps some day the warp shader/redirect shader won't be as heavy as it is and it would really speed up rendering. They both super slow down displacement and clouds. A V3 with a nice setup will render decently but as soon as you warp for shapes you're looking at double the render time +. It's a little unfortunate.

I agree that the warper slows down greatly. However, I am sure that the use of two fractals through a warper is a more complicated operation than the addition of two series-connected fractals. Because in case of a serial connection, you use the sum operation, and when using a warper, before reading the value of the new color value, you must first find its location relative to the current point of space (construct a vector from the target point). As a result, you get a very valuable effect - the preservation of the total amount of white color, i.e. compliance with the law of conservation of energy (in the form of matter).

On the other hand, it did not seem to me that the redirect shader works so badly. I think so because if you leave one fractal connected directly to the warper (I'm talking about the *tgc, which I shared before), the rendering speed becomes about 2.5-3 times lower, as it should have been if I could connect three fractals to the warper WITHOUT a redirect shader at all. At least in my case, I did not see a large tax on the use of the redirect shader.

I would also like to say a few words about optimizing the rendering of my *.tgс using a non-critical form simplification.

You can make a small optimization not due to the rendering settings, but at the expense of the nebula geometric form. Several elements of the node network contains opportunities for optimization.

One possibility is to use a warping fractals. You can use not three fractals on the warper through the redirect shader, but one, right on the warp shader. Three I used to avoid the appearance of explicit directional forms, which were sometimes deliberate. But sometimes they did not appear at all. This way you can get additional rendering speed by limiting the selection of good seeds. I could not go on this because I needed a lot of good seeds, a lot of differen good forms. Using only one fractal gives x2.5-3 rendering speed increase, as I already said.

However, reducing the number of fractals on the warper will require additional adjustment of these fractal. This is due to the fact that the form will also contain three times less distorted elements. This means that it will be necessary to complicate the shape of a single fractal by increasing its displacement, roughness, clumping of variations and other. This is to your taste.

The rendering speed can also decrease significantly if you use less gas (evenly distributed fog) in the spaces between the nebula dense structures. This can be done by increasing the contrast of the fractal so that pure black areas appear between the dense structures. Or using color offset. At the same time, it is necessary to ensure that the transition from black to white through gray is smooth, without any obvious lines (although in some cases they can imitate fine thin structures of gas or dust).

You can also reduce the smallest scale for a fractal. I set it up for images of 2048x2048 pixels. For smaller sizes it can be reduced.

Denis Sirenko

Hi guys! Several new backgrounds with our nebulae, I want to share.

bg-050
[attachimg=1]

Denis Sirenko

bg-051
[attachimg=1]

Denis Sirenko

bg-052
[attachimg=1]

WAS

#358
Quote from: Oshyan on October 28, 2018, 05:30:48 PM
Path tracing is not going to do anything different here. It basically only affects surfaces like terrain and objects, not atmosphere or clouds.

- Oshyan

Than again I wonder what it'd be like on a background sphere illumimated by stars and a nebula setup with clouds over that for extra magic. Than again PT seems to hate lots of clouds


That very last nebula. [052] is one of my all time favorites from you.

Dune