Corona Renderer Forum

General Category => Gallery => Topic started by: MatEvil on 2013-03-19, 15:25:42

Title: Corona Alpha 4 animation test
Post by: MatEvil on 2013-03-19, 15:25:42
My last test with Alpha 4, maybe I must increase the time per frame, now is  1 minute per frame...



Matteo
Title: Re: Corona Alpha 4 animation test
Post by: jjaz82 on 2013-03-19, 16:53:20
Great Mat :)
Title: Re: Corona Alpha 4 animation test
Post by: Ludvik Koutny on 2013-03-19, 17:05:22
The settings you use do not make any sense :)
Title: Re: Corona Alpha 4 animation test
Post by: Polymax on 2013-03-19, 17:25:19
I agree with Rawalanche, but still looks good :)
Title: Re: Corona Alpha 4 animation test
Post by: michaltimko on 2013-03-19, 21:19:11
Funky settings man, but anim looks promising ;)
Try to improve it
Title: Re: Corona Alpha 4 animation test
Post by: MatEvil on 2013-03-20, 10:47:16
The settings you use do not make any sense :)

Sorry... maybe I don't understand well the engine, can you help me to understand it? so...

I looking the interactive doc on line and old guide

use combination Path Tracing and HD Cache

"Path Tracing is brute force per-pixel global illumination method which takes a bit longer to converge to a clean state, but rewards you with very high accuracy and detail of indirect illumination. It is currently recommended primary GI solution and thus set as default."



"PT Samples

Amount of rays traced per sample. Higher count will result into slower convergence warm up but possibly faster convergence in a long run, while lower amount will give you fast feedback but slower convergence in a long run. Lower PT samples value will also increase amount of samples dedicated to antialiasing and depth of field.
 
For quick previews, try to use value of 4. For final renderings, value of 32 will do. For deeper areas with small light entrances, increasing value further to about 64 might pay off. In most cases, value of 16 works fine, providing even distribution of sampling."


So, for give more quality I need to more samples (?) it's similar to the subdivision of V-Ray with brute Force or Samples of Maxwell renderer ?



"HD Cache is quite funny name for radiance cache (has nothing to do with High-Definition ;) ). This solution is fairly close to Light Cache or Harmonics GI you can encounter in Vray renderer and finalRender. It is an efficient interpolated solution which samples surrounding environment  and lighting adaptively from the active camera. In combination with Path Tracing as primary GI solution can delivery very accurate results at very reasonable time."

For don't used the complete Undiased way I apply the secondary solver HD Cache

Precomp step (HD cache precomputation spacing)

This value sets the sapcing between HD Cache samples. Lower value will make samples more dense, higher will make them more sparse.

Use value of 10 for quick previews, to have very fast precomp, phase, and value of 1 for final rendering.


precomp step =  precomp density right? so... more space (10 for example) give a low details, but if I give a small space (0,5 ecc... under the 1 for example) I can give more detail because I give more density (if I understand I give a more small the "grid" between the point so I have plus information)....

PT Samples

This value defines quality of HD Cache samples. Increasing this value increases accuracy of HD Cache samples and can eliminate flickering in animation


increases accuracy = up more details and I can reduce the flickering... so I have more element in my scene is moved more fast and more around so it could generate more flickering (also because is moving object is not the simple flythroug).

What I don't understand? Where I wrong?

Thank you for now and for your time

Matteo
Title: Re: Corona Alpha 4 animation test
Post by: MatEvil on 2013-03-20, 11:17:02
and this the old test of tonight...


Mat

p.s: yes... have a wrong setting....
Title: Re: Corona Alpha 4 animation test
Post by: Javadevil on 2013-03-20, 12:27:03
Matteo,  just use PT+PT and reduce your PT samples.  You use some crazy settings :-)
Title: Re: Corona Alpha 4 animation test
Post by: MatEvil on 2013-03-20, 12:33:28
Matteo,  just use PT+PT and reduce your PT samples.  You use some crazy settings :-)

Ok, but someone can reaply about thie message?


The settings you use do not make any sense :)

Sorry... maybe I don't understand well the engine, can you help me to understand it? so...

I looking the interactive doc on line and old guide

use combination Path Tracing and HD Cache

"Path Tracing is brute force per-pixel global illumination method which takes a bit longer to converge to a clean state, but rewards you with very high accuracy and detail of indirect illumination. It is currently recommended primary GI solution and thus set as default."



"PT Samples

Amount of rays traced per sample. Higher count will result into slower convergence warm up but possibly faster convergence in a long run, while lower amount will give you fast feedback but slower convergence in a long run. Lower PT samples value will also increase amount of samples dedicated to antialiasing and depth of field.
 
For quick previews, try to use value of 4. For final renderings, value of 32 will do. For deeper areas with small light entrances, increasing value further to about 64 might pay off. In most cases, value of 16 works fine, providing even distribution of sampling."


So, for give more quality I need to more samples (?) it's similar to the subdivision of V-Ray with brute Force or Samples of Maxwell renderer ?



"HD Cache is quite funny name for radiance cache (has nothing to do with High-Definition ;) ). This solution is fairly close to Light Cache or Harmonics GI you can encounter in Vray renderer and finalRender. It is an efficient interpolated solution which samples surrounding environment  and lighting adaptively from the active camera. In combination with Path Tracing as primary GI solution can delivery very accurate results at very reasonable time."

For don't used the complete Undiased way I apply the secondary solver HD Cache

Precomp step (HD cache precomputation spacing)

This value sets the sapcing between HD Cache samples. Lower value will make samples more dense, higher will make them more sparse.

Use value of 10 for quick previews, to have very fast precomp, phase, and value of 1 for final rendering.


precomp step =  precomp density right? so... more space (10 for example) give a low details, but if I give a small space (0,5 ecc... under the 1 for example) I can give more detail because I give more density (if I understand I give a more small the "grid" between the point so I have plus information)....

PT Samples

This value defines quality of HD Cache samples. Increasing this value increases accuracy of HD Cache samples and can eliminate flickering in animation


increases accuracy = up more details and I can reduce the flickering... so I have more element in my scene is moved more fast and more around so it could generate more flickering (also because is moving object is not the simple flythroug).

What I don't understand? Where I wrong?

Thank you for now and for your time

Matteo

Mat
Title: Re: Corona Alpha 4 animation test
Post by: Ondra on 2013-03-20, 12:49:32
First of all, I would not use HD cache in this configuration. There is too little GI going on to justify its overhead. Use PT+PT instead
Second, higher samples value indeed makes the GI/direct noise go away faster, but it is not free. It is at cost of the Depth of field/Motion blur/aliasing noise going away slower. In the animation you have posted, the motion blur noise will be a bigger issue, even though it is hard to judge from the video, I would need a series of subsequent standalone images.

As for the settings being "crazy", I don't think it is necessarily true in all scenarios. If it works, then it is ok ;)
Title: Re: Corona Alpha 4 animation test
Post by: MatEvil on 2013-03-20, 13:06:22
First of all, I would not use HD cache in this configuration. There is too little GI going on to justify its overhead. Use PT+PT instead

ok, I try with PT+PT

Second, higher samples value indeed makes the GI/direct noise go away faster, but it is not free. It is at cost of the Depth of field/Motion blur/aliasing noise going away slower. In the animation you have posted, the motion blur noise will be a bigger issue, even though it is hard to judge from the video, I would need a series of subsequent standalone images.

I hope con help you...

(http://www.art2upz.com/__TEMP/Download/test_motion_blur_corona.jpg)
Title: Re: Corona Alpha 4 animation test
Post by: Ondra on 2013-03-20, 13:07:35
In the last picture I see no noise at all, becase there is no motion blur going on. So I would turn PT samples down to about 4 to get better motion blur
Title: Re: Corona Alpha 4 animation test
Post by: MatEvil on 2013-03-20, 13:10:36
In the last picture I see no noise at all, becase there is no motion blur going on. So I would turn PT samples down to about 4 to get better motion blur

Last image is the last frame so no motion blur :)

If you want I can send you the file of Max 2012 in private... (and i you can do some test).


I do a test tonight when I go to home...

Matteo
Title: Re: Corona Alpha 4 animation test
Post by: Ondra on 2013-03-20, 13:21:30
Here is my private upload thingy: http://corona-renderer.com/upload
Title: Re: Corona Alpha 4 animation test
Post by: MatEvil on 2013-03-20, 13:48:43
I send it ;)