Author Topic: Caustic pixelate issue on water  (Read 1510 times)

2020-01-04, 10:56:31

albertoiuav

  • Active Users
  • **
  • Posts: 9
    • View Profile
Hi all,

i'm using Corona 5 and Cinema 4d r21.
I'm testing caustics and i have to say that the result is quite impressive and precise. I activated caustic on refraction and in general settings of Corona.
But, at the moment i have strange darker square group of pixels.

The water is an high poly plane with a displace deformer.
I calculated both test for 35-40 passes.
I highlighted the problem in one of the images attached. Last one is with a plane without any displacer/bump, the probelm remains.

Hope you could help me.
Thank you in advance.


2020-01-04, 11:00:24
Reply #1

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 6525
  • Let's move this topic, shall we?
    • View Profile
    • My Models
This looks like a bug to me. I'm moving the topic to bug reporting board. It would be best if you could provide the scene for support team to investigate. https://corona-renderer.com/upload
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2020-01-04, 11:22:29
Reply #2

albertoiuav

  • Active Users
  • **
  • Posts: 9
    • View Profile
Thank you for your reply.
I've just uploaded the scene. And name file is "1578133291_Caustic-issue.zip"


Hope to hearing back from someone soon.
best

2020-01-07, 14:18:55
Reply #3

beanzvision

  • Corona Team
  • Active Users
  • ****
  • Posts: 1591
  • Bengamin
    • View Profile
    • Cormats
Hi There,

I had a look at your scene and am also getting those strange darker pixels. I was able to reduce the amount of them by disabling "generate caustics from environment" within the Corona render settings under the performance tab. We still need to investigate further to find the cause of these anomalies.


2020-01-07, 15:55:25
Reply #4

albertoiuav

  • Active Users
  • **
  • Posts: 9
    • View Profile
Thank you for your answer..so at the moment, if i understand correctly, the bug remains.
Do you have other similar cases (from other users) in which water caustics produce this kind of issues?

Hope you will find a solution for this problem.
Waiting for some great news.

Best
Alberto

2020-01-07, 16:32:07
Reply #5

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 9803
  • Marcin
    • View Profile
Is this some kind of region/crop, or a full-frame render?
Is there a difference if you use a flat plane with no thickness vs a solid box for the water geometry?

2020-01-07, 16:53:28
Reply #6

albertoiuav

  • Active Users
  • **
  • Posts: 9
    • View Profile
Hi Maru,

this is a full-frame render.
If i need to use a solid box, do i need to re-create a "negative" from the pool shape?
If yes, do i need to create it 1-2 mm smaller in order to do not have any overlapped polygons?
Because at the moment is a high dense plane that is a great solution also because i animated it with a procedural displacer.

Atacched you can find two simple sketch-sections.
At the moment i'm using a plane, do you mean i need the solid one? I will have some edge problems for sure because of the procedural diplacer animation.

Thank you

2020-01-07, 18:36:11
Reply #7

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 9803
  • Marcin
    • View Profile
It can be a huge box overlapping with everything. :) I am not even sure if this will help, just coming up with some ideas.

2020-01-07, 19:00:05
Reply #8

burnin

  • Active Users
  • **
  • Posts: 937
    • View Profile
Could HDRI be the cause? Have you checked with a different one?

as this is w/o HDRI, just sun&sky
« Last Edit: 2020-01-07, 20:26:24 by burnin »

2020-01-08, 09:52:44
Reply #9

albertoiuav

  • Active Users
  • **
  • Posts: 9
    • View Profile
Hi all,

i'm not using HDRI, only sun and sky.
Attached you can find other 4 tests.
1st and second one show the difference between a full frame render and with the same settings, only a region one.
The region has few square dark pixels...quite strange because the settings and camera are the same.

3d is a new filw and i don't have any problem. The only difference is that with the new scene i apllied a simple grey material to pool structure
Maybe something (i don't know what) went wrong with the scene.
Finally, last test is always with new scene but using a box and not a plane...but i can't see any difference.


2020-01-08, 11:50:07
Reply #10

burnin

  • Active Users
  • **
  • Posts: 937
    • View Profile
Few more Q, to mimic your scene as close as possible:

1. What are the dimensions, boundaries of your scene?
2. How many and what kind of other materials are near, which could influence the caustics?

Or better yet, tips on what and how to look...

A.
Elimination of material issues - Are materials physically plausible? Water first. If yes, then look for nearby specular, reflective, glossy mats - either way, first overwrite all to see how it behaves with everything else diffuse (~80%) but water...

B.
Elimination of geometrical or precision issues - Create a new scene. Copy stuff over, excluding everything but the part (or just with a nearby surroundings) in question. Only model what's needed. Large sizes can cause math. precision issues (alike here @ 1~2min)

C.
A must: Lighting models, lamps should be simple & light sources defined, set as close to real as possible. At least until you have enough experience to manipulate w/o getting unknown issues, which you can then consider and report as bugs.

D.
& a few more tips - "DOS & DON'TS" from Maxwell Render blog


Hopefully you'll find the culprit ;)

2020-01-08, 11:52:51
Reply #11

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 9803
  • Marcin
    • View Profile
What is the difference in water and pool structure materials between the working and squares scene?

2020-01-09, 13:42:08
Reply #12

beanzvision

  • Corona Team
  • Active Users
  • ****
  • Posts: 1591
  • Bengamin
    • View Profile
    • Cormats
I found that hiding the "Mare" object in your scene makes it render just fine. But it sounds like you have already solved it, Is that right?