Author Topic: Corona 3.0 Displacement Artifacts  (Read 326 times)

2019-03-05, 10:09:08

CASA_MT

  • Active Users
  • **
  • Posts: 36
    • View Profile
Hi Coronauts,

in our company we just upgraded to 3.0 (including Hotfix) for 3ds Max and ran into an issue regarding the new displacement.
I tried to find a solution in the forum, but did not find any.

We always use chamfered edges on our walls in order to make them more realistic.
When using displacement, these chamfered edges will become the typical distorted geometry known from older methods of displacement.
I attached some images to show what exactly I mean.
I do not recall this issue from any project we did with 2.0.
So far this problem occured in two seperate scenes with different kinds of geometry, so it might be safe to say that it is a general problem.
Is there a quick fix for this problem?

2019-03-05, 10:13:38
Reply #1

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 5612
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Might be autobump. You can disable it in dev/debug settings and see if that helps.
I'm not Corona Team member. Everything i say, is my personal opinion only.

2019-03-05, 11:14:45
Reply #2

CASA_MT

  • Active Users
  • **
  • Posts: 36
    • View Profile
Disableing AutoBump unfortunately does not do the trick, weirdly enough it was already disabled (although I did not alter any settings in dev/debug)...
Has nobody else come across this problem? Maybe for the most part it is not really visible when displacing large irregular surfaces like brick walls without a corner or uneven muddy terrain.

2019-03-05, 11:38:51
Reply #3

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 5612
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Other thing that comes to my mind, is that your chamfered corners consists of extremelly elongated polygons. Maybe Corona subdivision algorithm doesn't like them. Could you try to add some extra horizontal lines to see if that helps.
I'm not Corona Team member. Everything i say, is my personal opinion only.

2019-03-05, 13:32:09
Reply #4

CASA_MT

  • Active Users
  • **
  • Posts: 36
    • View Profile
I added some horizontal lines and the chamfered edges turn out a little bit better... Therfor another problem appears on the right flat side of the wall, which is really strange, as it does not seem to tesselate this simple geometry properly.
You can clearly see the new edges as artefacts in the displaced geometry, this has usually never been an issue before.
(I only attach the AO-pass, as the phenomenon is most prominent there).

2019-03-06, 19:13:27
Reply #5

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 8780
  • Marcin
    • View Profile
Did it really render different in some older version of Corona?
I think it is related to the UVW mapping of this object. The texture is just stretched/squashed in that area. Is it any different if you render a diffuse texture or bump, without displacement?

2019-03-06, 23:16:59
Reply #6

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 5612
  • Let's move this topic, shall we?
    • View Profile
    • My Models
I don't think it's UVW mapping issue. I did some tests and it seems that chamfered areas are easy to solve with some subdivision, but then strange artifacts arise in previously good areas. Exactly like it's shown in #4 post in this topic. I don't recall that i saw something like that in earlier versions of Corona. It might be worth to perform some comparisons, but i don't have much time for that at the moment. The only solution that i found to deliver clean result, is to set displacement size to very low value and quadrify mesh with very low step size, but that's hardly practical.
I'm not Corona Team member. Everything i say, is my personal opinion only.

2019-03-07, 10:29:29
Reply #7

CASA_MT

  • Active Users
  • **
  • Posts: 36
    • View Profile
The wall has a simple box-mapping applied, nothing fancy there.
In fact we had a case in which a wall with a relatively simple brick-texture applied to it(and chamfered edges) worked fine in 2.0 and had to be re-done in 3.0 - we deleted the chamfers and then it was ok (although I would consider this an unsatisfiying solution).
Unfortunately I cannot show these steps, as we did not save the image due to the corrupt geometry and currently we do not have the time to switch back to 2.0 to show exactly what I mean.
At that point we still considered it an isolated case.

I do agree with romullus, as an unnecessary subdivision might be an "ok-ish" solution, but seems redundant, especially as it worked fine before.

I attached the MAX-file of the wall including the textures, maybe you can have a look at it at some time.