Author Topic: Daily builds version 2  (Read 134589 times)

2018-01-04, 13:23:44
Reply #120

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12777
  • Marcin
    • View Profile
It was explicitly called "Significantly faster rendering of shadowcatcher in compositing mode" in the changelog, so I guess that's what it does. ;)
I think adaptivity and dark areas is a slightly different topic.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2018-01-04, 13:27:05
Reply #121

sprayer

  • Active Users
  • **
  • Posts: 794
    • View Profile
but you wrote about noise in shadows, not how much faster it is now -__- also at yout picture i can see what AA now has more priority so so AA vs Gi may be different in 2.0

2018-01-06, 22:05:39
Reply #122

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8862
  • Let's move this topic, shall we?
    • View Profile
    • My Models
I find that noise level readings in 2.0 greatly differs from 1.7 Something like 18 min. vs 5 min. to arrive at 5% in the same scene. But of course 2.0 is much noisier than 1.7 at the same noise level. Was that done deliberately, or it's just consequence of some changes? To me it looks like noise level makes less and less sense with each release... :/
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2018-01-07, 02:26:30
Reply #123

sevecek

  • Former Corona Team Member
  • Active Users
  • **
  • Posts: 197
    • View Profile
I find that noise level readings in 2.0 greatly differs from 1.7 Something like 18 min. vs 5 min. to arrive at 5% in the same scene. But of course 2.0 is much noisier than 1.7 at the same noise level. Was that done deliberately, or it's just consequence of some changes? To me it looks like noise level makes less and less sense with each release... :/

It is a consequence of the changes in adaptivity, but it's definitely undesirable. I already changed the computed noise level to be (approximately) the same as in Corona 1.7, it should be in daily build soon.

2018-01-07, 09:52:00
Reply #124

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8862
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Thank you! Looking forward to the next build then :]
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2018-01-08, 16:05:20
Reply #125

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8862
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Regarding constant IR refresh on mouse move over viewports, i found that it's like completely random thing. Over the period of same session, it came from stable IR, then to constant refresh and then to stable again. Nothing major was changed in the scene, appart from hiding/unhiding objects, creating and applying materials. Corona cameras are present in the scene, but were hiden at that moment and rendering was going through perspective view.
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2018-01-08, 16:10:10
Reply #126

TomG

  • Administrator
  • Active Users
  • *****
  • Posts: 5472
    • View Profile
I had it happening in a scene with Corona Proxies, and when I removed those, it worked normally again, so that seemed to be part of it at least, not sure if that relates to what you were hiding and unhiding :)
Tom Grimes | chaos-corona.com
Product Manager | contact us

2018-01-08, 16:12:33
Reply #127

denisgo22

  • Active Users
  • **
  • Posts: 700
    • View Profile
Regarding constant IR refresh on mouse move over viewports, i found that it's like completely random thing. Over the period of same session, it came from stable IR, then to constant refresh and then to stable again. Nothing major was changed in the scene, appart from hiding/unhiding objects, creating and applying materials. Corona cameras are present in the scene, but were hiden at that moment and rendering was going through perspective view.


Strange thing--- if selecting all objects in scene with IR, some time this working correct, and after some time return to refreshing

2018-01-08, 17:03:55
Reply #128

Ryuu

  • Former Corona Team Member
  • Active Users
  • **
  • Posts: 654
  • Michal
    • View Profile
We have a fix for the restarting IR. It will be in next daily (hopefully today).

Usual disclaimer: Not sure if this fixes all the IR restarting problems or just most of them.

2018-01-09, 06:49:22
Reply #129

danio1011

  • Active Users
  • **
  • Posts: 361
    • View Profile
I was just curious, is the recent memory footprint reduction of 8 bytes per triangle a lot?  Or a little?  How much does a triangle usually use?

2018-01-09, 10:13:50
Reply #130

Ryuu

  • Former Corona Team Member
  • Active Users
  • **
  • Posts: 654
  • Michal
    • View Profile
This doesn't have a simple answer.

In internal Corona representation this means ~12% reduction in memory per triangle. But each of Embree, Corona and 3ds Max has its own representation of the geometry, so the data is unfortunately triplicated (this is something we want to address in the future). To be honest, right now I have no idea of Embree or 3ds Max memory footprint for geometry data, but assuming it's the same as Corona, this results in a net ~4% memory reduction.

That's why we thought it was more important to just say it's 8 bytes per triangle so you can immediately see that "yay, my 1G poly scene will now consume 8G less memory" :)

2018-01-09, 11:03:20
Reply #131

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8862
  • Let's move this topic, shall we?
    • View Profile
    • My Models
1G triangles scene?! Tell that to my 16G computer :]
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2018-01-09, 11:08:55
Reply #132

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12777
  • Marcin
    • View Profile
So far I checked with displacement. In both cases the initial RAM usage for the whole system was ~10GB.
In 1.7 and v2 the curve look similar at the beginning, but with 1.7 it stayed at almost 18GB, and with v2 it stayed at just a bit over 16GB.
6:8 would mean 25% lower RAM usage in this case.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2018-01-09, 12:51:46
Reply #133

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
yep, it is "1 gigabyte memory saved per 125M unique polygons". It is nothing dramatic, I am currently working on making something way better for scenes with motion blur.
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2018-01-09, 12:52:58
Reply #134

Dionysios.TS

  • Active Users
  • **
  • Posts: 766
    • View Profile
    • Evolvia Imaging
Hi all and happy new 2018!
Sorry for the silly question but I can find the post whith the new Dropbox link to be able to download the latest Daily Build 2.0....

Any clues?

Thanks!

Dionysios -