Chaos Corona 14 for 3ds Max - Previews (Daily Builds) Discussion

[Aram Avetisyan]

Hi, I think there might be an issue with Gaussian Splats.  I attached two render comparing Vray GS vs Corona GS.  Vray's of course looks correct, not sure what is happening with Corona.

Hi,

We have already done GS color correction improvements internally, which should fix your issue too. Hopefully next daily will have it.

[v.p.vlasenko]

"Significantly improved parsing speed, especially on CPUs with large number of threads. Geometry stage now stays consistent between renders"

Haven't noticed any improvement. Just tested 13 h1 and daily. Scene parsing in 13h1 - 1m10s, daily - 1m11s. Same.

[Aram Avetisyan]

"Significantly improved parsing speed, especially on CPUs with large number of threads. Geometry stage now stays consistent between renders"

Haven't noticed any improvement. Just tested 13 h1 and daily. Scene parsing in 13h1 - 1m10s, daily - 1m11s. Same.

It is not supposed to show up on any scene. CPU model is important too, say a 86 core EPYC is affected, an 8/16 core Ryzen will not see much difference. Try sequence rendering with large geometry computation and you will see the difference.

[v.p.vlasenko]

Tried on 2 scenes, with a lot of instances (forest pack and tyflow). One 3min, second 1:10 (this one also have hair). Should I see some difference with 18/36?

Didn't tried sequence though.

[Aram Avetisyan]

One 3min, second 1:10

You mean first render took 03:00, second render took 01:10?
Geometry calculation is part of parsing, this is where the most of the difference should come. The other stages may be the same.

There should be clear improvement in geometry computation (and effectively total parsing time) - it used to increase slightly with each re-render/sequence. Significant difference should be seen with roughly 10-15 consecutive frames. V14 DB should be consistent in this regard.

[v.p.vlasenko]

Nope, I mean two different scenes.
First render 180s for the first and 70s for the second scene.
+/- same for 13h1 and latest daily, so I didn't see any improvement here, at least for those 2 scenes.

[_fosafosa]

Hey guys, looks like Vantage link is working properly!

[aaouviz]

Something odd is happening when using CIE;

- Render a scene (with latest daily)
- Open the *.cxr file
- Load a *.conf file for LightMix (the conf file was saved from a Corona 13 version of CIE)
- Save the image
- Image lighting is saved wrong (in my case, too bright).

If I copy the image from CIE and paste it into photoshop, the image/lighting looks correct.

Not sure if this is replicable or not, but figured I should share my experience today after installing the daily.

P.S. Amazing work with the new features thus far. Absolutely thrilled to have splats in Corona (hoping the color fix is coming soon).

Cheers!

[maru]

The simpler reproduction steps are:
- save anything in .cxr format in any version of Corona
- open with v14 CIE
- save with v14 CIE

:)

This is now reported as (Report ID=CMAX-1864).
Thank you!

[mferster]

Did some tests with the Gaussian splat object.

It feels like something is off with how the colours are being represented when brought in compared to when the same splat is viewed in an online viewer like poly.cam.

It kinda feels like the gamma is off perhaps. There is a distinct lack of contrast and some green and red splats show up in what should otherwise be gray.

I am using Gamma 2.2 colour management and default settings in the VFB. the Color space options in the  splat object modifier panel don't seem to make any meaningful changes.

Edit: WHOOPS, saw that there was another post reporting the same issue. Ahwell I will keep this up showing its not an isolated problem.

[Aram Avetisyan]

Internally, not only the color interpretation is better/fixed, but there is better color correction control for splats. Stay tuned for upcoming previews (daily builds).