Author Topic: How IR is used to be interactive  (Read 31477 times)

2019-06-27, 10:35:24
Reply #15

Frood

  • Active Users
  • **
  • Posts: 1922
    • View Profile
    • Rakete GmbH
We certainly want to address these issues.

This one is from 02/2018 for example. There has been absolutely no movement. You even do not need any special scene to reproduce, so have fun.


Good Luck



Never underestimate the power of a well placed level one spell.

2019-06-27, 13:20:46
Reply #16

felicialee

  • Users
  • *
  • Posts: 4
    • View Profile
It is indeed helpful:

I too have noticed a big slowdown in IPR, especially with Forest Pack.  I actually thought it might have been due to Forest Pack updating, particularly the big update where they made it so you could 'live edit' during IPR.

More specifically:

I notice Forest Pack point clouds slow down IPR massively.  Changing to a pyramid proxy helps a bit.

I also notice that base surfaces that are scattered on that have modifiers like FFD can really slow things down when using Forest Pack and IPR.  Collapse the base surface to an editable poly and it speeds up some

Moderator's note - removed spam links.
« Last Edit: 2020-05-27, 10:41:01 by romullus »

2019-06-27, 15:06:11
Reply #17

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12764
  • Marcin
    • View Profile
Guys, if you have scenes where IR used to work fine in an older version, and now is working worse, please:
-Explain in which exact version of Corona it used to work better
-Explain in which version it works worse
-Contact us either in this forum thread, or start a new one, or use this contact form: https://coronarenderer.freshdesk.com/support/tickets/new
We will then let you know about further steps.

Thanks in advance.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2019-06-27, 15:50:05
Reply #18

lupaz

  • Active Users
  • **
  • Posts: 954
    • View Profile
Hi Maru,

At least in my case, I don't and can't have different versions installed to actually test this. You probably do, right?
Isn't it just a matter of you testing and then telling us there are no differences or there are indeed differences?




2019-06-27, 15:58:29
Reply #19

lupaz

  • Active Users
  • **
  • Posts: 954
    • View Profile
Actually it would be really interesting and fun to see the difference between version 1.4 and forward, no?

2019-06-28, 14:39:59
Reply #20

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12764
  • Marcin
    • View Profile
Hi Maru,

At least in my case, I don't and can't have different versions installed to actually test this. You probably do, right?
Isn't it just a matter of you testing and then telling us there are no differences or there are indeed differences?

Actually it would be really interesting and fun to see the difference between version 1.4 and forward, no?

Hi Lupaz,

I never requested users to try their scenes with different versions of Corona. We only need user scenes, because the test scenes we create are usually much different than the ones users create in real production.
So if you have a scene which you see that is working super slow (or at least not as fast as expected) - let us know.
If you have a scene which you remember worked great in an older version of Corona, but it works much worse now - even better. Let us know too.

Once we have a scene, we will open it in different versions of Corona and Max, and will see if we are able to reproduce the slowdown, and what are the differences on our hardware.

There could be various issues here. Of course it could be some kind of regression in Corona. But it could as well be:
- User error
- Specific hardware / software combo issue
- Actual change in the scene (e.g. if a user forgot he changed some render setting, which made IR slower)

So if you ever spot something working incorrectly in Corona, please report it to us. It definitely should not take more than 2 minutes, and in case we need more info, we will just contact you: https://coronarenderer.freshdesk.com/support/tickets/new
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2019-06-28, 17:02:48
Reply #21

lupaz

  • Active Users
  • **
  • Posts: 954
    • View Profile
Thanks Maru.

"Super slow", "remember"...it's very subjective.

I don't think the issue people are talking about is a specific case.

I think it's just the doubt to whether you are sure or not that we are using the fastest Corona IR ever.

If you need a scene I can send you one, but if you have a scene someone sent to you a few years ago and do some testing with it with Corona 1.5 and 4.1, wouldn't that work the same way?
I see these comparisons all the time with Fstorm with that crappy kitchen scene. They had the same scene since the beginning.

2019-06-29, 00:12:43
Reply #22

Bormax

  • Active Users
  • **
  • Posts: 568
    • View Profile
One thing I noticed actually with any scene I work with. During the work on some project I bring a lot of different assets to the it's scene to try different variants of them (I suppose it's pretty common workflow). Of course, scene gets heavier and more complex because I keep many different objects in it until the final setup will be found, which slows the rendering, especially all pre-phases. It's kinda clear why it happens in such a heavy scene. But after I clean my scene keeping only the objects which are going to be involved in final rendering, rendering process doesn't get faster - pre-phases take about the same time, and it means that IR's respond doesn't get faster as well.
Simple trick I use to make IR works smoother is bring all objects from the current scene to the new empty scene. From my experience I get about 10-15% shorter final pictures render times for interior scenes, for exterior scenes difference is not so big. If we are talking about pre-phases they take 50-70% shorter time which makes IR muuuch more responsive and smoother.

2019-06-29, 12:07:02
Reply #23

Juraj

  • Active Users
  • **
  • Posts: 4761
    • View Profile
    • studio website
One thing I noticed actually with any scene I work with. During the work on some project I bring a lot of different assets to the it's scene to try different variants of them (I suppose it's pretty common workflow). Of course, scene gets heavier and more complex because I keep many different objects in it until the final setup will be found, which slows the rendering, especially all pre-phases. It's kinda clear why it happens in such a heavy scene. But after I clean my scene keeping only the objects which are going to be involved in final rendering, rendering process doesn't get faster - pre-phases take about the same time, and it means that IR's respond doesn't get faster as well.
Simple trick I use to make IR works smoother is bring all objects from the current scene to the new empty scene. From my experience I get about 10-15% shorter final pictures render times for interior scenes, for exterior scenes difference is not so big. If we are talking about pre-phases they take 50-70% shorter time which makes IR muuuch more responsive and smoother.

I wonder (Because I never tried it and measured it), do those cleaner plugins help with that at all or this is something that user cannot reach within 3dsMax, the internal "crap".

I used to do this (transfer to new clean scene) many years ago when 3dsMax 2013/14 used to have all those awful file-increasing issues, but never since. I think it would be good to try myself since I work on some scenes for months and maybe this could give them new life.
Please follow my new Instagram for latest projects, tips&tricks, short video tutorials and free models
Behance  Probably best updated portfolio of my work
lysfaere.com Please check the new stuff!

2019-06-29, 14:38:24
Reply #24

cecofuli

  • Active Users
  • **
  • Posts: 1577
    • View Profile
    • www.francescolegrenzi.com
Look at the first prototype or Corona IR (4-3 years ago).

As you can see, I think the technology behind the IR changed a lot from release to release.
At the beginning, Corona IR doesn't support the Subsampling.
and, I don't know, the old version looks more "fluid" to my eyes.
 I don' like the Subsamplig when I work, because I hate to see "huge pixel".
They distract me very much. I use it almost at 1, even if the update is not so responsive.

Honestly, I liked more the first version of IR, where, the new pixels were blended to the previous pixels.
That do you think?



t=138s


2019-07-08, 11:32:04
Reply #25

maru

  • Corona Team
  • Active Users
  • ****
  • Posts: 12764
  • Marcin
    • View Profile
Thanks everyone for the feedback. According to our initial (very simple) tests, it seems that:
- IR really used to be faster / more responsive in Corona versions from 1.0 to 1.5 (with 1.4 feeling particularly fast)
- It got a bit worse in the general "feeling" with 1.6 where subsampling was introduced (the pixelation of IR at the beginning)
- In V4 it feels pretty much the same as in 1.6
Note: this is a result of an INITIAL, SIMPLE test. We will investigate this further and will inform you about our findings.
Marcin Miodek | chaos-corona.com
3D Support Team Lead - Corona | contact us

2019-07-08, 13:52:01
Reply #26

pokoy

  • Active Users
  • **
  • Posts: 1862
    • View Profile
Hmm, subjectively I feel that IR got a tad slower, then again we asked for (and got) so much new stuff that it maybe had an impact on parsing and refresh times.

One thing that I absolutely hate is how when you are in IR and switch POV and then paint a region, all the pixels outside of the region revert back to the old IR content instead of showing the subpixel content of the new POV. This is something I didn't understand from the beginning, imo it should just display the unfinished result instead.

2019-07-08, 15:43:59
Reply #27

Nejc Kilar

  • Corona Team
  • Active Users
  • ****
  • Posts: 1251
    • View Profile
    • My personal website
Hmm, subjectively I feel that IR got a tad slower, then again we asked for (and got) so much new stuff that it maybe had an impact on parsing and refresh times.

One thing that I absolutely hate is how when you are in IR and switch POV and then paint a region, all the pixels outside of the region revert back to the old IR content instead of showing the subpixel content of the new POV. This is something I didn't understand from the beginning, imo it should just display the unfinished result instead.

Indeed, I find this a bit bothersome too. :) I manually click clear each time that happens just so I can have a better idea of what is going on in that particular camera angle...

Excited that Maru confirmed the team will look into the IR btw :) Yay!
Nejc Kilar | chaos-corona.com
Educational Content Creator | contact us

2019-07-08, 18:20:32
Reply #28

lupaz

  • Active Users
  • **
  • Posts: 954
    • View Profile
@Maru: That's great. Hopefully we'll get a boost.

2019-08-01, 15:34:46
Reply #29

lupaz

  • Active Users
  • **
  • Posts: 954
    • View Profile
Hi Maru,

I was wondering if there were more tests done.
Any further findings?

Thanks.