multiple max spotlights with projector map get mixed up by corona renderer

[hotdog]

max 2021 - corona 6 hotfix1

I got 11 max spot lights in my scene. every spot is unique and every spot has a projector light map assigned. i cant use corona native light because it is too blurred also when using full directivity of the light.

problem:
when rendering, the maps get mixed up. some light is projecting the bitmap from another light. sometimes two, sometimes more. no reasonable pattern in behavior.
the maps in the properties remain intact.
best results achieved when reassigning all maps and then hit render. next render try will all mix up again.

tried solutions - nothing worked so far:
simplified image bitmap files
simplified max texture names
simplified spot light names
randomized spotlight names
tested corona bitmaps or native max bitmap nodes

problem also occured while testing with 4 or 5 lights.

i am on a deadline here and this is pretty annoying. probably have to render final rendering 2 times and post it together :(

edit: attached a simple scene with 3 lights only. hit interactive rendering  a couple of times and you'll see the mess..

 

[hotdog]

Nobody cares to at least confirm this bug?
I know this is not top priority but nevertheless it would be nice to get some reaction.

I managed to get the right combination for the final render after abou 30 tries (simple luck IMO).
funny thing: I eventually stopped the rendering (it was good enough), then resumed and the new combination would mix itself into the existing image. Didn't even look so bad... Can't show you since it's confidential.

cheers

[maru]

Hi, sorry for the delay, and we will look into this ASAP.
Please note that if you need immediate support, it is definitely better to contact our official support at https://coronarenderer.freshdesk.com/support/tickets/new

I have one question for now:
Since this is about Max lights, have you tried the same thing with other renderer (Scanline, V-Ray, ...). It is possible it's a Max issue.

We will look into it anyway and will let you know about our findings.

[hotdog]

ok tnx

the problem does not occur with scanline, just tested.

did I forget to attach the linkled images (they are directly from corona webpage)?
so here the scene again with the 3 images.
- one scene corona
- one scene scanline

[GeorgeK]

ok tnx

the problem does not occur with scanline, just tested.

did I forget to attach the linkled images (they are directly from corona webpage)?
so here the scene again with the 3 images.
- one scene corona
- one scene scanline

Hi thank you for the scene, so this is highly likely an issue with spotlights and projection itself rather than anything to do with Corona, the issue on the scene I am sharing is reproduced with scanline alone and 3ds max bitmap/lights.

Each time you copy a light, the map projection slot still keeps "material#1". If you alter the bitmap on the copied new light (from within the material editor), you end up with two different projections and two different sources of "material#1". In short your scene has two different "material#1" bitmaps and each has a different image root, although the name and the ID of the material are the same.

The solution for this is to manually update the corresponding bitmap of each new copied light. either drag and drop from the ones you currently have in your Material editor to each new light or simply update each root bitmap path.

This seems to be a bug with 3ds max itself as the id of the bitmaps are identical not just their name tag, but the root path of the bitmap differs.

[GeorgeK]

max 2021 - corona 6 hotfix1

I got 11 max spot lights in my scene. every spot is unique and every spot has a projector light map assigned. i cant use corona native light because it is too blurred also when using full directivity of the light.

problem:
when rendering, the maps get mixed up. some light is projecting the bitmap from another light. sometimes two, sometimes more. no reasonable pattern in behavior.
the maps in the properties remain intact.
best results achieved when reassigning all maps and then hit render. next render try will all mix up again.

tried solutions - nothing worked so far:
simplified image bitmap files
simplified max texture names
simplified spot light names
randomized spotlight names
tested corona bitmaps or native max bitmap nodes

problem also occured while testing with 4 or 5 lights.

i am on a deadline here and this is pretty annoying. probably have to render final rendering 2 times and post it together :(

edit: attached a simple scene with 3 lights only. hit interactive rendering  a couple of times and you'll see the mess..

Unfortunately, I can't reproduce one part of this issue which is the bitmaps change/update per new IR session.

[hotdog]

tnx for your answer. I think I understand what you are trying to say about the identifiers.

but with my corona scene no further copying is needed to get the maps mixed up in IR and there are 3 unique maps and identifiers i think
(#2,#3,#4).

tested my scene (without further copying the lights) on another workstation. also tested max2020. same result.
are you sure you tested long enough in IR to reproduce? sometimes it takes 5 or even more tries...
attached a screen capture.

[GeorgeK]

tnx for your answer. I think I understand what you are trying to say about the identifiers.

but with my corona scene no further copying is needed to get the maps mixed up in IR and there are 3 unique maps and identifiers i think
(#2,#3,#4).

tested my scene (without further copying the lights) on another workstation. also tested max2020. same result.
are you sure you tested long enough in IR to reproduce? sometimes it takes 5 or even more tries...
attached a screen capture.

Curious, and thank you for the capture, I was unable to reproduce this no matter how many times i tried. We ll investigate this further.

[GeorgeK]

tnx for your answer. I think I understand what you are trying to say about the identifiers.

but with my corona scene no further copying is needed to get the maps mixed up in IR and there are 3 unique maps and identifiers i think
(#2,#3,#4).

tested my scene (without further copying the lights) on another workstation. also tested max2020. same result.
are you sure you tested long enough in IR to reproduce? sometimes it takes 5 or even more tries...
attached a screen capture.

Hi again, so after some further testing we've managed to reproduce this. While the bug is not directly related to Corona it did affect Corona version 6 hf1 with the random projection of textures while re-rendering/parsing the scene.

The issue has been addressed in the latest daily builds, the projector bug still remains as mentioned above but you won't be getting any randomization of the textures each time you re-render. Autodesk has also been contacted with regards to the above.

For the moment I would simply suggest you resort to using our latest daily build but do note that other things might be broken so always keep a backup of your current scene. You can find the latest daily build of Corona Version 7 here: https://www.dropbox.com/sh/mswauuv1afec8am/AAA5OKrd_QujokDM3mxQ7S40a

Best,

[hotdog]

sounds good. I'll wait for next major release, I don't need the feature at the moment. but good to know it will work next time.

what I don't understand is why the same result (at least on a single wall) could not be achieved by using a corona light with directivity set to 1.
Even if placed practically with zero distance to the lit object, the result will be blurred.

anyway. no problem. tnx

[romullus]

what I don't understand is why the same result (at least on a single wall) could not be achieved by using a corona light with directivity set to 1.
Even if placed practically with zero distance to the lit object, the result will be blurred.

That's because Corona lights don't have projector option. So called projector is when highly focused light beam goes through some coloured gel, placed in front of it and light rays get coloured by the gel. Corona light texmap is esentially when light source itself is coloured. It's like displaying picture on ipad and trying to project that picture on a wall. No matter what, you won't be able to get sharp projection that way. It would be nice if Corona light would get its own projector, but devs probably decided that suporting Max's native lights is good enough option.

[hotdog]

ok maybe I was wrong. it just takes its time.
here a test with two lights 100x100cm, one placed 0.01cm, the other 2 metres distance to wall. directivity 1. lights not wisible to cam cast no shadows etc.
300 passes.
the GI from the light that is really near looks kind of weird.

about your Ipad: I think an Ipad is not comparable with a light with directivity = 1.
:)
but I think if you place an ipad directly on a paper and make it invisible to your eye (how?) this should nearly give you a crisp "projection" :)

[romullus]

Of course you'll get sharp result if you'll press the light against the wall, but what's the practical usage of this? At this point you'd better replace the light with the plane and self illumination, which not only would be perfectly sharp, but also ultra cheap to render.

[hotdog]

... but what's the practical usage of this?

For example the wall could have a brick texture on it. I know one could then take a layered material, masking out the selfillum part, but I found out that corona does not like this very much regarding performance. Also it would mean additional mapping work.

The scene I was working on when I first discovered the map change bug was way more complex. The geometry where the images had to be projected on was generated by railclone containing translucent fabric parts where the projection even was on the backside, emiting light, but weaker. I just switched from octane to corona, so I was just thinking that after modelling I will somehow project the images on that geometry assuming it will be no big deal doing this. Just had the time now after the deadline to rethink if there would have been another option.

Anyway, the max lights did the job with a bit of a hassle because of the random changing (with a total of 11 images projected the randomizing was really not getting better and with final image resolution and longer uhd cache time - you'll imagine) :)

About the corona light directivity again:
A plane light with directivity 1 is probably very hard to find in real world if not impossible, but we are in CGI so for me the value 1 does not make sense looking at the rounded edges the light produces when further away from the wall. for me this looks like directivity 0.98 or 0.99.

Again, no complaints. just some tests and thoughts.

[romullus]

I'm not questioning the need for projector, i'm questioning your argument that it's doable with Corona light ;]

As for the directionality parameter, i could be wrong, but i think it does not make light rays to go more parallel at all, instead it acts like  barn doors light modifier and restrics the light to specified angle, but not alters physical light properties at all. Setting directionality to 1 does not mean that you will get laser like light, it means that barn doors are at 90° angle to the light's surface. It's still physically correct, with the small caveat - barn doors in Corona light are infinitely long.