Lights through glass displaying black

[Jpjapers]

Havent seen this in a few versions. Lights showing through glass as black discs?
max 2018, corona 3 with standard glass shader+ absorbtion colour.

[dzintas]

Usually this happens when you uncheck "visible in refractions" checkbox.

[Jpjapers]

Usually this happens when you uncheck "visible in refractions" checkbox.

Yes but in this instance that box is checked.

[alexyork]

I've also had this lately using a recent v4 DB although can't compare in v3/v2 for now. same kind of set up. super simple thick glass screen with a simple disc corona light visible through it - renders black. when you don't look through the glass it renders white. visible to refraction/reflection + visible in render all turned on, incl. prevent black appearance.

[Jpjapers]

I've also had this lately using a recent v4 DB although can't compare in v3/v2 for now. same kind of set up. super simple thick glass screen with a simple disc corona light visible through it - renders black. when you don't look through the glass it renders white. visible to refraction/reflection + visible in render all turned on, incl. prevent black appearance.

Yeah same exact setup it seems. I havent tried any dailies.
Ill experiment and see if i can find any cause.

[maru]

Already reported.
(internal id=223075528)

[sprayer]

In v5 still this problem exist
Is there any workaround? I can't turning off refraction for frosted glass =(

it seams this happens for ies light

[maru]

The workaround is to use two separate objects for lights:
1. The visible part, which will be visible directly, in reflections, in refractions - it does not emit light, but it is visible. Example: LightMtl with "emit light" off; rayswitch
2. The invisible part, which casts light into the scene - e.g. a Corona Light set to fully invisible

The issue here is that the "prevent black" option only works for direct visibility (not for reflections and refractions). We know about it and want to provide a fix as soon as possible.

[sprayer]

The workaround is to use two separate objects for lights:
1. The visible part, which will be visible directly, in reflections, in refractions - it does not emit light, but it is visible. Example: LightMtl with "emit light" off; rayswitch
2. The invisible part, which casts light into the scene - e.g. a Corona Light set to fully invisible

The issue here is that the "prevent black" option only works for direct visibility (not for reflections and refractions). We know about it and want to provide a fix as soon as possible.

Thank you i will follow in that way.

[jasonvdp]

Has this been resolved already? I am still having the problem with Corona 6.

[TomG]

No, it is actually a very very complex issue to resolve, so resolution is not on the cards, in the near future at least (you can check the roadmap to see what is on the cards https://trello.com/b/EfPE4kPx/corona-tentative-road-map-3ds-max ).

[Coco_74]

This still seems to be an issue, when is this going to be fixed?

[TomG]

There is no time frame for a fix, sorry.

[ferrarod]

Has this been fixed yet? I understand it's technically complex due to the previous comments, but at the same time, it seems like it should be a very standard ability for a high end rendering product.

Please advise. I'm often working on interiors with 100's of recessed lights and to duplicate those efforts every time is not the ideal workaround.

[TomG]

Still no change, sorry. It would require some fundamental reworking of some things in Corona, which of course you already have heard, and with the potential for that to break so many other things, it hasn't found a place in the scheduling so far.

[dj_buckley]

Still no change, sorry. It would require some fundamental reworking of some things in Corona, which of course you already have heard, and with the potential for that to break so many other things, it hasn't found a place in the scheduling so far.

Has any work been done to address it?  It's been nearly 6 years since the first post in this thread so just wondering if any attempt/progress has been made towards solving it?

[TomG]

A curious kind of question. It's not one of those things that can be "attempted" - you either do it, or don't, and it would be in its entirety not done a bit at a time. The cause of it in a sense is known, we know why it happens in the general principles of the algorithms Corona is built upon, so we know what it would take to fix it (in broad terms) and we know how complex and long that would be.

So this question is rather like knowing that something arises due to the foundation of a building, and so knowing that to fix it would involve closing the building and tearing a lot of the building down and then having to build it differently, and then asking "Has any attempt been made to fix it?". No - because we'd have to close the building, tear a lot of it down, and rebuild it differently. This is not just a problem in an isolated part of the code, where you can try a rewrite or workaround or two, but is down on the foundations that everything else is then built upon.

[dj_buckley]

A plan to fix it is progress :)

Just looking for transparency I guess.  Is it likely to be fixed or not?  At the minute it sounds like not.

How long do you keep telling people it's hard to fix before actually fixing it?

[TomG]

Well, it can't be branched off into a different code branch, and everyone else continues to do their work while one person tries to fix this (since this will change whatever they are working on) is what I mean by the above. So that means that for one release cycle, or possible two (that is 6 months, or a year), there would be no improvements, no bug fixes, nothing other than this one thing, and then some other things that used to work may not work, or could work very differently, and all existing scenes would likely be completely broken.

Finding the time to completely pause Corona development like that would be very hard, and in many ways more detrimental to users and to Corona itself than the benefit of this one issue being fixed - this only happens in specific situations, when the light has directionality, and there is a workaround which does take a bit of extra set up but is not impossibly hard to do.

So it comes down to that we know the issue is there, we know the impact it has on users, and we would like to fix it - and if we decide to work on this particular issue to the exclusion of all else, we will for sure let everyone know. And yes, like any issue, that may be "never" as it is not possible to do all the things, sorry. That also means if anyone asks again, we will have the same answer, and we'll be transparent and honest about it as always.