Author Topic: Light visibility with directionality  (Read 13989 times)

2016-02-11, 09:09:19

Alessandro

  • Active Users
  • **
  • Posts: 323
    • View Profile
    • DotLab Srl
This is not really a bug, but imho an incorrect way to manage light visibility.
See attached pic, using a directionality value >0 cause the light source become invisible out of the cone of directionality. Always, when I use a visible light I need to always see it, in this moment I have to add an auto illuminated surface to be sure to see light source.
Maybe there should be a flag to define if the light source is always visible with any directionality value.
My Ducati or a render with Corona.....mmm, hard question!

2016-02-11, 09:43:54
Reply #1

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8779
  • Let's move this topic, shall we?
    • View Profile
    • My Models
As you already figured it by yourself, it's not a bug, but correct behaviour. There is a methods how to overcome this if you don't like results. Try to search for solutions in these topics: https://forum.corona-renderer.com/index.php/topic,6542.0.html and https://forum.corona-renderer.com/index.php/topic,6539.0.html
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2016-02-11, 10:33:14
Reply #2

Alessandro

  • Active Users
  • **
  • Posts: 323
    • View Profile
    • DotLab Srl
Yes, exactly as I'm doing now. But it's a big annoy, I have to add a geometry, and move it with the light if I need to move the light, and so on...
I agree with Ondra, web described in IES file depends by a lot of value, maybe the most important is the shape of the light bulb and the shape of the "case" of the light. So, actually, if point of view is out of the web, light source is not visible.
But.
When I prepare conceptual render I often need this kind of solution. It could be very useful to have the capability to force light source visibility. And (Ondra, I apologize for the next words...) I think it is not thousand rows of codes :|||| (ok, I'm joking Ondra, don't be angry;) )


It's just me? If it's, ok, I'll go on with creating a lighted surface above the light.
My Ducati or a render with Corona.....mmm, hard question!

2016-02-11, 10:34:44
Reply #3

FrostKiwi

  • Active Users
  • **
  • Posts: 686
    • View Profile
    • YouTube
This is not really a bug, but imho an incorrect way to manage light visibility.
See attached pic, using a directionality value >0 cause the light source become invisible out of the cone of directionality. Always, when I use a visible light I need to always see it, in this moment I have to add an auto illuminated surface to be sure to see light source.
Maybe there should be a flag to define if the light source is always visible with any directionality value.
Rayswitch a self illum material in direct override slot, 0% performance loss, no geometry needed.
I'm 🐥 not 🥝, pls don't eat me ( ;  ;   )

2016-02-11, 11:42:43
Reply #4

Alessandro

  • Active Users
  • **
  • Posts: 323
    • View Profile
    • DotLab Srl
Rayswitch a self illum material in direct override slot, 0% performance loss, no geometry needed.

.. :| really, sorry, but I can't understand what you mean. Can you please explain me better what you suggest me to do?
My Ducati or a render with Corona.....mmm, hard question!

2016-02-11, 14:19:43
Reply #5

FrostKiwi

  • Active Users
  • **
  • Posts: 686
    • View Profile
    • YouTube
.. :| really, sorry, but I can't understand what you mean. Can you please explain me better what you suggest me to do?
The cheap solution is to apply the Corona RaySwitcher Material to the Light. I presume your light is a cricle, so get a small circle/disk and apply it. In all the slots you put a corona light mtl with the same settings as the light. At this point you have the light casting like it did before, nothing change, same look, same performance. now you apply to the direct ovveride, reflection and refraction the same lightmtl, but with 0 directionality.

What you get is a light that looks like a 0 directionality light, you see it from every angle, in reflections and refractions as well, but the light itself casts a directional light.
I'm 🐥 not 🥝, pls don't eat me ( ;  ;   )

2016-02-11, 15:10:36
Reply #6

Alessandro

  • Active Users
  • **
  • Posts: 323
    • View Profile
    • DotLab Srl
.. :| really, sorry, but I can't understand what you mean. Can you please explain me better what you suggest me to do?
The cheap solution is to apply the Corona RaySwitcher Material to the Light. I presume your light is a cricle, so get a small circle/disk and apply it. In all the slots you put a corona light mtl with the same settings as the light. At this point you have the light casting like it did before, nothing change, same look, same performance. now you apply to the direct ovveride, reflection and refraction the same lightmtl, but with 0 directionality.

What you get is a light that looks like a 0 directionality light, you see it from every angle, in reflections and refractions as well, but the light itself casts a directional light.

Thx for your long explanation, actually I was talking about a Light object, so I couldn't understand you ;)
It's all right for a surface with light material, your way works fine.
Best.
My Ducati or a render with Corona.....mmm, hard question!

2016-02-11, 23:42:58
Reply #7

FrostKiwi

  • Active Users
  • **
  • Posts: 686
    • View Profile
    • YouTube
Thx for your long explanation, actually I was talking about a Light object, so I couldn't understand you ;)
It's all right for a surface with light material, your way works fine.
Best.
Light Object and Light MTL applied to the same material are the same thing. There is no difference between CoronaLight and CoronaLight MTL, except user interface.
I'm 🐥 not 🥝, pls don't eat me ( ;  ;   )

2016-02-12, 09:45:45
Reply #8

Alessandro

  • Active Users
  • **
  • Posts: 323
    • View Profile
    • DotLab Srl
Light Object and Light MTL applied to the same material are the same thing. There is no difference between CoronaLight and CoronaLight MTL, except user interface.
Yes. And I find more usable light object interface than lighted surfaces for a lot of reasons.
My Ducati or a render with Corona.....mmm, hard question!

2016-02-12, 19:28:11
Reply #9

agentdark45

  • Active Users
  • **
  • Posts: 579
    • View Profile
Light Object and Light MTL applied to the same material are the same thing. There is no difference between CoronaLight and CoronaLight MTL, except user interface.
Yes. And I find more usable light object interface than lighted surfaces for a lot of reasons.

This.

I don't think anyone wants directional lights being dimmed no matter how physically correct the behavior may be. It would be nice to keep the light fully illuminated for all directional values as the default setting and then maybe have a tick box for those who want their directional lights to turn black.
Vray who?

2016-02-14, 12:16:33
Reply #10

karnak

  • Primary Certified Instructor
  • Active Users
  • ***
  • Posts: 76
    • View Profile
In real photography, when you want to create directional lighting, you apply for example a grid modifier to your light source.

Something like this:


When you look this light source from the side, you will see only black, because of the modifier.


Corona Academy (May 2017)

2016-02-14, 13:20:17
Reply #11

Alessandro

  • Active Users
  • **
  • Posts: 323
    • View Profile
    • DotLab Srl
In real photography, when you want to create directional lighting, you apply for example a grid modifier to your light source.

Something like this:


When you look this light source from the side, you will see only black, because of the modifier.




Of course it's true, but we can go over this kind of limits with render, why not?
As I've already say, often, most of all in conceptual renders, it's so usefull to see source place of a light, no matter its dirtectionality. I'm not talking about phisical or not phisical correct, I'm talking about having the best result with the minimal work.
And it seems I'm not alone in this .. ;)
My Ducati or a render with Corona.....mmm, hard question!

2016-02-16, 11:37:13
Reply #12

FrostKiwi

  • Active Users
  • **
  • Posts: 686
    • View Profile
    • YouTube
I'm talking about having the best result with the minimal work.
And it seems I'm not alone in this .. ;)
No you are not. Need physically correct directional light?
Do nothing.
Need did rectional light to be visible from all angles?
Sacrifice 5 seconds, rightclick rayswitcher and apply to GI, copy material with 0 directionallity to direct override.
I'm 🐥 not 🥝, pls don't eat me ( ;  ;   )

2016-02-16, 12:29:41
Reply #13

Alessandro

  • Active Users
  • **
  • Posts: 323
    • View Profile
    • DotLab Srl
I'm talking about having the best result with the minimal work.
And it seems I'm not alone in this .. ;)
No you are not. Need physically correct directional light?
Do nothing.
Need did rectional light to be visible from all angles?
Sacrifice 5 seconds, rightclick rayswitcher and apply to GI, copy material with 0 directionallity to direct override.

Of course, my dear (maybe young?) friend. But, this is your opinion. That I respect. Maybe you shoud learn to respect the other one too ;)
My Ducati or a render with Corona.....mmm, hard question!

2016-02-16, 13:30:57
Reply #14

FrostKiwi

  • Active Users
  • **
  • Posts: 686
    • View Profile
    • YouTube
Of course, my dear (maybe young?) friend. But, this is your opinion. That I respect. Maybe you shoud learn to respect the other one too ;)
Didn't mean to offend you, I just fail to see the issue.

The description of the renderer states physical accuracy and eas of use as it's primary goal - anything else is faking.
The devs included an option to do exactly that.

[Directionality is usually a lamp shade thing, or a ceiling light being in a little cave from my experience, no light is just glued to the ceiling, seeing directional lights from all angles, implies the light casts rays to other directions as well, which by definition is the opposite of directionality, in archviz we aim for realism and physically accurate lighting. And again if you want it, the option is there, including it as a check box in the light modifier means including a physically wrong option, which in turn goes a bit against what archviz stands for]
And yes, I use idirectionallity + fake to see from all angles non stop, all the time, I don't do archviz, but animation, I do respect your opinion.
I'm 🐥 not 🥝, pls don't eat me ( ;  ;   )

2016-02-16, 14:27:07
Reply #15

nauticus25

  • Active Users
  • **
  • Posts: 63
    • View Profile
I'm talking about having the best result with the minimal work.
And it seems I'm not alone in this .. ;)
No you are not. Need physically correct directional light?
Do nothing.
Need did rectional light to be visible from all angles?
Sacrifice 5 seconds, rightclick rayswitcher and apply to GI, copy material with 0 directionallity to direct override.
I am using light fixtures specified by the lighting designer.  The workflow is currently to plop in a CoronaLight, add the IES file, and set the output in lumens.  Then I add an additional object the size of the light to provide the glow to the camera.  If I could get away with just the CoronaLightMtl, that'd be great, but you can't use IES files with the material (for good reason), the material doesn't show up in the light lister, and the intensity is only in default W/(sr.m^2) units.  You can't use lumens, candelas or lux (lighting designer always specifies lumens to me).  It's usually impossible to get the directionality to match that of an IES file.  And if you want to tweak the size of the disc, you have to recalculate the conversion from lumens to W/(sr.m^2). A direct visibility intensity override for the CoronaLight object, while not physically correct, would save having to make (and keep track of) a separate object just for camera visibility.

The specific case where I would use this all the time is for recessed can lights with a diffuser lens.  To make it look physically real, you would put the CoronaLight inside the fixture, above the lens.  But the IES file already assumes the lens in the output, so to put the IES light above the lens means you're diffusing the light twice.  So you have to exclude the lens from casting shadows.  Plus, now you have to cut a hole in your ceiling to put the fixture in.  I'm as much of a perfectionist as anyone else here, but ain't nobody got time for that! :) I just want a glowing disc stuck to the ceiling.
i9-12900K @ 3.2GHz, 64GB RAM, 3090ti
Max 2024, Corona 10