Many lights, different approaches

[maru]

Hi, I made a small comparison between different techniques of placing many lights in scene (1600 here). Surprisingly there isn't much difference between time and quality of each approach. I'd like to see some other examples of such tests or read suggestions on creating another test scene that would significantly show when to use what. :)

[Ondra]

This is because of how your scene looks like. If you would increase spacing for example 5 times, you would get more different results.

[Alex Abarca]

Hi Maru, would it be significant in your test to add walls around the model. This is interesting

[maru]

Ok, will increase spacing and will add some GI hardcore.

[agentdark45]

I do a lot of open plan hotel layout renders, often requiring hundreds of lamps / down-lights / strip lights e.t.c (rendered as night shots, so the lights are the main source of illumination) so this would be really useful for me in nailing the right setup.

[Alex Abarca]

I like to use IES lights and maintain accuracy in kelvin and lumens, but just on the foreground. The background lights can be similar to what your showing Maru.

[maru]

Ok, here comes a better (I hope) comparison.

Settings:
PT+PT, 16 samples
LSM: 1
MSI: 0 (otherwise self-illumination got killed)
filtering turned off (otherwise gradient map got really ugly)
16 passes

Observations:
-self-illumination of CoronaMtl is pretty useless in this situation (this is probably why you have a warning there)
-40x40 lights and 40x40 planes with CoronaLightMtl give virtually the same results, even fireflies are in exactly same pattern, noise caused by GI looks slightly different but is equally strong
-there are visual differences between:
1 big textured corona light vs 1 big textured coronalightmtl plane
1 big textured corona light vs 40x40 corona lights
1 big textured corona light vs 40x40 coronalightmtl planes
1 big textured coronalightmtl plane vs 40x40 corona lights
1 big textured coronalightmtl plane vs 40x40 coronalightmtl planes with no texture
Which is kind of strange but might be caused by some kind of mistake of mine.

[Ondra]

ad self-illumination: the way selfillum and MSI behaves is for a reason. You used selfillumination in a way not recommended in the warning, and set MSI to value not recommended for general rendering, it should not be surprising you got bad results. It makes no sense to duplicate functionality of corona lights with self-illumination. That is not the purpose of self-illum.

[maru]

Just for clarification: the idea of this comparison is not to produce the best result, just to compare how different settings will work.

[Ondra]

yes, but your comparison should not be "If I use this feature in completely wrong way, it does not work", but rather "If I use this feature the right way, it produces different (dimmer) results"

[maru]

Ok, understood.

[Siahpoosh]

nice , i`m also working hard for creating a complete information about this situation , the only solution for solving a scene with many lights is grouping near lights together and total of light groups should be under a value  and upper from a value too .
i will share the infos soon.