Page MenuHome

Cycles, inconsistent power units between mesh light and rectangle area light
Confirmed, NormalPublicKNOWN ISSUE

Description

System Information
Operating system: Windows 10 x64

Blender Version
Broken: version: 2.82 (sub 6), branch: master, commit date: 2019-12-17 23:52, hash: e0cd5b980fc5, type: Release

As seen in T53517 comment :

Lamps are in Watts, [...]. Mesh lights are in Watts/m^2 [...]

Rendering the following two emissive items should give the same lighting :

  • rectangle area light, 1m x 1m, power 1W
  • plane mesh, 1m x 1m, Emission shader, strength = 1.0

This is not the case, area light power must be multiplied by 4 to get the same result.

Exact steps for others to reproduce the error

With the attached blend file,

  • enable 'Rendered' viewport shading
  • the 'Emissive plane' brings more light to the 'Ground' that the 'Area light'
  • change the 'Area light' power to 4W

Event Timeline

As a side note, the Emission documentation is unclear about the Mesh lights units :

_Strength_
Strength of the emitted light. For point and area lights, the unit is Watts. For materials, a value of 1.0 will ensure that the object in the image has the exact same color as the Color input, i.e. make it ‘shadeless’.

Area lamps are in Watts, emission shaders are in Watts/m2.

From the manual:

"Emission shaders on meshes are also in Watts/m2"

Doesn't sound like a bug at all.

Thanks for the manual except, I missed this part in the description.

But for my case, as my area light is 1W and 1m2, it should light the same as a 1W/m2 emisive plane measuring 1m2.

Philipp Oeser (lichtwerk) lowered the priority of this task from 90 to 50.Dec 18 2019, 7:27 PM

Area lamps are in Watts, emission shaders are in Watts/m2.
From the manual:
"Emission shaders on meshes are also in Watts/m2"
Doesn't sound like a bug at all.

Well, Strength of 1 on the Emission shader [with a 1m² face], should give the same as a 1m² Area Light with 1W power on first sight, then, no?
Checking....

Not sure, I guess it depends on how the intensity of an area lamp is distributed. If you scale up emitting geometry but leave the intensity the same, it gets dimmer (as far as I recall), but I don't think an area lamp does. I don't have Blender in front of me so I can't confirm ATM.

If you scale up emitting geometry you get more light because it keeps its brightness (which makes total sense because otherwise you would have to know the total size of the emitting surface).
If you scale an area light it is normalized so that the total amount of emitted light stays the same. Like in real life, your lamp emits a certain amount of light, but if you change the size of the diffuser in front of it you don't get more light but softer / dimmer light.

Don't even get me started on units and getting results from Cycles that are quantitatively correct in a radiometric sense...

Personally, I very much disagree with showing any unit for lamp strength in the UI, because it leads exactly to this confusion.

For example, the unit for a zero-radius point lamp is different from that for a point lamp with radius, the units for most lamps, sunlamps and emissive materials are different.

Similarly, there's the whole radiometry vs. photometry mess, see e.g. the IES loader.

Additionally, the problem with Watt in particular is that while it is meant in the radiometric sense in the UI, Watt has for a long time been used to quantify the intensity of real-world light sources based on their power draw, so now many people assume that a point lamp set to 100W should produce a properly exposed render of a room.

On the topic of exposure: Even if all units were handled correctly, the values in the rendered buffer would be in W/(m^2*sr) (or cd/m^2 - again, radiometry vs. photometry), which means that a realistically exposed scene would have values in the hundreds and completely blow out on the screen because we expect values relative to the screen brightness...

Personally, I think that as long as we don't have physical exposure controls (f-stop and ISO) for the camera in Cycles, there's no point in pretending that the output is quantitatively valid by showing units in the UI.

Sorry for this rant, but I hope this makes it clear that this topic is very complex and almost impossible to get right...

@Lukas Stockner (lukasstockner97) does this mean this should be classified as 'Known Issue'?

Brecht Van Lommel (brecht) changed the subtype of this task from "Report" to "Known Issue".

This can indeed be classified as known issue. It's more an area where we can improve things than necessarily a bug.

I'm not sure that the conversion math is even that simple, that these are supposed to match with the given units. It is also affected by the emission distribution function, two-sided vs. one-sided emission, and possibly other factors.