Directional Area Lights

Hehe, sure thing! Apparently, then, “directional” must just be the way Bulgarian people like to refer to it… :smile:

I do like “spread” a lot! But it would make more sense then to reverse the values. 1 would mean maximum spread, thus the default behavior. If we go for spread, it would be even clearer to express it as an angle, moreover as, internally, that parameter is in fact processed as an angle. The default would then be 180°, to be reduced to 0° for maximum focus. I like the idea.
Would this cause any impracticality anywhere else?

8 Likes

I might be wrong, but I think if’ve hear direction-related words (sometimes angles) more often than “spread” across softwares. So maybe it would make it easier for everyone to stay in this lexicon?
Anyway it’s nothing that important, as long as it’s consistent at least within Blender.

I think “spread” is fine, it’s short and understandable.

2 Likes

Here’s LuxCore, it’s called “Spread Angle” here

2 Likes

I was just about to say that. Expressing directionality as an angle would be clearer and a whole lot easier to visualize conceptually.

1 Like

In stage lighting there are several tools for shaping the directionality of the light. Barn doors, top hats. They can be characterized not only by an angle at which they can be opened, but also by the length of these shutter-like element, so probably there should be a parameter for shutter length, in addition to directionality (angle, spread). As you can see, on the screenshot, light shutter angle and length combined give pretty dramatical difference on penumbra light distribution. Emissive plane is hidden inside the light constraining meshes.

7 Likes

The ‘Photographer’ addon tackles the intensity issues, might be good to have a look at that.
Having a spread option would be great btw. And ‘Spread’ is used a lot of this option in other renderers.

For barndoor type of light effects, in Arnold this is done via a shader node that is linked to the light shader. Maybe that would be a interesting route to take, as it opens up more possibilities for light related things as it is not ‘bolted on’.

cheers.

1 Like

It’s like IES profiles


So supporting this in blender would be great

yes this seem cool, but is better the control on lamp’s spreads this library is limited to the lamps specifics that are constrained by glass and shapes, and are good for fast architectural viz rendering. Is always better having custom controls on lighting, without external library.

My favourite solution so far is this: https://blender.stackexchange.com/a/57046
It gives great results, but I’m pretty sure there’s energy loss.

In my opinion, both spot and area lights should natively have this directional intensity curve, not just a parameter or two, (and possibly directional colour gradient) since the immediate increase in realism is significant. This would be an easy way–even for beginners–to create real looking light sources and to match reference photos without knowing what actual light source was used.

2 Likes

Barndoors and honeycombs (directionality) have different use cases. Honeycombs can be used to create harsher shadows while keeping a soft spec.

Sorry for the long absence…

I’ve submitted a patch for revision here: https://developer.blender.org/D10594
The patch does not compensate for the luminosity loss, as neither the spotlight does. To maintain some commonality among the two types of light, I thought that we could come up in the future with a specific switch, a checkbox called “preserve luminosity”, that would handle that aspect, making both lights luminosity-constant. In fact, I believe users should be able to decide which mode they prefer to operate in, as there are some situations in which the current behavior might be preferable: for example when the light is used as a key light and the variating luminance at the center of the beam could require an otherwise unnecessary exposure correction. Or during animations, the parameter could be animated to create interesting light effects. Ultimately, for those who come from traditional studio photography, this behavior is pretty much what they expect from a physical honeycomb in front of their lights.

22 Likes

I have a couple of questions about the new Area Light Spread setting as I am incorporating it into the Physical Lights units of my Photographer addon.

  • The spread value conserves energy, meaning using a narrower spread will focus the beam and make it brighter. However, this is not how the Spotlight Size (cone angle) works, does it make sense to have two different behaviors? I see that you wanted to give the option but it’s not what has been committed at the moment.
    I will support both depending if it’s lumen or candela anyway, but I am curious about this choice.

  • Spread is not supported by EEVEE and breaks compatibility between the two engines now, especially because of the brightness difference. Will it be added before 2.93 release?

Best,
Fabien

It makes sense to have the option for both, it just was decided this was not a requirement for the patch to be added.

Probably not for 2.93.

It’s not obvious how to do this in realtime, at least I’m not sure how you’d adapt the Linearly Transformed Cosines method to this. But that shouldn’t hold back features in Cycles.

2 Likes

hello Matteo thanks for the new feature very useful.

I have a question regarding using area lights as a projector.
I tested it with a video file and it works. However currently in order to get a sharp projection on a wall, the area light has to be close to the wall like this:

As soon as I move it away the projection gets blurry.

My guess is that it has to do with the current minimum of 1 degree spread?
Would it be possible to have the option to set it to 0, in order to use the area light as a projector?

Hi, it is already being proposed, discussed and (I guess) planned

1 Like

ok thanks. Can you please share the link to the discussion?

It was mentioned in the rendering meeting notes from 2021-04-27:

Link to the topic:

4 Likes

The current model “goes crazy” with spread values very close to 0 (number of rays hitting the light approaching zero, while their individual intensity skyrockets towards infinity), that’s the reason behind the decision of capping it to 1 degree. I would personally be very committed to making it work for spread 0, but I’m overwhelmed with work for at least the whole next month. Hopefully, someone will be able to get it done sooner, otherwise it is on my to-do list for sure (given that the problem is not above my skills and I’ll be able to crack it).

4 Likes