I remember when I was recreating my bathroom and I modeled the lamp and gave it a glass material, but when I came to the lighting it got confusing. I modeled a sphere in the lamp because in real life the lighting comes from the lamp obviously. And then when I render the scene, no matter how strong the sphere light is I always get a pretty low light.

Is Cycles a realistic or close to realistic render engine? Or what lamps should I use? Do I have to fake the light? How can I replicate the lighting as in real life?

Edit: I'm not sure if the spherical emitter will light up the room better if it is not inside a glass material

  • $\begingroup$ Could you provide a link to your .blend-file and post some screenshots/render-output so one can see what is wrong? Your description of the problem is not very clear about what has gone wrong except "something with the light, it's too dark", thanks $\endgroup$ Commented May 16, 2014 at 20:01
  • $\begingroup$ So you have a lamp object inside a mesh object? or have you used an emission node on the mesh material? a mesh around a light object will block light emission like an object casting a shadow. $\endgroup$
    – sambler
    Commented May 17, 2014 at 0:52
  • $\begingroup$ See blender.stackexchange.com/q/2558/599. The glass object around the emitter will make it very difficult for the emitter to illuminate objects, as a ray must bounce off an object and through the glass in order to illuminate the object. While this approach might work okay with Bidirectional PT and MLT, you probably want to use some tricks to render this with backwards PT as cycles uses. $\endgroup$
    – gandalf3
    Commented May 17, 2014 at 1:14
  • $\begingroup$ Also related: blender.stackexchange.com/questions/8645/…. If you want to be as physically accurate as possible, you might want to look at using luxrender. $\endgroup$
    – gandalf3
    Commented May 17, 2014 at 3:07

1 Answer 1


The problem

The issue here is that cycles, which uses backwards path tracing (tracing light rays from the camera to the light sources), is not very good at rendering caustics in these kinds of situations.

This is because it's not very likely for a ray to bounce off a surface, transmit through the glass bulb, and hit the element within. Multiple Importance Sampling (MIS) usually helps with this by guiding more rays to hit light source(s), however the glass bulb surrounding the light source effectively prevents this from working.

From left to right: Element with MIS, Element without MIS, Element with MIS and glass bulb.

enter image description here enter image description here enter image description here

To properly render such a situation, solutions like Bi-direction path tracing and Metropolis Light Transport (MLT) exist. Unfortunately, cycles does not yet have this functionality, and it is considered "not a priority" according to the cycles roadmap.

Even so, there is a working patch for MLT (and adaptive sampling/stopping) which is being actively developed as of the time of this posting. You will have to compile blender yourself (it's really not hard) to use this patch. There is also an older patch for Bi-dir here.

Here is an example render of the element and bulb with the MLT patch:

enter image description here

It's better, but not at all the desired result (I don't really know any of the MLT related settings yet, so it's possible this could be improved some more).

See also:


enter image description hereenter image description here

  • Use a point lamp for doing the actual illumination, and make the bulb and filament only visible to the camera by mixing the glass or emission shader with a transparent shader based on the is camera ray output of the light path node.


There are a few things which can prevent indirect bounces from transmitting through objects entirely:

Make sure No caustics is disabled and any lamps which you want to shine through or reflect off transmissive or glossy materials have MIS enabled.

  • $\begingroup$ thanks. Since I'm a sort of beginner to blender it confused me a little but I'll catch up. $\endgroup$
    – Vladimir
    Commented Jun 2, 2014 at 17:37

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .