Am 15.11.2016 um 02:33 schrieb [GDS|Entropy]:

>> It may come as a surprise to you, but the glow feature is entirely
>> independent of media, so there's no reason to tie modifications of one
>> to the other.
> 
> It is surprising, both because Glows look the way the do, and because I had
> thought there was perhaps a greater tie between the two functions due to
> your previous mention that resolving the glow bug was related to scattering
> media in some way.

What I meant was that the glow artifacts I was thinking of were inherent
to the simplicity of the glow algorithm, and that repairing them would
mean redesigning the feature to use a much more complex algorithm very
similar to that of scattering media.

> How do glows work? Is it like the fog keyword? They tend to have a fog sort
> of effect on their own without media or fog.

The feature with the most similarities is probably ground fog.

> It may be interesting at a later date to make them an actual volumetric
> effect if they do not work that way currently.

Define "volumetric effect".

If you mean "an effect depending on properties of a 3D region of space,
as opposed to properties of a surface", then glow already is a
volumetric effect.

If you mean "an effect computed by numeric integration along a ray, as
opposed to symbolic integration", then no, glow does not match that
definition. But I don't think it makes much sense to redesign the glow
feature in such a manner: Like fog it is a lightweight feature, thanks
to the use of symbolic integration, which comes at the cost of
imperfections; changing it to use numeric integration could eliminate
those imperfections, but would also turn it into heavyweight feature. If
you want to realistically model halos caused by some scattering medium,
use... (drumroll)... scattering media.

Post a reply to this message