image filtering

radiance-general
1

Hi all,

Currently I'm working on a booklet showing my Radiance work.

Also some of the technicalities of Radiance are discussed.

I have to agree with what 'Rendering with Radiance' states on this topic;
its hard to explain precisely what makes Radiance fundamentally different
from other rendering solutions. Call me a pretentious prick, but I find it
effective in communicating Radiance' value to consider an image a virtual
prototype rather than just a rendering. To get this point across, its
important to show a potential client the different levels of significance a
Radiance rendering has. It all comes down to filtering in the end.

I think its important to explain how a pcond -h image differs from a
non-filtered image (well, pfilt)

I'm trying to sum up the different levels of information one can extract
from a single Radiance image using:

Pcond,

Glare,

Falsecolor (nits, Dayfact, lux)

Which is basically what one can expect from someone who has reasonable
control over Radiance.

Of course, there's lots of more ways of extracting different levels of
information from Radiance / a Radiance image.

Speaking for myself, I have been messing around piping rtrace intersections
to a .dxf file, showing a beautiful diagram of the way Radiance is computing
its light interactions. But I'm not that funky with image filtering.

Any suggestions of interesting / original ways of extracting information of
a Radiance image?

Cheers,

Jelle.

2

Hey Jelle,

Just a few thoughts here. Explaining rendering systems to clients is always difficult. I think that unfortunately clients typically just want good looking images and do not neccessarily care how the images is "made." However, I think that you are probably on a good track to consider the results of a Radiance simulation as a virtual prototype, which means something more than just a "rendering." But in order for this to hold true, the underlying data, (lighting levels, materials, geometry and other fundamental assumptions in the scene) used to generate the simulations needs to be "accurate" to some standard suitable for the given objective. Filtering an image with tone mapping offers one way to view the scene in the image.

-Jack

Jelle Feringa / EZCT Architecture & Design Research wrote:

···

Hi all,

Currently I'm working on a booklet showing my Radiance work.

Also some of the technicalities of Radiance are discussed.

I have to agree with what 'Rendering with Radiance' states on this topic; its hard to explain precisely what makes Radiance fundamentally different from other rendering solutions. Call me a pretentious prick, but I find it effective in communicating Radiance' value to consider an image a virtual prototype rather than just a rendering. To get this point across, its important to show a potential client the different levels of significance a Radiance rendering has. It all comes down to filtering in the end.

I think its important to explain how a pcond -h image differs from a non-filtered image (well, pfilt)

I'm trying to sum up the different levels of information one can extract from a single Radiance image using:

Pcond,

Glare,

Falsecolor (nits, Dayfact, lux)

Which is basically what one can expect from someone who has reasonable control over Radiance.

Of course, there's lots of more ways of extracting different levels of information from Radiance / a Radiance image.

Speaking for myself, I have been messing around piping rtrace intersections to a .dxf file, showing a beautiful diagram of the way Radiance is computing its light interactions. But I'm not that funky with image filtering.

Any suggestions of interesting / original ways of extracting information of a Radiance image?

Cheers,

Jelle.

------------------------------------------------------------------------

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

3

Hi Rob,

Sure, but this isn't an issue directly related to a specific scene or
geometry?

···

-----Original Message-----
From: Fitzsimmons, Rob [mailto:[email protected]]
Sent: Tuesday, July 19, 2005 6:51 PM
To: 'Jelle Feringa / EZCT Architecture & Design Research '
Subject: RE: [Radiance-general] image filtering

Jelle,
I would be interested in seeing your dxf models of the Radiance solution.
Could you post a rendering of the dxf solution?

Thanks
Rob Fitzsimmons

4

I have to agree with what 'Rendering with Radiance' states on this topic;
its hard to explain precisely what makes Radiance fundamentally different
from other rendering solutions.

I like your ideas on this. I'm fond of saying that Radiance is a
lighting simulator which can be used for rendering, rather than a
rendering tool. Usually I demonstrate this by showing a simulation
image next to a Radiance-generated diagram of numerical lighting
levels, emphasizing the point that this is an attempt to approximate
luminous reality with some level of accuracy.

I think its important to explain how a pcond -h image differs from a
non-filtered image (well, pfilt)

My approach here would be to overlay isolux contours on the unfiltered
image, and set it next to the filtered image. Does this sound like
the kind of thing you're looking for?

Randolph

···

On Tue, Jul 19, 2005 at 04:22:34PM +0200, Jelle Feringa / EZCT Architecture & Design Research wrote: