>> Yes, it should be perfectly identical to if you'd set up the identical 
>> scene
>> in real life and taken a photo with a perfect camera.  This is the goal 
>> of
>> POV.
>
> Well, wake me up when you get there. I really want to see that. And even 
> if you
> do get there, I'm not sure you will like what you see. I suggest you check 
> a
> little article I read on how photographers like Ansel Adams had to use 
> filers
> that I posted somewhere here.

All those ideas (filters, lighting etc) are equally applicable to your POV 
scene as they are to real life.  Again, the goal of POV is that whatever you 
do IRL (eg add a reflector, light or filter to your scene) would give the 
same result in POV.  Note there is still a long way to go, but that doesn't 
mean it shouldn't be the goal.

> There was similar quest years back about using computers to imitate visual
> perception, that is, computers that see. And in the end, with all the 
> variables
> that couldn't be computed, like size constancy and object recognition and 
> such,
> they gave up.

Fortunately not everyone gave up!

Post a reply to this message

"scott" <sco### [at] scottcom> wrote:
>
> All those ideas (filters, lighting etc) are equally applicable to your POV
> scene as they are to real life.  Again, the goal of POV is that whatever you
> do IRL (eg add a reflector, light or filter to your scene) would give the
> same result in POV.  Note there is still a long way to go, but that doesn't
> mean it shouldn't be the goal.

Beating a dead horse here, but a comment on this point. Yes, POV, as it is, does
allow such "filters". What concerns me in what appears to be the direction of
the new release is that many of these techniques seem to be either deprecated or
made so complicated and cumbersome to impliment they are in effect impossible to
use. Especially "assumed_gamma" which I consider essential to the process but
which some here have described as a kind of cheap and quick artistic trick.

Post a reply to this message

> Beating a dead horse here, but a comment on this point. Yes, POV, as it 
> is, does
> allow such "filters". What concerns me in what appears to be the direction 
> of
> the new release is that many of these techniques seem to be either 
> deprecated or
> made so complicated and cumbersome to impliment they are in effect 
> impossible to
> use.

What specifically has changed?  You've always been able to put objects in 
front of the camera to mimic filters, in fact recently more features have 
been added to the camera to give much more flexibility.

> Especially "assumed_gamma" which I consider essential to the process but
> which some here have described as a kind of cheap and quick artistic 
> trick.

assumed_gamma is not meant to be used as an artistic tool, if you really 
want such an effect then just do some post-processing in Photoshop the same 
way you would with a real photo - it's much more flexible.

Post a reply to this message

"scott" <sco### [at] scottcom> wrote:
>
> assumed_gamma is not meant to be used as an artistic tool, if you really
> want such an effect then just do some post-processing in Photoshop the same
> way you would with a real photo - it's much more flexible.

This is true and, as any experienced photographer knows, if in the darkroom or
Photoshop, you usually do. But as Ansel Adams discovered it always comes out
more natural looking to put the filter directly on the lens. Compared to that
dodging and burning highlights and shadows is kind of trick. But a trick still
necessary to "calibrate", one might say, the physical input to the human eye. I
refer to the article quoted in this thread about how Adams used lenses:
http://www.kenrockwell.com/tech/filters.htm. Just as you have to calibrate a
monitor to faithfully reflect the picture, you have to adjust the picture so
that it looks convincing to the brain.

The article goes on "That said, the best images come when nature is at her best"
But POV does not have a beautiful moonrise over Yosemite to work with. But in
the dark with only our imagination to guide the way.

Post a reply to this message

> This is true and, as any experienced photographer knows, if in the 
> darkroom or
> Photoshop, you usually do. But as Ansel Adams discovered it always comes 
> out
> more natural looking to put the filter directly on the lens.

Which is why I mentioned that POV also allows you to put arbitrary objects 
in front of the virtual camera to simulate many lens effects.  Off the top 
of my head a polarising filter would be one of the few that is not possible 
to simulate.  I don't get what this has to do with assumed_gamma though, 
which lens IRL are you trying to simulate by toying with assumed_gamma?

Post a reply to this message

"scott" <sco### [at] scottcom> wrote:
>
> assumed_gamma is not meant to be used as an artistic tool, if you really
> want such an effect then just do some post-processing in Photoshop the same
> way you would with a real photo - it's much more flexible.

I don't think you appreciate how much "artistic effect" goes into just seeing
things realistically in the first place. It's not an "effect". It's a necessary
part of the visual process. The brain pre-processes everything which then, in a
sense, has to reciprocally post-processed to make it look right. All the concern
about calibrating monitors precisely and adjusting for gamma seems to overlook
this ultimate "monitor" which works nothing like raytracing or photography.

Painters have discovered this, even naturalistic painters like Vermeer, long
before Ansel Adams or digitized photographers. The eye is not a camera. It works
by different rules than physics, rules discovered mainly by artists, and which
science is now just beginning to understand.

Post a reply to this message

> Which is why I mentioned that POV also allows you to put arbitrary objects
> in front of the virtual camera to simulate many lens effects.  Off the top
> of my head a polarising filter would be one of the few that is not possible
> to simulate.  I don't get what this has to do with assumed_gamma though,
> which lens IRL are you trying to simulate by toying with assumed_gamma?

You know I never really tried actually putting a virtual filter on the POV
camera. I've never really had to, what POV can simulate being so far from nature
in all her glory, using, say a red filter to bring out the depth in clouds. It's
hard enough just getting something that looks like a cloud. I've always used
assumed_gamma as not just the easiest but only possible way of adjusting
"curves", as they say, relative tonal values, like the red filter to bring out
depth. As you know too, from the experiments of Polaroid and others, that
filters used with actual light is very different from "filters" that could be
put in front of a POV lens, where say a red filter probably (but never tried it)
I'm guessing would just tint everything red.

Post a reply to this message

> I don't think you appreciate how much "artistic effect" goes into just 
> seeing
> things realistically in the first place. It's not an "effect". It's a 
> necessary
> part of the visual process. The brain pre-processes everything which then, 
> in a
> sense, has to reciprocally post-processed to make it look right. All the 
> concern
> about calibrating monitors precisely and adjusting for gamma seems to 
> overlook
> this ultimate "monitor" which works nothing like raytracing or 
> photography.
>
> Painters have discovered this, even naturalistic painters like Vermeer, 
> long
> before Ansel Adams or digitized photographers. The eye is not a camera. It 
> works
> by different rules than physics, rules discovered mainly by artists, and 
> which
> science is now just beginning to understand.

I think you need to separate the process of calculating the correct light 
values (which is what POV is designed to do) and mapping those to some pixel 
values that looks realistic (which POV does is a very primitive way).  The 
issues you mention only exist because it's not possible to display the full 
range of colours and intensities human's can see yet on displays or paper. 
There are plenty of complex techniques for mapping actual physical light 
data (either from sensors or program like POV) to 8bit images suitable for 
displaying on a standard PC monitor, none are perfect.  POV doesn't pretend 
(yet) to be able to do any of this, if you want this then use an HDR output 
format and an external tool.

Post a reply to this message

> You know I never really tried actually putting a virtual filter on the POV
> camera. I've never really had to, what POV can simulate being so far from 
> nature
> in all her glory, using, say a red filter to bring out the depth in 
> clouds. It's
> hard enough just getting something that looks like a cloud. I've always 
> used
> assumed_gamma as not just the easiest but only possible way of adjusting
> "curves", as they say, relative tonal values,

And what happens when one part of your scene (eg the clouds) requires a 
different assumed_gamma than another part to look realistic?  Or you want to 
use an object from the object library (or any other source) in your scene? 
IMO it's far better to play about with the light values, media properties or 
material properties to get the scene looking how you want.  That way you can 
be sure it's going to render the same on any machine, and you will be free 
to put external parts in to your scene.

Post a reply to this message

I GET IT!

I apologize for raising such a firestorm here and hope you can forgive us
old-timers for being so accustomed to our habitual ways of doing things. It is
hard sometimes to wrap our ingrained minds around changes that we are not used
to. Especially when they seem to contradict the fixed reference points,
mathematically and graphically and "commonsensically", we've used to test what's
true. My focus is graphics which uses a different approach. Not that one is
right or wrong. We can all get to the same place. Vermeer did it centuries ago,
but with different brushes and a different palette of tools. And like some of
us, we're skeptical of anything we can't prove to ourselves.

There are still issues with the limits of raytracing and its need for artistic
expression but that's somewhat beyond the immediate point. At least I now see
more clearly how they relate.

What convinced me was a test of the well-known shadow illusion, just a cylinder
on a black and white grid:
http://scienceblogs.com/startswithabang/upload/2010/02/weekend_diversion_a_question_o/772px-grey_square_optical_illusio
n.PNG.png
In trying to reproduce it, it became obvious that 3.7 is, at least on a
functional level, basically identical to previous versions with assumed_gamma
set to 1. The only difference being that ambient_light, which previously I
seldom ever considered, and assumed_gamma, which I used all the time, now play a
very different part.

So, sorry and thanks especially to clipka for his patient and thorough attempts.

SK

Post a reply to this message