In article <web.3ecf8a69a14066c984d065310@news.povray.org>,
 "Matti Karnaattu" <nomail@nomail> wrote:

> Photographers use gray surface which reflects 18% light back to make
> adjustments because it looks half bright.
> 
>   0.18^x = 0.5
> 
>   x = ln(0.5)/ln(0.18)
> 
> I'm not sure about this gamma. I'm still looking more information about
> simulating camera and human eye dynamic range.

The gamma is an approximation to the non-linear response curve of CRT 
phosphors: 50% gray is neither 50% intensity nor 50% apparent intensity. 
Gamma currection compensates for this to give a nearly linear response. 
(LCD's have an S-shaped curve, probably based on cosine, but modern LCD 
displays probably compensate for this electronically.)

I don't know if a logarithmic scale would be useful given the limited 
dynamic range of monitors and typical depth resolutions (usually 8 bits 
per component, for 255 non-black levels). Also, people who work with 
color learn to use linear values, they would have to re-learn everything 
to use a logarithmic color space. It isn't more "correct", it is just a 
different system.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: chr### [at] tagpovrayorg
http://tag.povray.org/

Post a reply to this message

> I solve it myself and it is based on basic geometry formulas. I also
create
> Cornell box scene to test it. Instrutions to Cornell Box scene are here:
>
>   http://www.graphics.cornell.edu/online/box/data.html
>
> It works. I can't explain better because my notes and files are home and I
> have to use net from cafe until I get own Internet connection.


I want write a GPL python class (mainly to use it in Blender) to
write a Pov code to make a realistic-physical correct light simulation.

Can I use your formula?
Have you a site? Can you explain more?
Can you make a very simple example code?

Best regards,

                  Manuel

Post a reply to this message

>I don't know if a logarithmic scale would be useful given the limited
>dynamic range of monitors and typical depth resolutions (usually 8 bits
>per component, for 255 non-black levels). Also, people who work with
>color learn to use linear values, they would have to re-learn everything
>to use a logarithmic color space. It isn't more "correct", it is just a
>different system.

It is useful because realistic radiosity requires real luminances.

Matti

Post a reply to this message

>I want write a GPL python class (mainly to use it in Blender) to
>write a Pov code to make a realistic-physical correct light simulation.

>Can I use your formula?

Of course. I forgot to say.. fade_distance formula works only with flat
light sources.. For omnidirectional light you should set fade_distace to
light source radius.

>Have you a site? Can you explain more?

No, I dont have site. fade_distance formula is derived from circle area.. I
try to explain fade_power formula and give you example code when I have
more time.

Matti

Post a reply to this message

> No, I dont have site.

:-(

>fade_distance formula is derived from circle area.. I
> try to explain fade_power formula and give you example code when I have
> more time.


OK. I'm waiting :-)
thanks

Best Regards,

           Manuel

Post a reply to this message

Hi Matti,  I'm a newbie in radiosity so I think my question is a stupid one.
I can't understand what you really mean with "photograph dynamic range".
Is it referred to the challenge of simulating the human eye dinamic exposure
range? Or is it referred only to the photographic devices simulation? Like
film exposure range, diafragm/time exposure calculation?

Thank you in advance.
bye, save


Matti Karnaattu wrote:
>>I mean the simulation with the radiometry (or photometric) value of
>>a lamp (power, light intensity, or Photometric Solid, etc...).
>
>
>Best way to use POV-Ray light simulation is to multiply light source color
>with real light luminance (cm/m^2). Then set fade_distance using following
>formula:
>
>  sqrt(pi * A) / pi
>
>Real world fade_power is 2. Because POV-Ray doesn't have dynamic range
>compression, this isn't enough. I recommend you to define two adjusting
>variables "Gamma" and "Max_Luminance" to make dynamic range compression.
>Gamma is the gamma used to compress image dynamic range and Max_Luminance
>is the luminance (cd/m^2) where pixel luminance at range 0..1 is "1".
>Max_Luminance controls image brightness. When using these variables you can
>calculate light source color multiplier using formula:
>
>  pow(light_source_luminance / Max_Luminance, Gamma)
>
>And global Fade_Power to *all* light sources by using formula:
>
>  (ln(-exp(-ln(2) * Gamma) + 2) + ln(2) * Gamma) * 2 / ln(3)
>
>To simulate photograph dynamic range compression I think that Gamma should
>be ln(0.5)/ln(0.18) because gray card which reflects 18% of light looks
>half bright.
>Light source color luminance (without multiplier) must be one. Good starting
>value for Max_Luminance is 50. Light source colours luminances without
>multiplier must be one. Example for white light, light color is <1, 1, 1>.
>
>Because you probably want to make other light sources than omnidircetional,
>you need to simulate light distribution using spotlights. If light source
>is flat, use spotlight radius -90 falloff 90 tightness 1 for cosine
>falloff.
>
>I don't recommend to use Jaime's lightsys because it doesn't use physically
>accurate lightning. (Sorry Jaime).
>
>Matti
>

Post a reply to this message