On 08/04/2014 07:49 AM, clipka wrote:
> Am 04.08.2014 13:31, schrieb James Holsenback:
>> On 08/04/2014 06:57 AM, clipka wrote:
>>> Am 04.08.2014 12:04, schrieb James Holsenback:
>>>> On 07/23/2014 07:27 PM, clipka wrote:
>>>>> I'm currently implementing a new UberPOV feature that allows bypassing
>>>>> the caching of radiosity samples, effectively resulting in purely
>>>>> stochastic unbiased Global Illumination computation. (See recent
>>>>> discussion on povray.unofficial.patches.)
>>>>
>>>> here's the image that prompted my parameter question in p.u.patches ...
>>>> the alcove is made of box primitives and has cylinders to round the
>>>> insides of the corners, used merge not union.
>>>
>>> I think what you're seeing is not an artifact at all.
>>>
>>> A surface like the floor typically gets darker near other surfaces such
>>> as the wall; but the strength of this effect depends on the wall's
>>> brightness. A very bright wall - whether just due to its colour or also
>>> due to the angle of illumination - may cause the floor nearby to even
>>> become /brighter/ than elsewhere, while a very dark wall - whether just
>>> due to its colour, angle of illumination or shadows - will cause the
>>> nearby floor to become /extremely/ dark.
>>>
>>
>> Hmmm ... ok then just for sake of discussion here's how
>>
>> #include "rad_def.inc"
>> Rad_Settings(Radiosity_Fast, off, off)
>>
>> handles things
>
> ... because it happens to utterly miscompute it.
>
> Use more pretrace steps and a lower error_bound for starters.
>

well yeah ... my point was to compare no_cache with bare bones radiosity 
settings. i'm thinking it (rad) get's closer with what amounts to "quick 
pass" settings.

Post a reply to this message