On 06/06/2011 07:50 PM, clipka wrote:
> Am 06.06.2011 23:20, schrieb Jim Holsenback:
>
>> I've worked on a few thinks, but this post is about needed a bit of
>> radiosity help. These are the settings I'm using:
>>
>> pretrace_start 1/16
>> pretrace_end 1/64
>
> I would reduce the pretrace_end...
>
>> count 256
>
> ... set count to "256,10000" or some such (not sure whether it will make
> much of a difference though)
>
>> error_bound 0.256
>> low_error_factor 0.125
>
> ... increase the low_error factor (0.5 seems a sane value)...
>
>> nearest_count 12
>
> ... and set nearest_count to "20,10" or something alike, to make use of
> adaptive pretrace.
>
>> gray_threshold 0
>> normal on
>>
>> .... but as you can see it's still a bit patchy. I don't think count is
>> the answer and was wondering if what's needed here is in the missing
>> ingredient ... min/max reuse ... right now it's default.
>
> Recently an error has been reported in 3.70RC3 causing noticeable
> splotches when using only very few pretrace steps with a high
> pretrace_end setting; maybe that's related to what we're seeing here.
>
> Unless you're using the current developers' version, that is. In that
> case you're already running with a fix for that error.

No I'm running the latest incantation ... initially I'd thought that was 
my problem and when it was 1st reported I pulled the fix you'd posted. 
Question about pretrace_end ... won't a smaller value make more of the 
samples to be gathered in the final pass? I thought it was better to 
shoot for having the final pass sample count about half the total. Oh 
and thanks for the nudge, I'd forgotten about the adaptive settings for 
nearest ... had been bouncing between 12 and 20 so (lol) I guess 20,12 
it is then.

Post a reply to this message