On 08/23/2018 06:16 AM, clipka wrote:
> Am 23.08.2018 um 02:36 schrieb Jim Holsenback:
>> well i'm kind of chasing my tail so i thought i'd ask for some guidance.
>> i'm using radiosity and a very dim point light to get some shadow
>> action. check out the lower left hand corner of the image. that's what
>> i'm shooting for! why don't the other spheres have the same affect?
>> other that the color there's no diff...the emission component is the
>> same level for all the colors. is it more about the emission of the
>> object or the surface that is showing the affect.
> 
> Looks to me like a radiosity setting issue, with too low sample density.
> 
> In the bottom left corner, by chance a sample was taken very close to
> the base of the green... tron unicycle?... and picked up the emission
> very nicely, while all other samples were taken comparatively far away
> from the gizmos.
> 
> There are various knobs you might try tweaking here, but my personal bet
> is that `error_bound` is set too high. You'll probably want a value of
> 0.5 or even lower. The default of 1.8 certainly won't cut it.
> 
> In reducing `error_bound` you may find that the result looks unnaturally
> splotchy. If it is a general effect, try increasing `nearest_count`
> and/or `count` to compensate. If instead (or after you adjust
> `nearest_count` and/or `count`) the splotchiness is limited to nooks and
> crannies, try reducing `minimum_reuse`.
> 
> Reducing `error_bound` (and, in some cases, increasing `count`) may also
> introduce block-like artifacts. Those should be easy to fix by reducing
> `pretrace_end`. A too high `pretrace_end` may also prevent changes to
> `nearest_count` from being fully effective.
> 

your response wasn't here when i was following up to thomas ... but yep 
lowering nearest_count started getting close to expected results... a 
little splotchy but i think you've set me on the right path... danke

Post a reply to this message