|
|
On 1/14/22 6:06 AM, Warren wrote:
> Cousin Ricky <ric### [at] yahoocom> wrote:
>> On 2022-01-13 10:06 (-4), Warren wrote:
>>> Hi,
>>> I post a message here, because I get a weird render with povray 3.8.0 beta 1 of
>>> a scene with the following resolution : 1280x960 (see attached image with this
>>> message) , you can see a white stain that has the shape of a heart near the
>>> pentagram. When I render this same scene with stable v3.7.0 the white stain
>>> doesn't appear. And even weird, when I render this scene with povray "3.8.0
>>> beta1" but with the resolution of 640x480, I don't get this white stain in the
>>> final image.
>>
>> What I've found so far:
>> - The artifact varies with the image size. A 960x720 render leaves an
>> oblong stain on the upper left vertex.
>> - The artifact isn't completely absent with 640x480. At 640x480, tiny
>> white spots appear along the veins of the pentagram, but they are
>> absent in a 3.7 render. I presume that this means you did not put
>> them there deliberately, and that they are related to the big heart-
>> shaped stain.
>> - 3.8 beta 2 does not solve the problem.
>
> There are other things to consider:
> - First off, set the boolean 'PlaceWell' (line 19 of main pov file) to false, to
> get a faster rendering, the heart shape artifact being always there.
> - Changing the minimum_reuse value from '0.015' to '0.01' make the 'heart shaped
> artifact' to disappear, but the little white stains you talked about are there,
> it seems.
>
The following might, or might not, be of help to you with official
versions of POV-Ray.
I took a quick look at at this bug with my current povr branch and found
along with some recent bounding code changes it got me to a more stable
fail for a bug I've been chasing for years. Details posted in the v4.0
group.
My povr branch is now considerably different than official POV-Ray
offerings, but after translating the pentagram up by 2x the thread
radius(a) I got away from the 'white regions.' I never saw the heart
shaped one - no idea why not in povr. Image attached.
(a) - Media needs clean start and stop intervals. Near coincident
shape/surfaces can cause confusion and leave intervals open - and with
emission media this results in extremely bright regions.
Where you had media I moved from min, max samples and interval
specifications to just 'samples 10'.
In many places you are using the srgb keyword with channel values
outside the 0-1 range. My povr branch doesn't allow this because any
'srgb calculation' outside the 0-1 range is wacky. See post elsewhere
for details. If you need values outside 0-1, use rgb to exactly specify
what you want for internal linear values.
includes/outdoorsLamp.inc -> debug statement line 218 uses LampZpos and
it should be LampZPos.
Bill P.
Post a reply to this message
Attachments:
Download 'image_povrtracefix.jpg' (105 KB)
Preview of image 'image_povrtracefix.jpg'
|
|