See thread in p.b.i.  The scene uses two light sources, one to trace a 
cylindrical parallel beam through two prisms and across a room, and 
another, with media interaction turned off, to light the scene.  The 
latter light has a looks_like for what reason I don't remember.  (This 
was 12 years ago!)

When the scene is rendered as an animation, the looks_like is omitted in 
frame 2.

File beam0-pared-by37.log is the statistics from a v3.7.0 run with 
+KFF2.  File beam0-pared-by36.log is the statistics from a v3.6.1 run 
with +KFF2 +EF1.

POV-Ray versions:
   3.6.1 (g++ 3.4.1 @ i686-pc-linux-gnu)
   This is an official version prepared by the POV-Ray Team.

   3.7.0.unofficial [no source code alterations]
   Compilation settings:
     Build architecture:  x86_64-unknown-linux-gnu
     Built/Optimized for: x86_64-unknown-linux-gnu (using -march=native)
     Compiler vendor:     gnu
     Compiler version:    g++ 4.8
     Compiler flags:      -pipe -Wno-multichar -Wno-write-strings 
-fno-enforce-eh-specs -s -O3 -ffast-math -march=native -pthread

OS: openSUSE 13.2 (GNU/Linux)

Post a reply to this message

Attachments:
Download 'beam0-pared.zip' (5 KB)

Cousin Ricky <ric### [at] yahoocom> wrote:
> See thread in p.b.i.  The scene uses two light sources, one to trace a
> cylindrical parallel beam through two prisms and across a room, and
> another, with media interaction turned off, to light the scene.  The
> latter light has a looks_like for what reason I don't remember.  (This
> was 12 years ago!)
>
> When the scene is rendered as an animation, the looks_like is omitted in
> frame 2.

Don't know the reason behind it but simply making the looks_like sphere hollow
allows the photons to show.

Post a reply to this message

Am 07.07.2017 um 20:10 schrieb omniverse:
> Cousin Ricky <ric### [at] yahoocom> wrote:
>> See thread in p.b.i.  The scene uses two light sources, one to trace a
>> cylindrical parallel beam through two prisms and across a room, and
>> another, with media interaction turned off, to light the scene.  The
>> latter light has a looks_like for what reason I don't remember.  (This
>> was 12 years ago!)
>>
>> When the scene is rendered as an animation, the looks_like is omitted in
>> frame 2.
> 
> Don't know the reason behind it but simply making the looks_like sphere hollow
> allows the photons to show.

Well, /I/ know the reason behind it now ;)

During refactoring for 3.7.0, a bug was introduced that may cause
objects in the same CSG group to mess with each others' `hollow` flag,
with the last object winning. In this case that happened to be the
non-hollow `looks_like` object.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Am 07.07.2017 um 20:10 schrieb omniverse:
> > Cousin Ricky <ric### [at] yahoocom> wrote:
> >> See thread in p.b.i.  The scene uses two light sources, one to trace a
> >> cylindrical parallel beam through two prisms and across a room, and
> >> another, with media interaction turned off, to light the scene.  The
> >> latter light has a looks_like for what reason I don't remember.  (This
> >> was 12 years ago!)
> >>
> >> When the scene is rendered as an animation, the looks_like is omitted in
> >> frame 2.
> >
> > Don't know the reason behind it but simply making the looks_like sphere hollow
> > allows the photons to show.
>
> Well, /I/ know the reason behind it now ;)
>
> During refactoring for 3.7.0, a bug was introduced that may cause
> objects in the same CSG group to mess with each others' `hollow` flag,
> with the last object winning. In this case that happened to be the
> non-hollow `looks_like` object.

Much appreciation to you for finding that.

I hadn't thought of the looks_like being part of CSG if a lone object is used.

Wow, code reading is fun and confusing. Just before you answered to that I was
finally getting a look at both the 3.7 and 3.62 source for comparisons of what
might have changed, in particular the parse.cpp file, and I sure get lost fast.

I just wanted to see what parts looks_like dealt with, although I guessed it was
probably the object or hollow causing the trouble when conveyed to the parser
somewhere. I was thinking about the "children" parts and if that might be the
reason.
Please ignore me for my lack of certainty, clarity and definition there. I'm way
out of my league!

Anyway, thanks again for checking on that... both of you Christoph and Richard.

Bob

Post a reply to this message

Am 08.07.2017 um 09:07 schrieb clipka:
> Am 07.07.2017 um 20:10 schrieb omniverse:
>> Cousin Ricky <ric### [at] yahoocom> wrote:
>>> See thread in p.b.i.  The scene uses two light sources, one to trace a
>>> cylindrical parallel beam through two prisms and across a room, and
>>> another, with media interaction turned off, to light the scene.  The
>>> latter light has a looks_like for what reason I don't remember.  (This
>>> was 12 years ago!)
>>>
>>> When the scene is rendered as an animation, the looks_like is omitted in
>>> frame 2.
>>
>> Don't know the reason behind it but simply making the looks_like sphere hollow
>> allows the photons to show.
> 
> Well, /I/ know the reason behind it now ;)
> 
> During refactoring for 3.7.0, a bug was introduced that may cause
> objects in the same CSG group to mess with each others' `hollow` flag,
> with the last object winning. In this case that happened to be the
> non-hollow `looks_like` object.

... actually, no: The bug wasn't introduced in v3.7; it was a
pre-existing condition.

The reason your scene renders fine in v3.6 is that the symptoms were
ever so slightly different: Instead of the last object, it was the first
object that won the `hollow` flag war.

Place the box after the light source, and it will render fine in v3.7
but break in v3.6.


Can someone please open a GitHub issue for this one? It may need some
work before it can be fixed.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> The reason your scene renders fine in v3.6 is that the symptoms were
> ever so slightly different: Instead of the last object, it was the first
> object that won the `hollow` flag war.
>
> Place the box after the light source, and it will render fine in v3.7
> but break in v3.6.

I'm surprised I hadn't noticed this problem before with it around all that time,
or maybe I just change the scenes until they work right... or give up on them.

> Can someone please open a GitHub issue for this one? It may need some
> work before it can be fixed.

Done, whether correct enough is another matter but I tried my best.  :)

Bob

Post a reply to this message