Hi, I'm trying to render a surface I exported from a VTK visualisation software,
but there are some strange lights under the surface

The bottom part of my figure is here:

https://s32.postimg.org/7y35aegp1/sample.png


I'm using this settings at the beginning of my POV file

#include "rad_def.inc"
#include "colors.inc"

// For the mesh2 textures:
#declare my_texture = finish { phong 1 phong_size 20 ambient 0
                                   reflection{metallic 0}
                              };

global_settings {
      max_trace_level 30
      assumed_gamma 1.5
      radiosity { Rad_Settings( Radiosity_Fast, on, on ) }
      photons { spacing 0.005 }
      }

background { color rgb <1.000000, 1.000000, 1.000000>}

camera {
 perspective
 location <8.196152, 8.196152, 8.196152>
 sky <0.000000, 0.000000, 1.000000>
 right <-1, 0, 0>
 angle 30.000000
 look_at <0.000000, 0.000000, -0.000000>
}

light_source { <-3, -3, 4.>
               color White
               area_light <6, 0, 0>, <0, 6, 0>, 6, 6
               area_illumination on
               photons {area_light}
               jitter
               // adaptive 1
             }

plane
{
    <0, 0, 1>, -1
    texture{ pigment { color rgb < 1, 1, 1 > }
             finish{ ambient 0 diffuse 0.5 }
         }
}

....


Is there a way to fix that? I tried increasing the max_trace_level but it
doesn't change, neither with the photons although I didn't specify photons in
the main figure (it gets too slow)

Post a reply to this message

On 06/22/2016 06:28 AM, aradan.d wrote:
> Hi, I'm trying to render a surface I exported from a VTK visualisation software,
> but there are some strange lights under the surface
>
> The bottom part of my figure is here:
>
> https://s32.postimg.org/7y35aegp1/sample.png
>
>
> I'm using this settings at the beginning of my POV file
>
> Is there a way to fix that? I tried increasing the max_trace_level but it
> doesn't change, neither with the photons although I didn't specify photons in
> the main figure (it gets too slow)
>
>

I'd guess the first issue is the area_light is very near your object for 
the vectors you have specified(1). Probably the low quality radiosity 
settings play a part too. Changing the max_trace_level is unlikely to 
affect anything.

Suggest:
- Comment out all the photon stuff.
- Change the first line of your light source to say :

light_source { <-3, -3, 4.>*100

- Change your radiosity settings to:

radiosity { Rad_Settings( Radiosity_Final, on, on ) }

Aside: If the triangle count on your mesh is low you might also be 
seeing sharp edges in the base mesh - smoothing the mesh might help.

Bill P.
(1) - There is a circular / auto-orient area light option which is 
useful too.

Post a reply to this message

On 22-6-2016 12:28, aradan.d wrote:
> global_settings {
>       max_trace_level 30
>       assumed_gamma 1.5
>       radiosity { Rad_Settings( Radiosity_Fast, on, on ) }
>       photons { spacing 0.005 }
>       }

Most probably the splotchiness come from the radiosity settings. You 
should try at least these parameters:

error_bound 1.0 //start here and decrease slowly until splotchiness 
disappears
nearest_count 10 //or higher, up to 20
recursion_limit 2 //should be fine for this

You can increase count substantially without problems. If you use the 
Rad_Settings, set the two on's to off as you have no normals (have you?) 
nor media to take care of.

A smaller pretrace_end is also to be recommended: 0.01 or 0.005 for 
instance.

Take out the photons altogether.

The preset values are fine but generally you get better results with 
your own settings, adapted to your particular scene.

-- 
Thomas

Post a reply to this message

Thanks, the radiosity settings were the ones that caused the weird shadows.
Using an error bound of 1.5 makes it look really nice. I will also check the
other suggestions and see how they work.

Cheers

Post a reply to this message

On 06/22/2016 07:36 AM, William F Pokorny wrote:
>
> I'd guess the first issue is the area_light is very near your object for
> the vectors you have specified(1).

Wondering if maybe the near area light (like 36 discrete lights being 
placed so near) would still more or less work out OK because of jitter. 
In other words, now wondering if the jitter is in the area light to 
light space and not the pixel AA space(1).

It is! 15+ years using POV-Ray and I'm still getting my head around 
fundamental things...

The result in my local test was still noisy so suppose it might still 
affect radiosity. Increasing the area light count might help if true.

Bill P.

(1) For what it is worth, the following from the Making animation 
section of the documentation is what confused me about how jitter really 
works. "Jitter is a very small amount of random ray perturbation 
designed to diffuse tiny aliasing errors...". The jitter being in the 
area light to light space I believe means the ray to light perturbation 
can be any magnitude from really tiny to really large.

Post a reply to this message

> Hi, I'm trying to render a surface I exported from a VTK visualisation software,
> but there are some strange lights under the surface
>

Looking at your image, it realy look like radiosity artefacts.

Things that can help:
Increasing the count value and add the second parameter. Set that second 
parameter to about 10 to 20 times your count value. It set a pool of 
sampling directions that are used in rotation.
Try to use a smaller value for minimum_reuse. Don't go overboard, 
something like 0.001 should be enough. Default at 0.0015.
Use a smaller pretrace_end. You can go as low as 0.00125 for a large image.
Reducing error_bound may help, as reducing low_error_factor from the 0.5 
default value. Note that the value for that parameter are larger than 
that in rad_def.inc as they are based on the older default of 0.8 from 
version 3.5.


It don't look like your area light have problems.
You can increase the sampling a lot if you use the adaptive option.
using adaptive 0 turn it ON and start sampling using only the 4 corners 
points. adaptive 1 start sampling with an array of 3*3, adaptive 2 start 
at 5*5 and adaptive 3 at 9*9. Using adaptive, you can use arrays of 
65*65 with almost no slowdown over a 9*9 one.

If you want to use the orient option, you also must use a square array 
and use the circular option.



Alain

Post a reply to this message

On 22-6-2016 14:49, aradan.d wrote:
> Thanks, the radiosity settings were the ones that caused the weird shadows.
> Using an error bound of 1.5 makes it look really nice. I will also check the
> other suggestions and see how they work.
>

Alain's suggestions are very sound.

Something you may consider, even in a wip, is to use a micronormal on 
the ground plane. It will not necessarily decrease radiosity artefacts 
but it improves the view without to much additional render time. 
Something like normal {granite 0.5 scale 0.001} for example.

-- 
Thomas

Post a reply to this message