Hi(gh)!

While rendering a new version of my Phobos fly-around movie, I pretty 
soon found that the mesh2 of the Martian moon is cut off on one or, 
sometimes, even on both sides. I tried many settings of overall scaling 
values, even translated the whole Solar System so that Mars and its 
moons were placed close to the origin, thus avoiding extreme coordinate 
values - but whatever I do, the mesh2 is clipped once again (see 
attached image below)!

What can I do to get rid of this?

See you in Khyberspace!

Yadgar

Post a reply to this message

Attachments:
Download 'povsolar-at-sc.png' (253 KB)

Preview of image 'povsolar-at-sc.png'

Am 01.01.2015 um 13:02 schrieb "Jörg \"Yadgar\" Bleimann":
> Hi(gh)!
>
> While rendering a new version of my Phobos fly-around movie, I pretty
> soon found that the mesh2 of the Martian moon is cut off on one or,
> sometimes, even on both sides. I tried many settings of overall scaling
> values, even translated the whole Solar System so that Mars and its
> moons were placed close to the origin, thus avoiding extreme coordinate
> values - but whatever I do, the mesh2 is clipped once again (see
> attached image below)!
>
> What can I do to get rid of this?

Make sure that the scene is translated so that the camera is close to 
<0,0,0>; the further away you go, the more problematic the bounding 
becomes, as it uses low ("single") precision floating point numbers to 
minimize the bounding tree memory footprint.

Post a reply to this message

Hi(gh)!

Am 01.01.2015 um 18:35 schrieb clipka:
> Am 01.01.2015 um 13:02 schrieb "Jörg \"Yadgar\" Bleimann":

> Make sure that the scene is translated so that the camera is close to
> <0,0,0>;

O.k., I understand - currently the camera is still at about <6000, 0, 
6000> (or so), while Mars sits at the origin.

> the further away you go, the more problematic the bounding
> becomes, as it uses low ("single") precision floating point numbers to
> minimize the bounding tree memory footprint.

May that change in the future? Will there be POV-Ray versions which use 
double or even long double floats for bounding?

A smaller epsilon value also sounds tempting...

See you in Khyberspace!

Yadgar

Post a reply to this message

On 1-1-2015 23:32, "Jörg \"Yadgar\" Bleimann" wrote:
> A smaller epsilon value also sounds tempting...

[hint] That should be easy to implement by the Team shouldn't it? and 
would solve a lot of related problems ;-) After all, a hack in that 
sense was made some years ago to accommodate the Ring World.

-- 
Thomas

Post a reply to this message

On 01/02/2015 04:45 AM, Thomas de Groot wrote:
> On 1-1-2015 23:32, "Jörg \"Yadgar\" Bleimann" wrote:
>> A smaller epsilon value also sounds tempting...
>
> [hint] That should be easy to implement by the Team shouldn't it? and
> would solve a lot of related problems ;-) After all, a hack in that
> sense was made some years ago to accommodate the Ring World.

Thorsten has already addressed this, and it appears (tentatively) to be 
a no-go.  See Flyspray #297.

Post a reply to this message

On 2-1-2015 15:38, Cousin Ricky wrote:
> On 01/02/2015 04:45 AM, Thomas de Groot wrote:
>> On 1-1-2015 23:32, "Jörg \"Yadgar\" Bleimann" wrote:
>>> A smaller epsilon value also sounds tempting...
>>
>> [hint] That should be easy to implement by the Team shouldn't it? and
>> would solve a lot of related problems ;-) After all, a hack in that
>> sense was made some years ago to accommodate the Ring World.
>
> Thorsten has already addressed this, and it appears (tentatively) to be
> a no-go.  See Flyspray #297.
>
Hmmm right. I see. It did help a bit though but seemingly it is not the 
ultimate solution. I retract my comment :-)

-- 
Thomas

Post a reply to this message