On 5/16/19 3:16 AM, IGM wrote:
> Hi!
> I'm trying to use the user_defined camera, but the explanation in the reference
> and the examples I find around are still obscure to me. Is there any for-dummies
> introduction? Is there anyone who can try to describe its use in different
> words?
> 
> Thank you
> igmar
> 
> 
Have you played with the v3.8 sample scene? It was aimed to be pretty 
simple and has a defined perspective camera too so you can see - in a 
more usual way - the shapes which the user_defined inward cylindrical 
camera is 'viewing' as rays are defined in a rotating fashion around the 
y axis from top to bottom.

<install location>/scenes/camera/user_defined.pov

or on github at:

https://github.com/POV-Ray/povray/tree/master/distribution/scenes/camera

The location block defines the origin of each ray. The direction block 
the direction from each ray. Change, for example, the { y } in the 
locations block of the example scene to { y+0.5 } and render the scene 
anew. (The inward looking cylindrical set of rays now sits higher in y 
relative to the shapes in the scene.)

Bill P.

Post a reply to this message

William F Pokorny <ano### [at] anonymousorg> wrote:
> Have you played with the v3.8 sample scene?

Yes, but maybe I didn't spend enough time with it...
Thank you, I will give it another try.

igmar

Post a reply to this message

After some testing, I've a question: with user_defined camera, camera
transformations (rotate and translate) won't work. Once created the three
functions{} (for direction and location) suitable to obtain a particular effect,
is there a way to "rotate" the whole camera? The insertion of translation is
very simple, but for rotation things seem more complicate..

Post a reply to this message

"IGM" <iar### [at] gmailcom> wrote:
> After some testing, I've a question: with user_defined camera, camera
> transformations (rotate and translate) won't work. Once created the three
> functions{} (for direction and location) suitable to obtain a particular effect,
> is there a way to "rotate" the whole camera? The insertion of translation is
> very simple, but for rotation things seem more complicate..

By the way, the second example found in
"http://wiki.povray.org/content/Reference:Camera#User_defined_projection"
uses "rotate", but it has no effect...:

// Only direction functions specified
#declare Camera01 = camera {
    user_defined
    direction {
      function { x }
      function { y }
      function { 1 }
    }
    location  <0,0,-2>
    rotate    y*5
}

Where am I wrong?
igmar

Post a reply to this message

On 5/18/19 3:58 PM, IGM wrote:
> "IGM" <iar### [at] gmailcom> wrote:
>> After some testing, I've a question: with user_defined camera, camera
>> transformations (rotate and translate) won't work. Once created the three
>> functions{} (for direction and location) suitable to obtain a particular effect,
>> is there a way to "rotate" the whole camera? The insertion of translation is
>> very simple, but for rotation things seem more complicate..
> 
> By the way, the second example found in
> "http://wiki.povray.org/content/Reference:Camera#User_defined_projection"
> uses "rotate", but it has no effect...:
> 
> // Only direction functions specified
> #declare Camera01 = camera {
>      user_defined
>      direction {
>        function { x }
>        function { y }
>        function { 1 }
>      }
>      location  <0,0,-2>
>      rotate    y*5
> }
> 
> Where am I wrong?
> igmar
> 
> 

I think your right the documentation doesn't match the functionality 
with respect to camera modifiers. As to what should be or not, others 
better know.

I've not thought about a functionally defined camera set up where the 
functions enable later rotation. With functions suppose possible though 
complicated. I'll let the idea rattle around my skull for a while.

I was involved in the original ODS work that led to creation of the 
user_defined camera and I created the shipped sample scene. But, I've 
never worked in the camera source code so I don't know it - specifically 
how the modifiers get applied with cameras / or camera rays. Having that 
functionality with user_defined cameras would be convenient.

I've used the user_defined camera extensively in my own work, but always 
straight up never needing to transform the camera after the functional 
definition. I have a vague recollection of someone else posting the 
transform stuff didn't work for user_defined cameras (or 'maybe' some 
other camera type) and an expert's reply. I had no luck turning it up in 
a couple quick searches.

If you need some user defined, transform-able camera immediately, the 
mesh camera is another kind of user defined camera where transforms - 
applied to the mesh itself at least - should work fine. Unfortunately 
the mesh camera's mesh load (parse time) tends to be expensive. And 
don't forget Jerome's hgpovray38 has additional inbuilt cameras.

Bill P.

Post a reply to this message

Hi,
I'm following also the discussion
http://news.povray.org/povray.general/thread/%3CXnsA9E3A3F647A49seed7%40news.povray.org%3E/

The approach, in my opinion, is still far from being user friendly, but I
_think_ that the problem is similar to mine.

igmar

Post a reply to this message

"IGM" <iar### [at] gmailcom> wrote:
> Hi,
> I'm following also the discussion
>
http://news.povray.org/povray.general/thread/%3CXnsA9E3A3F647A49seed7%40news.povray.org%3E/
>
> The approach, in my opinion, is still far from being user friendly, but I
> _think_ that the problem is similar to mine.
>
> igmar

Yes, I recall briefly looking at that thread.
TOK has a very thorough understanding of a lot of the underlying math, and
excels in applying it and translating it into code.

Since this (in my mind) seems very similar to isosurfaces, you might take a look
at
http://www.econym.demon.co.uk/isotut/substitute.htm

to see how various transforms are implemented in functions.


As for rotations, I looked into the transforms for that a number of times.
There should be some useful links in this thread
http://news.povray.org/povray.general/thread/%3Cweb.5b9ed475f76fb4baa47873e10%40news.povray.org%3E/

as well as others to show you how to change the expressions for x, y, and z to
rotate the function you want to use.

Post a reply to this message

On 5/20/19 12:57 PM, IGM wrote:
> Hi,
> I'm following also the discussion
>
http://news.povray.org/povray.general/thread/%3CXnsA9E3A3F647A49seed7%40news.povray.org%3E/
> 
> The approach, in my opinion, is still far from being user friendly, but I
> _think_ that the problem is similar to mine.
> 
> igmar
> 
See post today to povray.binaries.images with the title:
"Method to apply transforms to user_defined camera."

It's at least a start on something useful that works today.

Bill P.

Post a reply to this message

William F Pokorny <ano### [at] anonymousorg> wrote:
> On 5/20/19 12:57 PM, IGM wrote:
> > Hi,
> > I'm following also the discussion
> >
http://news.povray.org/povray.general/thread/%3CXnsA9E3A3F647A49seed7%40news.povray.org%3E/
> >
> > The approach, in my opinion, is still far from being user friendly, but I
> > _think_ that the problem is similar to mine.
> >
> > igmar
> >
> See post today to povray.binaries.images with the title:
> "Method to apply transforms to user_defined camera."
>
> It's at least a start on something useful that works today.
>
> Bill P.

I tried the macro by Olav, but I found an unexpected behaviour:
https://news.povray.org/povray.general/thread/%3CXnsA9E3A3F647A49seed7%40news.povray.org%3E/?mtop=427321&moff=10

igmar

Post a reply to this message

On 6/4/19 3:16 AM, IGM wrote:
> William F Pokorny <ano### [at] anonymousorg> wrote:
>> On 5/20/19 12:57 PM, IGM wrote:
> 
> I tried the macro by Olav, but I found an unexpected behaviour:
>
https://news.povray.org/povray.general/thread/%3CXnsA9E3A3F647A49seed7%40news.povray.org%3E/?mtop=427321&moff=10
> 
> igmar
> 
I noted that you'd run into an issue, but not dug any deeper. If I find 
a little time and energy, I'll take a closer look. It'll be later this 
week - at best.

I have been thinking about the general problem as I go about life. 
Probably extra standard set up would need to be always be available with 
user defined cameras - more vectors or whatever? - and then the camera 
functions written against that extra stuff. The transform mechanisms 
would act on the extra 'stuff' and the the camera's functions having 
been written against that 'stuff' would too transform in a defined way.

I mean, with functional definition of just rays origins and directions, 
nothing stops users from creating sets of rays that have nothing in 
common within one user defined camera. A set of camera rays might just 
be generating some information stored as an image that some other 
process later interprets perhaps controlling shape movement in some 
animation say. What would transforms there mean or ever be used to do...

Aside: Since creating that depth map patch that can capture information 
in addition to ray depths, I've found myself playing with it and related 
ideas. Thinking in particular of coupling that functionality with user 
defined cameras. We can now shoot rays collecting all sorts of 
information that might, for example, allow us to create a file of 
photons where POV-Ray's internal shooting process isn't involved at all. 
A file of photons loaded an just creating some end image effect...

Another thought I've had is some sort of two step process where an 
explicit camera rig consisting of explicit splines for rays origins and 
directions are defined(1). We transform then those splines. The user 
defined camera functions become then nothing more than indexing 
functions into the splines.

(1) Perhaps adding shapes in the camera rig too. Something I first 
hacked at with the newer lemon shape was to use it to create a curved 
arc of traces in a two step process. Two linear splines defining the 
origin and direction shooting traces at the inside of the lemon - its 
end points the same as the direction spline. The collected intersections 
and normals were captured and then themselves used as origins and 
directions for the actual inward arc of traces I wanted. Basically I got 
lazy and off loaded a bunch of math onto POV-Ray itself. Maybe similar 
tricks can be used with user defined cameras that enable transforms too.

Bill P.

Post a reply to this message