Both the Barracuda and Bel Air on display.  Fixed the mesh smoothing on the
Barracuda door and textured the light lenses (not really noticeable here
though)

-tgq

Post a reply to this message

Attachments:
Download 'belaircuda.jpg' (246 KB)

Preview of image 'belaircuda.jpg'

"Trevor G Quayle" <Tin### [at] hotmailcom> wrote:
> Both the Barracuda and Bel Air on display.  Fixed the mesh smoothing on the
> Barracuda door and textured the light lenses (not really noticeable here
> though)
>
> -tgq

They're beauties!

Looks like they and the enclosure need to be scaled up a bit, though, to
match the background. OR, perhaps putting the camera further back with more
of a telephoto lens would do the trick.

Perspective and lens-matching isn't easy! I was wondering: Do HDRI images
or light probes "contain" that information in some accessible form? Or is
it just up to the user to place any synthetic objects by trial-and-error?

Ken W.

Post a reply to this message

"Kenneth" <kdw### [at] earthlinknet> wrote:
> They're beauties!
>
> Looks like they and the enclosure need to be scaled up a bit, though, to
> match the background. OR, perhaps putting the camera further back with more
> of a telephoto lens would do the trick.
>
> Perspective and lens-matching isn't easy! I was wondering: Do HDRI images
> or light probes "contain" that information in some accessible form? Or is
> it just up to the user to place any synthetic objects by trial-and-error?
>
> Ken W.

Thanks.  Yeah, when the background is visible, getting scale to 'look' right
is very difficult.  The easiest way to get around it is to try to hide the
background.  Other than that, I think, as suggested, telephoto/wide zooming
is what needs to be used, but probably the opposite way you suggested.  This
image uses a 45deg viewing angle, so you see 45 degrees of the background.
Moving the camera closer and widening the view, the apparent size of the
objects relative to the image can stay relatively the same (discointing
distortion effects) while seeing more of the background.

As for automatically getting the depth info, I don't think it's possible.
(This isn't a file format issue, kits for any 360 panoramic image, which
can be non-HDR.  HDRs simply contain more luminance information about the
scene).  Everything radiates out from the center of the probe.  This can
cause difficulties with close/far objects in the probe and the scene,
especially when simply mapped to a sphere (eg. parallax issues depending on
view point).  Obviously, probes with no relatively near objects work best to
minimize the effects.  You can also try to 'recreate' the basic geometry of
the background, and sperically map this rather than using a sphere
(difficult if you know nothing of the dimensions of where the probe was
created, but it can be done through trial and error).  The simplest way to
deal with these spherical maps is to try to keep the mapping sphere
realtively large compared to the scene, and try to keep the objects and
line of view close to the centre (centering on the camera location gives
the least parallax problems for direct line of sight, centering on scene
objects give the least problems for reflection).  If there are any close
objects in the spherical map, try to hide them from view as best you can.

All that said, I may try to adjust the viewing angle of the scene to see how
it works out...

-tgq

Post a reply to this message

"Trevor G Quayle" <Tin### [at] hotmailcom> wrote:
> All that said, I may try to adjust the viewing angle of the scene to see how
> it works out...
>
> -tgq

A quick render with the viewing angle increased to 60deg and the camera
moved closer.  Perspective looks a little better.  I think one of the major
problems with the other version is not so much the apparent scale factor,
but the relative sharpness of the focus of the background and near objects.
 Our brain tells us that for the focus to look the way it does, the
background must be closer, which then dictates the realtive scale of the
objects/scene (i.e. for that kind of zoom, we expect a blurrier
background).

-tgq

Post a reply to this message

Attachments:
Download 'belaircuda2.jpg' (289 KB)

Preview of image 'belaircuda2.jpg'

"Trevor G Quayle" <Tin### [at] hotmailcom> wrote:

>
> A quick render with the viewing angle increased to 60deg and the camera
> moved closer.  Perspective looks a little better.

Oooh, much nicer!

> I think one of the major
> problems with the other version is not so much the apparent scale factor,
> but the relative sharpness of the focus of the background and near objects.
>  Our brain tells us that for the focus to look the way it does, the
> background must be closer, which then dictates the realtive scale of the
> objects/scene (i.e. for that kind of zoom, we expect a blurrier
> background).
>
> -tgq

Hey, try blurring the foreground to match!  :-P

Ken

Post a reply to this message

"Trevor G Quayle" <Tin### [at] hotmailcom> wrote:

>
> As for automatically getting the depth info, I don't think it's possible.
> (This isn't a file format issue, kits for any 360 panoramic image, which
> can be non-HDR.  HDRs simply contain more luminance information about the
> scene).

Good to know.  Not having delved into HDRI yet, I had assumed the images
were more "complex" (whatever that means!)

> Everything radiates out from the center of the probe.  This can
> cause difficulties with close/far objects in the probe and the scene,
> especially when simply mapped to a sphere (eg. parallax issues depending on
> view point).  Obviously, probes with no relatively near objects work best to
> minimize the effects.

Thanks, a good tip.

> You can also try to 'recreate' the basic geometry of
> the background, and sperically map this rather than using a sphere
> (difficult if you know nothing of the dimensions of where the probe was
> created, but it can be done through trial and error).

Yes.  As I understand it, many of the CGI special effects facilities in
Hollywood and elsewhere do this, by taking detailed measurements on-set
with laser rangefinders or some-such.

> The simplest way to
> deal with these spherical maps is to try to keep the mapping sphere
> realtively large compared to the scene, and try to keep the objects and
> line of view close to the centre (centering on the camera location gives
> the least parallax problems for direct line of sight, centering on scene
> objects give the least problems for reflection).  If there are any close
> objects in the spherical map, try to hide them from view as best you can.

Very useful info, much appreciated!  Thanks.

Ken

Post a reply to this message