>
> >> The camera/mirror sphere/HDRshop method can handle every advanced
> >> scene illumination (should even work with V6 GI) but it's
> very laborious...
> >> Mark
> >
> > The "camera/mirror sphere/HDRshop method" ? I'd be
> guessing you are
> > talking about using RS to capture HDRi images , in the same way one
> > would do so , in real life ? However , it is just a guess .
> >
>
>
> 100% correct!
>
> > The "very laborious" is also a bit of a mystery to me . Was this
> > method covered here and I missed it's journey into the void, or is
> > this considered to be 'common knowledge' , somehow ?
> >
> > thanks in advance
> >
> > garry
>
>
> Here's a brief description:
>
> - make a small analytical sphere with mirror material (just
> color=0 and
> reflection=1)
> - put a camera exactly underneath it and zoom in on the sphere
> - put these in a level, and place it at a strategic position
> in the scene and render. Make the camera Current if you
> render to a file.
>
> I uploaded an example scene with such a setup:
> http://www.athanor3d.com/pub/pano_volfog.zip
>
> Next, the HDRShop part. I noticed that they went commercial,
> how about USD 400 for a single license and USD 16500 for a
> 100-seat license!
> http://projects.ict.usc.edu/graphics/HDRShop/
> Version 1 is still free however, for non-commercial use.
>
> OK, here's the procedure:
> - load a fisheye image rendered with the above method
> - set Select->Draw Options to Circle
> - draw a circle as accurately as possible around the sphere's edge
> - crop the image (Image->Crop)
> - Now the fun part: open the Image->Panorama->Panoramic
> Transformations dialog
> - Source image should be the loaded image, Format = Mirrored Ball
> - click the Arbitrary Rotation button, then Settings: set
> X-axis to 90 degrees
> - Destination image: new image, Format = Latitude/Longitude
> - OK and save!
>
> Now you have a similar projection to the UVimage evaluated
> one but with full shading.
>
> I applied an image generated this way in RS to view it
> http://www.athanor3d.com/pub/OGLpanoviewer.zip
> Higher quality can be achieved with higher resolutions, but
> the bottleneck is the 1024 max OGL texture resolution, for
> rendering it can be much higher, of course.
>
>
> hope this helps,
> Mark
Nice one Mark, I just went through the whole sequence. Had to adjust the cam
for the mirroring bal, it's aspect is at 0, has to be 1, but that's ok.
The clear advantage above the UVimage method is that a full rendered image
is the result, with atmospheric effects and all.
The drawback however is that the resulting image covers only half the
sphere, where the UVimage method evaluates the full scene. I suppose that
could be overcome by a second rendering, and merge the two images in an
image editor.
Don't know what I like best, depends on the use I suppose.
One of the uses that you have mentionend is 360 degree HDR images for GI
rendering.
The UVimage method evaluates the illumination values from the reflective
sphere and puts them in a BMP image. If we would/could scale the
illumination values for the chrome material in a number of steps (7 or so)
and assemble those in HDRshop? Would that make a HDR image?
Regards,
Frank Bueters
