Cousin Ricky <ric### [at] yahoocom> wrote:
> On 2025-03-14 13:59 (-4), Cousin Ricky wrote:
> > On 2025-03-14 08:16 (-4), Mr wrote:
> >>
> >> a) From afar, the brightness seems sligthly off... as if the
> >> texture-to-lighting-to-final-image gamma was somehow too high ...doubled? / not
> >> reverse corrected as should happen in linear workflows allowing best contrast
> >> preservation or maybe a media is at play in the room?
> >
> > Probably just the overall lighting.  I did not use radiosity, and the
> > ambient may be a little high.
>
> Two more renders.  The first has a darker environment--brown walls,
> although they don't show in the image.  Radiosity is still not used, but
> my pre-fab environment computed a darker ambient because of the darker
> walls.  The render time was 8 minutes for the photons and 33 minutes for
> the render.
>
> The second render is an outdoor patio scene using radiosity.
>
> The glass design was tweaked for exactly 215 ml capacity, and the wine
> level was tweaked to get exactly 50 ml.  Still no meniscus (yet).
>
> >> b) Then opening the image at maximum size... Doesn't the anti-aliasing look to
> >> be off? or too low? Is it Method 3 and/or no AA value combined with a camera
> >> aperture? Where this shows most is on edge of the glass where some kinds of
> >> fireflies appear.
> >
> > It's the hyper-white of the highlights.  Since POV-Ray 3.6, color values
> > are clipped post-anti-aliasing, which results in jaggies on bright
> > highlights.  The only way to solve this is post-processing.  Sam Benge
> > and I have both written tools in POV-Ray SDL that can do this, although
> > thus far only Sam has published his tool.
> >
> > Prior to 3.6, colors were clipped before anti-aliasing, which made for
> > anemic-looking, albeit jaggies-free, highlights.  I actually prefer the
> > post-clipping of the newer POVs, because that can be fixed--especially
> > if the image is rendered to EXR.
>
> Both renders were post-processed to add light bleeding.  The colored
> fireflies on the rim and at the base of the stem are from the
> high-dispersion crystal glass material.  I think the random appearance
> results from complex multiple reflections that are too fine to bring out
> the true detail at this resolution, and they are not bright enough for
> my post-processing algorithm to smooth out.

I think the contrast curve is much better now, especially on the glowy
reflection of glass and wine surface, A slight tweak could be to allow
absorption to restitute a little further penetrating light rays to have more of
a gradient in the red shadowed area, currently appearing too dark as if the wine
was almost opaque.

Post a reply to this message