On 2022-02-28 17:56 (-4), Bald Eagle wrote:
> 
> [snip]
> So --- we might need a real-life Cornell Box with standardized illumination to
> eliminate the many influential variables.

I believe I have figured out a setup for photographing metals while
minimizing environmental biases.  Alas, lack of availability of some
metals has been a disincentive to me for proceeding with measurements of
those that I have on hand.

> [snip]
> 
> Not sure what else we have, but I'd say that at some point in the future we
> ought to strive to have have iron, steel, nickel, aluminum, tin, and zinc.
> And maybe titanium nitride.

How far should we extend the scope of the standard include files?
Should we even discard the current colors altogether and start with a
new set?  Or should we limit ourselves to the original goal of improving
the finishes?  The only reason I brought up the colors was because they
are significantly changed by the improved textures and the scene gamma
change.

> I'm sure they're all wildly similar, but maybe trying to look at the real world
> objects and figure out rendering textures for them might teach us to see what
> we're overlooking, discover better lighting and environment for objective
> observation, and maybe lump a few of those metals/platings into a catch-all
> texture that works for all of that group.

I would be very suspicious of a catchall finish.  The variability of
metallic textures is probably why metals.inc contains 5 finishes, why
even those 5 were culled from 9, why golds.inc's 5 finishes are
different from those of metals.inc, and why textures.inc has even more
metallic textures, most of which ignore the catchall finish that's
defined in that very same include file!  But the finishes F_MetalA
through F_MetalE should be sufficient for most purposes.

I wrote RC3Metal for the Object Collection precisely to overcome the
shortcomings of a fixed set of metallic textures, in addition to
resolving the obsolescence of the standard textures.  These are the
metallic colors defined in RC3Metal; however, the colors were not
derived from measurements:

Deduced naively from online spectral data:
  aluminum
  warm gold
  yellowish gold
  nickel plating
  silver
Eyeballed:
  yellow brass
  pale yellow-orange brass
  soft yellow-orange brass
  deep orange brass
  chrome plating
  brand new copper
  a darker copper
  old copper, on the orange side
  copper-nickel
  dark stainless steel
  light stainless steel
  regular steel
  new zinc plating
  weathered zinc plating
Pure guesswork:
  new bronze of the kind used for Olympic medals or ancient mirrors
  3 dark bronzes

Not long after publishing RC3Metal 1.0, I learned two things from Ive
and Christoph: that conversion of spectral data to a color, especially
of metals, is not straightforward; and that human vision is even more
horribly unreliable than I thought, so eyeballing colors, especially of
metals, is insufficient.  RC3Metal's existing colors definitely need
revision.  The following is what I have so far.

Metals already measured, though not published:
  freshly minted USA pennies, plating 95% Cu, 5% Zn

Metals already photographed for measurement (sampling quality unverified):
  yellow brass doorknob
  deep orange brass shelf mount
  chrome plated bathroom fixture
  shiny nickel bathroom fixture
  light stainless steel knife

Metals available, but not yet photographed:
  aluminum foil
  pale yellow-orange brass binder clip (probably in storage)
  soft yellow-orange brass clock
  old USA pennies, on the orange side, plating 95% Cu, 5% Zn
  dark stainless steel spoon or fork
  steel scissors
  copper-nickel plated coins
  weathered zinc plated fence posts

Metals that may be lying around:
  regular steel, if it's not rusted out
  indoor zinc plating

Metals photographed behind glass or from afar:
  gold of unknown purity

Metals not in hand, but easily purchased:
  moderately new copper piping
  lead (I'd be nervous keeping it around, though)
  titanium nitride
  new zinc plating

Metals of dubious availability:
  mirror or medal quality new bronze
  dark bronzes that aren't spray-on
  14K gold
  18K gold
  24K gold
  silver
  mercury
  niobium
  pewter
  platinum
  rhodium plating
  ruthenium plating
  speculum
  tin

Post a reply to this message