Glass is evil... the smoking gun has been found

Started Mar 2, 2017 | Discussions thread
Marianne Oelund Veteran Member • Posts: 7,390
Two smoking guns, rather
34

fPrime wrote:

TAP postulated that the more elements you have in play the lower the overall light transmission of the lens,

This understanding existed well before TAP did.

the more the fragile blue wavelengths are lost, and the more that all wavelengths are pushed out of phase with each other.

No, "phase" which as used here refers to wavelength-dependent axial and lateral displacements, does not steadily degrade with an increase in the number of elements.  With careful design, it actually improves.  With clueless design, such as using nothing but HRI glass for all elements, it would of course degrade - but no professional designer does this.

What TAP didn't have at the time was the data to prove this.

I'm sure he did.  You are bringing nothing new to the table here.  Detailed glass specs have been available for decades.

Today that's changed. I've just downloaded and analyzed the complete test data for all of Nikon's optical glass.

That would only change things for you personally.

You can quickly tell that 78% of all Nikon's glass falls at or above the old HRI 1.6 specification. In fact 66% has an RI between 1.6 and 1.83 below Nikon's HRI cutoff. And 13% of it meets or exceeds Nikon's HRI 1.85 cutoff!

The average RI of their glass is 1.70 and the median is 1.68 which infers a particular Nikon interest in variants of these glasses. Nikon makes five RI 1.70 glasses alone. A nodal analysis of the graph reveals a definite population concentration between RI 1.6 and 1.80 further confirming this.

Statistical approaches are inappropriate to design and analyze lenses and other deterministic systems.

Nikon doesn't detail what glass it uses in each of its lenses but you'd have to be pretty naive to think most of it wasn't RI 1.6 or better.

The AFS 70-200/2.8G VR II has total element thickness of 90mm.  About 49mm of that has RI under 1.6 and another 8.8mm has RI right on the 1.6 boundary.  Who were you saying is naive?

This obviously excludes the one or two LD or ED elements added to control dispersion from having too many elements in play.

More elements does not correlate with more dispersion, unless you apply them carelessly.

But these bring their own phase problems.

No, it's the high-RI dense flint glasses that bring the phase problems.

A hypothetical lens that used 10cm of such "low" RI glass would minimally absorb 20% of the blue light if not more

Your hypothetical lens isn't representative of real lens designs, which do not venture into "20% if not more" territory.  Further, you are again completely disregarding the fact that absorption occurs across the entire spectrum, not just in the blue range, so the blue losses alone do not dictate the lens color; the color skew is less than the blue loss.

I'm sure that the "modern must be better" critics will say 1) the issue is overblown, no lens is missing 20% of the blue spectrum, or

You are overblowing it, not just by exaggerating numbers, but also by singling out blue absorption as the root of all evils.  We all know most lenses have a warm cast.  Many users prefer it, and if it's undesired, it's easily corrected.  Lens color skews are smooth and do not produce uncorrectable metamerisms.  Lighting sources, on the other hand, can and often do.

2) you can't prove anything because Nikon doesn't specify the type of glass it uses in lenses and you haven't taken apart the lens to measure the RI of the elements and transmission,

That's your own claim.

or 3) the theory isn't valid because it doesn't explain my outlier lens "X" having a magenta skew. LOL, I get it. Manufacturer's obviously correct for high dispersion effects in the barrel when needed by adding ED elements...

It's just not that simple.

it's even easier to correct for yellow color skew with filtered coatings when need. Filtering some red and green out as well just means an even higher T-stop (another hidden sign of poor light transmission in modern lenses).

There is in fact quite a balancing act carried out by lens designers to achieve the tradeoff they want between T-stop and color neutrality.

Definitive proof will ultimately be difficult to get to here.

No, it's not difficult at all.

But the smoking gun has been found.

Actually, two . . . and yours isn't smoking as much as you think it is.

Following is a study of the AFS 70-200/2.8G VR II.  I selected this lens as an extreme case, and also because I own one which I can test for color response.  Even for this complex 21-element zoom lens, overall losses across the spectrum are dominated by interface reflections, but the study does show that bulk glass losses are important, especially for the blue channel.  You are not entirely wrong in your avenue of research - but you do tend to mislead by exaggeration and overlooking or misunderstanding other factors.

Here is a snapshot of the spreadsheet containing the calculations.  I will not attempt to explain the formulas in detail now, but will answer specific questions later.  Each glass type has its own column of transmission figures.  The rows under the types divide the spectrum into bands which the glass spectral data was given for.  Column O contains the net glass transmission figures for each band.

To the right under "CFA % weight per band" are the weighting factors which describe the spectral shape of the camera's filters (taken from a D3s).  These numbers are scaled to add up to 100 (%) for each filter.  The dot product of the weighting factors with the overall figures in column O, gives the channel responses, again in %.

The results at lower right show bulk glass absorption runs around 7-8% in the red and green channels, and almost 14% in the blue channel - a skew of 6%.  The "Lens" row is channel transmission empirical measurements  for the complete lens, and the "Interface" row shows the transmission for the air-glass interfaces.  The latter are producing loss rates around 12% on average.  Without modern coatings, this figure could easily double, so it is clear that complex lens designs are highly dependent on their element coatings.

There are lessons here for everyone:  Glass transmission is a larger contributor than I expected, but lens coatings cannot be disregarded in analysis of lens efficiency and color.

-- hide signature --

Source credit: Prov 2:6
- Marianne

Post (hide subjects) Posted by
(unknown member)
(unknown member)
tko
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow