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

Started Mar 2, 2017 | Discussions
fPrime
fPrime Senior Member • Posts: 2,592
Glass is evil... the smoking gun has been found
11

This thread is the continuation of several earlier ones asking why some users still prefer Nikon's older AI, AI-S, and D lenses over Nikon's newer G and E lenses. Beyond their smaller size and lower cost, the reasons often cited include more faithful color fidelity and better color saturation for D-generation and earlier lenses. Broadly speaking, discriminating users find that many of Nikon's newer lenses render with flatter rendition and a yellow cast (as do many of Sigma's ART lenses). This post will explain the probable cause.

The core reason can perhaps best be expressed by the phrase "Glass is evil." The less glass there is in the lens, the better. Credit for the concept actually goes to Ken Wheeler, aka The Angry Photographer (TAP), who popularized it a few years ago on his YouTube channel. TAP postulated that the more elements you have in play the lower the overall light transmission of the lens, the more the fragile blue wavelengths are lost, and the more that all wavelengths are pushed out of phase with each other. What TAP didn't have at the time was the data to prove this.

Today that's changed. I've just downloaded and analyzed the complete test data for all of Nikon's optical glass. The results are illuminating to say the least. But before I divulge the contents, let me review the fundamentals. As you can see below, visible light occupies a relatively narrow band between 380nm and 720nm:

The perfect lens would capture the complete visible range save for one problem... short wavelengths in the violet to blue range are absorbed to varying degrees by the glass itself. Let's look at a common color transmittance curve for one of Nikon's newer high refractive index optical glass types compared to the older version:

Notice that before 550nm all wavelengths fall below 100% transmittance. Blue light sits at 475nm and is significantly degraded for this type of optical glass. Green sits at 510nm and is also affected, albeit less. And this is only the loss induced by a 1cm slice of this glass!

Now some of you might be interested to understand why blue wavelengths would be differentially absorbed in glass. This is because blue wavelengths are always the most fragile wavelengths when passing through a solid medium. Even in the days of film manufacturing the blue emulsion was placed on top of the green and red emulsions in order to effectively capture the blue channel. Sigma's Foveon sensor borrowed the concept in designing their sensor stack:

The reason is self evident. Just like in film and glass, the blue wavelengths are lost first when passing through silicon:

The correlation is clear. Blue is the canary in coalmine for measuring absorption.

Glass is a little more complicated than silicon by virtue of the fact that there are many different types of optical glass. In general, low refractive index glass is better at preserving the violet-blue-green range than high refractive index (HRI) glass. In order to draw better distinctions on refractive index (RI) values let's spend a moment to define the range in a few other ostensibly clear mediums. Vacuum has an RI of 1, air is 1.0003, water is 1.33, plate glass is 1.52, flint glass is 1.62. In lenses it's been common design practice now for decades to use higher RI optical glass where possible to minimize the weight and size of lenses. The reason is simple... the higher the refractive index of the glass is, the less glass it takes to bend light. HRI 1.6 optical glass was even doped with radioactive Thorium as early as 1939 in order to minimize the dispersion caused by the HRI glass.

These days newer substitutes for Thorium have been developed and the old RI 1.6 glass that was once considered HRI has now been superseded by glass with RI's that exceed 2.0. To wit, Nikon now doesn't even consider it's own glass HRI until the RI reaches 1.85. Let's see what their optical glass catalog looks like in 2017:

Refractive Index is on the y-axis, Abbe number (dispersion) on the x-axis

Nikon's 125 different optical glasses are displayed on the graph above. The low dispersion glass is in the lower left quadrant, the strongest HRI glass is in the upper right quadrant. 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. 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. This obviously excludes the one or two LD or ED elements added to control dispersion from having too many elements in play. But these bring their own phase problems.

Nikon have made all of their glass test data available for downloaded. I've used it to specifically study the spectral transmission values for the blue channel (their closest measured value was 480nm). The best Nikon glass was J-SK11 which transmitted 99.9% of blue through a 1cm slice. Not surprisingly it's RI was a lowly 1.56. The worst performer was J-LASFH17HS which only transmitted 94.5% of blue through a 1cm slice with an RI of 2.0. A more typical performance was given by J-BASF7 which transmitted 98% of blue at an RI of 1.70. A hypothetical lens that used 10cm of such "low" RI glass would minimally absorb 20% of the blue light if not more given a that blue light likely has a non-linear absorption rate (see how quickly silicon blocks blue light in silicon compared too the slower absorption rate for red and green).

The bottom line is that TAP was right all along about excess glass degrading blue light. 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 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, 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 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). Definitive proof will ultimately be difficult to get to here. But the smoking gun has been found.

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
DAVID MANZE Veteran Member • Posts: 5,305
Re: Glass is evil... the smoking gun has been found
13

Yawn.......continuation??

Personally, I was hoping that this subject would lay down and go away, most has already been said too many times. Those who believe one way or another, believe one way or another!

-- hide signature --

Dave's clichés

 DAVID MANZE's gear list:DAVID MANZE's gear list
Nikon D750 Nikon D500 Nikon AF-S DX Nikkor 16-85mm f/3.5-5.6G ED VR Nikon AF Nikkor 24mm f/2.8D Nikon AF Nikkor 28mm f/2.8D +16 more
fPrime
OP fPrime Senior Member • Posts: 2,592
Re: Glass is evil... the smoking gun has been found

DAVID MANZE wrote:

Yawn..

Personally, I was hoping that this subject would lay down and go away, most has already been said too many times.

Wise. Ostriches do the same!

Marianne asked for spectral data just before the last thread closed out. Now she has it.

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
DAVID MANZE Veteran Member • Posts: 5,305
Bury my head Oh no, I'm a proud mum to be!
3

Loads of elements are here around me! 

-- hide signature --

Dave's clichés

 DAVID MANZE's gear list:DAVID MANZE's gear list
Nikon D750 Nikon D500 Nikon AF-S DX Nikkor 16-85mm f/3.5-5.6G ED VR Nikon AF Nikkor 24mm f/2.8D Nikon AF Nikkor 28mm f/2.8D +16 more
Ben Stonewall Regular Member • Posts: 269
Zombie thread alert…
3

No text.

 Ben Stonewall's gear list:Ben Stonewall's gear list
Apple iPhone 6s Plus Apple iPhone 8 Plus 645 Pro Mk II +2 more
fPrime
OP fPrime Senior Member • Posts: 2,592
Re: Bury my head Oh no, I'm a proud mum to be!

DAVID MANZE wrote:

Loads of elements are here around me!

  Haha!  Too funny.

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
fPrime
OP fPrime Senior Member • Posts: 2,592
You obviously don't know this forum, lol
2

These threads are the only ones to ever top out at 150... every time.

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
Kubicide Regular Member • Posts: 456
Agree with your bottom line...
4

1. Yes you are overblowing this. No lens is going to limit 20% of the 'blue' spectrum (compared to what is transmitted in the red and green). Please go measure one to verify.

2. Yes you cannot prove anything without the facts which would include details on the lens design, materials and coatings, etc. This post is... err... interesting theory-crafting only.

3. Yes the theory - if there is one being proposed here - isn't valid.

So basically, yes, I totally agree with your bottom line! But it has nothing to do with modern or older designs.

Reilly Diefenbach
Reilly Diefenbach Forum Pro • Posts: 12,789
Re: Glass is evil... the smoking gun has been found
18

Snake Oil

 Reilly Diefenbach's gear list:Reilly Diefenbach's gear list
Nikon D500 Nikon D850
fPrime
OP fPrime Senior Member • Posts: 2,592
Re: Agree with your bottom line...
2

Kubicide wrote:

1. Yes you are overblowing this. No lens is going to limit 20% of the 'blue' spectrum (compared to what is transmitted in the red and green). Please go measure one to verify.

You're completely right about no lens being released with a 20% blue spectrum loss.  The manufacturer will try their best to filter some of the remaining spectrum with coatings to offset whatever blue absorption occurs from the elements.  But that also goes towards explaining the higher T-stops in later lenses.

Still, I find it interesting that that Sigma appears to only partially correct their yellow shift.  Nikon seems to also not fully correct it in the G lenses or 105E either.  Maybe these guys like warm lenses.  I certainly don't.

2. Yes you cannot prove anything without the facts which would include details on the lens design, materials and coatings, etc. This post is... err... interesting theory-crafting only.

Would be great to rip apart the lenses to test each element, but let's face it, that's not going to happen.

3. Yes the theory - if there is one being proposed here - isn't valid.

Unfortunately this conclusion exhibits flawed logic.  Theories don't required proofs.  Some simply await data or means to test.

So basically, yes, I totally agree with your bottom line! But it has nothing to do with modern or older designs.

Thanks, we'll disagree on that one.  It is a fact that many older primes have fewer elements and I believe they are better off for it in color saturation and color fidelity.

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
fPrime
OP fPrime Senior Member • Posts: 2,592
Re: Glass is evil... the smoking gun has been found
5

Reilly Diefenbach wrote:

I understand, big data and analysis scares you so much that you lost the ability to speak. Things will be fine again in the morning after your nap.

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
Kubicide Regular Member • Posts: 456
Re: Agree with your bottom line...

Well, it seems like you're not familiar with optical engineering which is okay. But, assuming you can make an apples-to-apples comparison, comments like the t-stop being higher would be better assumed simply due to the increased number of elements (used for correction - to get the higher performance). It's not from using 'thinner glass' or directly related to the refractive characteristics of the material. It's the increased number of surfaces more than anything else. (But even still many other factors have to be known and considered).

Didn't see any theory really in your post. If you can explain it in simple terms maybe that would help the debate.

Mackiesback
Mackiesback Senior Member • Posts: 5,135
Re: Glass is evil... the smoking gun has been found
2

Can't all this "rendering" be adjusted in post?

Ok then. Who wants ice cream?!?

 Mackiesback's gear list:Mackiesback's gear list
Nikon Coolpix A Nikon D50 Nikon D100 Nikon Df Nikon D810 +16 more
fPrime
OP fPrime Senior Member • Posts: 2,592
Re: Glass is evil... the smoking gun has been found

Mackiesback wrote:

Can't all this "rendering" be adjusted in post?

Ah, great question.  Color skew can certainly be adjusted and saturation can be boosted.  But then that's an artificial digital push versus the lens's natural analog rendering.  Personally I prefer to let my lenses speak versus post processing them.  I know not everyone feels that way, that's ok.

One thing that can't be fixed in post... phase compression by ED/LD glass if that has to be used to correct for the high dispersion of other elements.

Ok then. Who wants ice cream?!?

I have pistachio, only the best!

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
Kubicide Regular Member • Posts: 456
Re: Glass is evil... the smoking gun has been found
2

fPrime wrote:

Mackiesback wrote:

Can't all this "rendering" be adjusted in post?

Ah, great question. Color skew can certainly be adjusted and saturation can be boosted. But then that's an artificial digital push versus the lens's natural analog rendering. Personally I prefer to let my lenses speak versus post processing them. I know not everyone feels that way, that's ok.

One thing that can't be fixed in post... phase compression by ED/LD glass if that has to be used to correct for the high dispersion of other elements.

Ok then. Who wants ice cream?!?

I have pistachio, only the best!

fPrime

Oh my goodness! What is the world is "phase compression by ED/LD glass"????

Mackiesback
Mackiesback Senior Member • Posts: 5,135
Re: Glass is evil... the smoking gun has been found
5

fPrime wrote:

Mackiesback wrote:

Can't all this "rendering" be adjusted in post?

Ah, great question. Color skew can certainly be adjusted and saturation can be boosted. But then that's an artificial digital push versus the lens's natural analog rendering. Personally I prefer to let my lenses speak versus post processing them. I know not everyone feels that way, that's ok.

One thing that can't be fixed in post... phase compression by ED/LD glass if that has to be used to correct for the high dispersion of other elements.

Ok then. Who wants ice cream?!?

I have pistachio, only the best!

fPrime

Your version of Lightroom doesnt have a Phase Compression slider? Upgrade man!

 Mackiesback's gear list:Mackiesback's gear list
Nikon Coolpix A Nikon D50 Nikon D100 Nikon Df Nikon D810 +16 more
Nikonlinks Regular Member • Posts: 439
Re: Glass is evil... the smoking gun has been found
4

I am enjoying the discussion in layman's terms as I do not have enough first hand knowledge to have an opinion.  Whatever the technical reason, there is no doubt the lenses I own that are older/simpler designs produce different results than the newer ones.  My 28/2.8 AI-s and 105/2.5 manual focus lenses produce images that I like.  I can't imagine anyone invested in the Nikon system who would not own those two inexpensive gems - but I'm not giving up the new lenses either!

-- hide signature --

don
www.donferrario.com

Marianne Oelund Veteran Member • Posts: 7,390
Incomplete analysis => incorrect conclusion
38

fPrime wrote:

Reilly Diefenbach wrote:

I understand, big data and analysis scares you so much that you lost the ability to speak. Things will be fine again in the morning after your nap.

Very pretty post with lots of graphics, however, you're nowhere near arriving at the definitive answer yet.

You need to take an actual lens design, list the glasses used, calculate the transmissions for each type from the spectral data and lens thicknesses, combine those results into the complete transmission profile, and then you still need to apply the CFA filter responses for R, G and B.  The latter step is critical for various reasons, such as the fact that the blue channel of Nikon sensors cuts off at 420nm, so losses below that wavelength have zero impact to the image.

I'm gathering data for the 70-200/2.8 VR II right now, and will have the results of those calculations within a few days (you may be interested to know that particular lens uses 12 different glass types).  Then we can compare it to the complete-lens transmission for the three channels and determine how much of its total losses are attributable to "evil glass" as opposed to "evil interface reflections."

A few notes regarding glass types:

Nikon generally chooses glass types within the RI range of 1.45 to 1.85, at least for prime lenses and pro zooms.

Glass at n=1.5 and v=82 is Nikon's ED glass, type J-FKH1.  At n=1.45 and v=91, Nikon calls it "super ED," type J-FKH2.

And by the way, TAP doesn't call glass "evil" merely because of the color cast, in fact that's a relatively minor consideration.  He mainly accuses it of causing scattering (or "wrangling" as he calls it) that leads to loss of contrast and color saturation - which it doesn't do - and denies the critical importance of lens coatings.  If bulk glass were that destructive to light, long-distance optical-fiber links would be infeasible.

-- hide signature --

Source credit: Prov 2:6
- Marianne

fPrime
OP fPrime Senior Member • Posts: 2,592
Regarding phase
  1. Kubicide wrote:

fPrime wrote:

Mackiesback wrote:

Can't all this "rendering" be adjusted in post?

One thing that can't be fixed in post... phase compression by ED/LD glass if that has to be used to correct for the high dispersion of other elements.

Oh my goodness! What is the world is "phase compression by ED/LD glass"????

Light wavelengths get out of phase when strongly refracted by high dispersion elements... red, green, and blue travel at different speeds and arrive at different points hence Marianne's little joke about uncorrected HRI lenses producing a rainbow effect.

An ED or LD lens can be used to compensate for this induced dispersion somewhat by slowing down the faster wavelengths to bring them back into phase with the slower wavelengths. The problem with this compression is that it generally introduces is a flatter rendering from the often less than perfect reconstruction of the wavelengths involved. Some lens designs show it worse than others.

fPrime

 fPrime's gear list:fPrime's gear list
Canon EOS 5D Fujifilm FinePix S5 Pro Nikon D1X Nikon D200 Nikon D700
photoreddi Veteran Member • Posts: 7,973
Re: Glass is evil... the smoking gun has been found
6

fPrime wrote:

This thread is the continuation of several earlier ones asking why some users still prefer Nikon's older AI, AI-S, and D lenses over Nikon's newer G and E lenses. Beyond their smaller size and lower cost, the reasons often cited include more faithful color fidelity and better color saturation for D-generation and earlier lenses. Broadly speaking, discriminating users find that many of Nikon's newer lenses render with flatter rendition and a yellow cast (as do many of Sigma's ART lenses). This post will explain the probable cause.

The core reason can perhaps best be expressed by the phrase "Glass is evil." The less glass there is in the lens, the better. Credit for the concept actually goes to Ken Wheeler, aka The Angry Photographer (TAP), who popularized it a few years ago on his YouTube channel. TAP postulated that the more elements you have in play the lower the overall light transmission of the lens, the more the fragile blue wavelengths are lost, and the more that all wavelengths are pushed out of phase with each other. What TAP didn't have at the time was the data to prove this.

Today that's changed.

No, nothing has changed. TAP and his minions are as clueless as ever. Somehow when I think of you a "mini-meenion" comes to mind. TAP may say "Glass is evil" but it's TAP and his gang of gullibles and trolls that are having a play with evil. Lucky for us they can't shoot straight, they only end up shooting themselves in the foot.

Keyboard shortcuts:
FForum MMy threads