Disqus - Friends of planetMitch

Re: How the Fuji X-Pro2 Resparked my Passion for Photography

Patrick Chase — Thu, 22 Dec 2016 02:25:00 -0000

Why exactly do you think that SLR lenses need to be "thoroughly re-engineered" to work on mirrorless? What "optical demands" are you referring to?

In the case of wide-angles SLR lenses are more complicated than they need to be for mirrorless, because SLR wide-angles use retrofocus designs to satisfy the higher flange distance. That doesn't mean they won't work as well as on an SLR, though. Also, with the advent of affordable aspheric elements and advanced coatings the manufacturers have become quite good at mitigating the penalties associated with retrofocus designs.

If you're going to say something about "glass over the sensors" then please note that SLR lenses typically have quite long exit pupil to image plane distances (see Roger's compilation thereof), and that makes them much less susceptible to stack thickness variation than conventional lenses. This compatibility benefit actually arises directly from the aforementioned use of retrofocus layouts.

Your comments about filter thread diameter are complete gibberish. From a light collection perspective the entrance pupil diameter is what matters when comparing lenses of equivalent FOV in systems with different sensor sizes. If the entrance pupil diameter and FOV are the same, then the total flux (photons/sec) at the sensor will be the same, assuming similar transmission losses. The smaller sensor will have higher illuminance (flux per unit area) but that's offset by the correspondingly smaller pixel size. Total flux is what matters.

Re: Front Element Lens Protection Revisited

Patrick Chase — Thu, 29 Dec 2016 15:25:21 -0000

Do you have any sense for how/whether quality filters noticeably impact MTF on wider-angle lenses?

As you demonstrated in your stack height comparisons, the MTF impact of flat glass increases with angle of incidence. That's why lenses with short exit pupil to sensor distances are most sensitive to stack height.

The 50 mm lens that you tested has a maximum incident angle of ~23 deg. Do you think that the result would have been different on, say, a 16-35 at the wide end (~54 deg)? Beyond that the point is moot, since you typically can't use filters on wider lenses anyway.

Re: Front Element Lens Protection Revisited

Patrick Chase — Thu, 29 Dec 2016 16:59:27 -0000

That's true if the filter is perfectly flat and uniform. If it isn't then wider FOV will magnify the impact IIRC.

Re: Real World Testing of the New Canon 16-35mm f/2.8L III

Patrick Chase — Sat, 07 Jan 2017 23:46:36 -0000

EDITED to reflect Brandon's feedback (he knows optical HW better than I do).

Lens designers use vignetting specifically to suppress aberrations on the periphery (by blocking peripheral rays). If it weren't done then the wide-aperture corner performance would probably crater. Most wide-angle lenses vignette to some degree for that reason. I should be clear that when I say this I'm distinguishing aperture-dependent "vignetting" from aperture-independent "falloff", which arises for different reasons.

Within digital workflows falloff and blurring are actually strongly coupled and exchangeable. They both cause information to be lost. Falloff does so in the intensity domain (signal is reduced, which means that the signal/noise ratio is reduced) while blurring does so in the spatial domain (neighboring pixels are correlated). The two forms of information loss can be exchanged in post, by sharpening and denoising. Sharpening (partially) removes spatial correlation but also amplifies noise and thereby degrades S/N. Denoising improves S/N but also increases spatial correlation.

I have the Canon 16-35 III, and based on a 15-year career in image-processing (including sharpening and denoising algorithm design), I think that Canon struck a reasonable balance with that lens given the current state of the post-processing art. Yes, the extreme corners are vignetted, but they don't need anywhere near as much sharpening as with some other lenses, and the final result (after boosting the corners and sharpening to correct for aberrations) is as good as any other zoom in that class.

Re: Real World Testing of the New Canon 16-35mm f/2.8L III

Patrick Chase — Mon, 09 Jan 2017 01:14:56 -0000

Edited my comment to reflect your correction. Thanks!

Re: Lensrentals.com Review of the Sigma 85mm f/1.4 Art Series Lens

Patrick Chase — Wed, 25 Jan 2017 20:05:05 -0000

I don't agree with the assertion that the Sigma's size is "seemingly unnecessary".

It you look at the optical formula it appears that they're using a bunch of elements well out in front of the entrance pupil to manage off-axis aberrations (real OEs please feel free to correct me). To achieve that with a 60 mm entrance pupil the lens has to be wide, and to do it at 85 mm FL means that it has to be long.

The Otus has the same filter size, but is lighter, likely because it has fewer elements/groups. Of course you probably don't need as much glass if most of it (7 of 11 elements) is LD or aspheric...

EDIT: The Otus is slightly heavier, not lighter. 1.2 kg vs 1.13 for the Sigma. All of the weights in this article appear off.

Re: Lensrentals.com Review of the Sigma 85mm f/1.4 Art Series Lens

Patrick Chase — Wed, 25 Jan 2017 21:48:51 -0000

For use on film I agree.

For digital the brutal truth is that color is fixable, but loss of spatial information is not (sharpening can only amplify what remains, it can't bring back what is lost). Also Sigma's coatings have gotten a *lot* better.

Where the Otus really spanks the Sigma is in axial CA and spherochromaticism. Those 5 LD elements doing their thing, no doubt.

Re: Lensrentals.com Review of the Sigma 85mm f/1.4 Art Series Lens

Patrick Chase — Wed, 25 Jan 2017 22:52:00 -0000

One factual correction: The Sigma weighs 39.9 oz, which is 2.5 lbs, not 3.8 as stated in the review.

Also, the Canon 85/1.2 II weighs 2.3 lbs, not 2.9.

Re: Lensrentals.com Review of the Sigma 85mm f/1.4 Art Series Lens

Patrick Chase — Thu, 26 Jan 2017 13:43:23 -0000

One sample shot on an adapter doth not a valid comparison make.
The thing that is clear to me from that comparison is that the 85GM vignettes less, which is why the bokeh balls from point sources stay more circular moving off-axis. That wouldn't change from sample to sample.

Re: Lensrentals.com Review of the Sigma 85mm f/1.4 Art Series Lens

Patrick Chase — Thu, 26 Jan 2017 23:35:33 -0000

As a engineer (though not an OE) I would clarify that to "no engineer who is worth anything would add unnecessary size to a design unless forced to do so at gunpoint by marketing". It's been known to happen. For example look at all of the lenses with cosmetic metal shells around plastic innards...

Re: Lensrentals.com Review of the Sigma 85mm f/1.4 Art Series Lens

Patrick Chase — Sun, 29 Jan 2017 15:29:05 -0000

Indeed. I'd assumed that it went without saying that the Otus' liberal use of exotics was only feasible at an Otus price point :-).

Re: Opening up the New Canon 70-300mm f/4-5.6 IS II

Patrick Chase — Wed, 01 Feb 2017 12:43:14 -0000

You probably already know this, but for something that complex it's very much a team effort, spanning optics, mechanics, electronics, and (in modern lenses) controls and firmware.

Many companies additionally have a "systems" discipline dedicated to pulling everything together and optimizing the architecture as a whole. Canon seems to be doing a particularly good job on that last bit. Like the Fuji Roger tore down a while back, this strikes me as a very nice (and nicely balanced) design for a consumer lens.

Re: Opening up the New Canon 70-300mm f/4-5.6 IS II

Patrick Chase — Thu, 02 Feb 2017 12:50:16 -0000

The parts look to me like a variation on an "inchworm" motor: https://en.wikipedia.org/wi...

Re: MTF Tests for the Sigma BBL: The Big, Beautiful, 85mm Art Lens

Patrick Chase — Fri, 03 Feb 2017 15:03:30 -0000

I have a question about the measurements: Are you using a broadband light source for these? If so what distribution? If not what specific wavelength[s]?

I've looked through your blog posts from when you started working with the image bench, but didn't see that spelled out. My apologies if you've already documented this.

EDIT: The datasheet for your bench says "white LED or optional halogen" fed through a collimator. Is it one of those?

Re: MTF Tests for the Sigma BBL: The Big, Beautiful, 85mm Art Lens

Patrick Chase — Fri, 03 Feb 2017 18:47:55 -0000

That's reasonable and fairly common practice. I mostly asked because the Sigma is known to have a bit of axial CA and spherochromaticism, and was wondering how that would be reflected in your MTF results.

FWIW when I was an imaging system architect (working closely with OEs) we typically used through-focus curves at multiple wavelengths. You can learn a *lot* about how a system behaves that way. For example astigmatism can be distinguished from other anisotropic aberrations, and things like axial CA are easily identified. I also looked a lot at PSFs when designing digital filters, so I appreciate it when you post Olaf images as well...

EDIT: Thanks for posting the field of focus graphs. Those are also really useful for understanding the "why" instead of just the what (and if you convert the colors to back to MTF you can construct rough through-focus curves from them)

Re: MTF Tests for the Sigma BBL: The Big, Beautiful, 85mm Art Lens

Patrick Chase — Fri, 03 Feb 2017 22:49:20 -0000

The OEs I worked with mostly used Zemax.

I suspect that the PSF may be more useful to somebody (like me) who was/is in the business of correcting images than to an OE. Needless to say the Fourier transform of the PSF (SFR) is *very* useful in filter design, which is the use case I cited :-).

Here's an admittedly simplified example: If you have a lens with significant astigmatism, and you don't have great control of focus, then you probably don't want to overdo it with anisotropic sharpening based on the lens' behavior at nominal focus, for obvious reasons. If on the other hand the lens is consistently softer on one axis than the other through focus then you can be more aggressive. Seeing the PSFs as f(image distance) is really useful when making those sorts of judgments.

Re: MTF Tests for the Sigma BBL: The Big, Beautiful, 85mm Art Lens

Patrick Chase — Fri, 03 Feb 2017 23:56:15 -0000

Sorry about the repeat-reply, but here's a real-world example where PSFs were indispensable: Several years ago I was asked to help with a system that was having IQ issues, including grotesque sample variation. The simulated through-focus curves and the MTF measurements in production both looked passable, but the images from bad samples had glaringly obvious astigmatism (far worse than their production measurements said they did).

I "politely asked" the OE to simulate PSFs at 0.01 mm intervals through-focus, at which point the problem became obvious: This particular system had an off-axis all-reflective lens (and not a particularly simple one). As you are perhaps aware (I was not) those can exhibit astigmatism along axes other than tan-sag, in this case 45-135. Unfortunately the through-focus curves were tan-sag, as were the test patterns used for alignment and QA in production.

Arguably the OE screwed up royally by not generating 45-135 curves to begin with, and by not insisting on 45-135 targets in mfg, but once we were in that situation examination of the PSFs was the fastest way to debug it and figure out how to proceed (aligning with the right targets helped a lot, as you might expect. Also, off-axis systems are just different, such that there was a fair bit of learning and process tweaks involved).

EDIT: Fixed 0.1/0.01 typo. Also fixed "can exhibit astigmatism..." per Brandon's feedback.

Re: MTF Tests for the Sigma BBL: The Big, Beautiful, 85mm Art Lens

Patrick Chase — Sun, 05 Feb 2017 17:12:30 -0000

One last small remark on the topic of PSFs: The other place where I've found them useful is for simulation, particularly to get product buy-off for statistical/test limits. By this I mean that I convolved the spatially-variant PSF with reference images to model expected nominal- and worst-case behaviors.

A quick search of the literature reveals that this is a well-known technique (I don't think I knew about that literature at the time, but thought it such an obvious thing to do as to not be worthy of note). See for example http://www.imagescienceasso...

Re: Things You Didn’t Want to Know About Zoom Lenses

Patrick Chase — Fri, 10 Feb 2017 15:05:36 -0000

I'm going to go out on a limb and guess that the un-named prime you used for the set of 9 maps was a supertele, maybe a 400/2.8.?

Re: Things You Didn’t Want to Know About Zoom Lenses

Patrick Chase — Fri, 10 Feb 2017 15:53:03 -0000

Yeah, I realized it could be a 100 mm macro lens after I posted. I didn't know about the ripple issue when measuring superteles though.

Re: Things You Didn’t Want to Know About Zoom Lenses

Patrick Chase — Fri, 10 Feb 2017 16:03:53 -0000

I think that the measurements are extremely helpful to establish an acceptability range once you suspect a problem.

For example I recently bought a very highly-regarded zoom and thought I saw softness on one side, and particularly one corner on that side, at longer focal lengths in "real" images. I shot an ISO target, and observed that my sample had zero MTF at ~30 lp/mm and negative MTF (8 lines instead of 9) above that in the suspect corner.

You had published variability data for that lens, which made it easy to determine that that corner was >3 standard deviations below average, so I sent it back (the first time I've ever done so with a new lens btw). Without that data it would be a lot harder to make a rational decision.

Re: Things You Didn’t Want to Know About Zoom Lenses

Patrick Chase — Wed, 15 Feb 2017 20:47:00 -0000

The problem with that chart is that it's obviously a geometric simulation that doesn't include diffraction, and is physically impossible in the real world.

The theoretical upper limit for visible light (using 560 nm as a nominal) at 30 lp/mm and f/5.6 is about 88% MTF IIRC. Any chart that shows an f/5.6 lens performing better than that is bogus.

Re: Things You Didn’t Want to Know About Zoom Lenses

Patrick Chase — Thu, 16 Feb 2017 18:20:34 -0000

I agree that electronics and optics are very different, but you could say the same about just about any pair of product categories or industries you might care to pick. The causes and nature of variations are always unique to some degree, and engineers (including the optical sort) are remarkably good at "drawing distinctions" that make them seem more so.

With that said, the statistics of processes and process control are remarkably universal, and that extends to optical systems. I say this having been intimately involved in all too many production issues where optics were implicated. We missed a LOT of issues by doing sampling inspection, that were fairly easy to find with more modern process control methodologies. A classic example is where you have a batch of ever-so-slight slightly out-of-spec barrels suddenly show up in one bin on the production line.

Re: Things You Didn’t Want to Know About Zoom Lenses

Patrick Chase — Thu, 16 Feb 2017 18:51:47 -0000

Sure, and that's an incredibly universal problem in process control.

Overall system behaviors (what the customer cares about) may be very sensitive to some changes and relatively insensitive to others. You gave an example of the latter in the optical domain, but every engineer who's ever done process control can come up with a litany of similar examples in other domains. Ideally you spec your components in a way that emphasizes parameters that matter, but it isn't always feasible to do so. You often end up with specs that are "too tight" most of the time but have to be that way to prevent disaster.

It's also alarmingly common to have cases where two components are out of spec such that they cancel each other out, and introducing a batch of *good* parts to the line breaks the product. Such is life. That's why companies that manufacture stuff (ideally) employ engineers so sort it all out.

Capability-based controls such as "showmeyourpics" described are useful precisely because they help us detect those cases.

Re: Things You Didn’t Want to Know About Zoom Lenses

Patrick Chase — Thu, 16 Feb 2017 19:07:29 -0000

I can think of contrived/degenerate examples where some of those things would at least partially cancel, but for real lenses I think you're right.

It's important to recognize that modern lenses and cameras are very complex opto-mechanical systems, where customer-perceived quality is driven by more than just the traditional parameters you cite. Throwing something out completely off the cuff: Could you have offsetting errors between the optical power of a stabilizing group and the gain of the piezo actuators that move it? I realize that such an error in the stabilizing group would impact optical performance of the optical formula as a whole, but then you have the question of sensitivity.