[indytax]

This "open letter" on Luminous Landscape is fascinating. Using data complied by DxO Labs, the post asserts that DSLR manufacturers program cameras to covertly raise the ISO from the selected setting in order to compensate for light loss at the sensor. Consequently, the author suggests:

Due to the complexity of design and manufacture (let alone the high cost and weight) of large aperture lenses, one may actually end up with better results at virtually the same ISO and depth of field using lenses with more modest maximum apertures.

An Open Letter To The Major Camera Manufacturers

[TaoMaas]

I don't know that you can say we're better off with slower lenses because that really depends upon how you're plan to use them. If you're shooting sports...no, you're not better off with slower lenses. One extra stop can mean the difference between stopping the action or getting motion blur. Or it can mean that you shoot one click lower on your ASA and thereby have less noise. But if you're shooting landscapes and are planning on using the lens at f8 to f22...yes, you're probably better off with slower lenses if for no other reason than they're much cheaper and you could probably afford to buy a more complete set of primes.

[RioRico]

Conventional wisdom: Use the fastest possible lens because 1) wide-open lets you shoot faster, 2) you have more control over DOF, and 3) you hit the "sweet spot" (optimal FL) at a wider aperture.

Counter-argument: Use slower lenses because 1) they're cheaper, 2) they're smaller and lighter, 3) they're sharper wide-open, and 4) they're cheaper.

I try to exploit both sides of the equation. Carry the big Sears-Tomioka 55/1.4 (310g) but carry the tiny Industar-50/3.5 (60g) too. It gets heavier if not so extreme at longer FL's: my Lentar-Tokina 200/3.5 (690g) vs my Meyer Telemegor 180/5.5 (250g). Optimizing for slower glass can result in a MUCH lighter kit. But that faster, heaver glass sure is necessary at times.

[elho_cid]

sweet spot is a myth. there are not so many challenges in designing a slower lens and a slower lens can perform better at all aperture settings that is shares.
But DOF control is strong argument.

[indytax]

While I understand the advantage of more control over DOF the point of the article is that to obtain the other advantages, DSLRs are programmed to increase ISO beyond that set by the user in order to compensate for the greater light loss over film. Some cameras raise ISO levels nearly 1 stop to compensate. That increase in ISO negates whatever advantage in image quality you'd see by a wider-aperture sweet spot, and minimizes whatever advantage you got by faster shutter speed at the "chosen" ISO.

[Ron Kruger]

Faster lenses simply give you more latitude with light and more creativity for DOF. From my experience, the only situations were slower is better is for landscapes, where all my shooting is between f-8 and f-11 anyway.
The only other situation where slower might be better is for studio photography, where light intensity and angle is controlled, because there is some truth to Elho Cid's statement about slower lenses preforming better at all apertures, but this really depends upon the quality of the lens.
So I don't agree about the "cheaper" arguement. Even with slower lenses, you get what you pay for.

[reeftool]

A slightly slower lens is cheaper to manufacture and will probably sell more. This debate is raging on in various threads about the new DA35/2.4. Why not f/2 or faster? How many people would buy one at 3 times the price? The camera manufacturers are running a business. They have to make a profit and balance the fine line of price vs. quality. If a lens ends up being too expensive, very few will buy it.

[opiet70]

Not to be a naysayer or anything, but is 2.0 vs. 2.4 (if the glass is equivalent) a big deal when the K-5 can go up to ungodly ISOs? The K5's high ISOs are better than my K10D at 800... When you can gain 3-5 stops from the camera's ISO ability, the only real thing fast glass gives you is better DOF control.

This statement excludes certain lenses that are simply phenomenal...the bokeh masterpieces, the contrast masterpieces, etc. No camera can make the glass better. But, when everything's equal on the glass portion, a 1/2 stop makes little difference for the majority of people if they have the ISO range on their cameras.

Just my 2c...

[brofkand]

sweet spot is a myth. there are not so many challenges in designing a slower lens and a slower lens can perform better at all aperture settings that is shares.
But DOF control is strong argument.

How many lenses can you point out that are just as sharp wide open as they are at all other apertures?

I don't know about you, but virtually every lens I've seen is softer / less contrast / etc wide open. Whether or not it's a huge difference is one thing, but it is there. It's measurable, like a cup of sugar.

So "sweet spot is a myth" is in and of itself a myth.

I would rather shoot at ISO 100 at f/1.4 than I would at ISO 800 at f/5.6

There may not be much noise at ISO 800, but that noise went somewhere. It wasn't never there; the camera removed it. And it's very likely that in the process it removed some detail that I would have had, had I shot at ISO 100.

High ISO is such a load of baloney. I can shoot at ISO one million, but if there is no detail there what's it mean? I see photos shot at ISO 6400 all the time, people say they're gorgeous. I say yeah, but why can't I count the number of eyelashes the subject has?

I would get more resolution out of instamatic film!

It amazes me the "trends" in photography....Pictorialism, HDR, toy cameras, high ISO, etc etc.

[bpjod]

Ummm, only one person commenting here (before me) so far seems to have actually read the referenced article before posting. The article has nothing to do with the traditional pros/cons of fast vs. slow lenses.

He argues that a lot of light transmission is lost by complex lens designs. For example a complex f/1.4 lens compared to a simple f/3.5 lens may lose nearly one stop of light transmission due to the extra elements and lens interfaces. So if both lenses are set to f/8, the simple (and slower) lens actually transmits almost twice as much light (one-stop) to the sensor as the more complex (and faster) lens. To compensate, the camera bumps the ISO of the sensor to reach an equivalent exposure, but this happens without the photographer's knowledge. For example the camera is set to ISO 100, f/8, 1/125s, but with the more complex lens the camera bumps the sensor to nearly ISO 200 to compensate for the light loss of the more complex lens without informing the photographer-the EXIF still reports ISO 100. This results in decreased image quality because of the ISO bump.

My first thought was that this bodes well for Pentax's DA Ltd. lenses as they are fairly simple designs and aren't terribly fast. My second thought was that while I can understand how the camera manufacturers might be duping us by bumping the ISO of the sensor without our knowledge, how on earth did Fuji and Kodak bump the ISO of my slide film without my knowledge?

An interesting article, and while I don't doubt that some light loss likely occurs in a complex lens design, I suspect that it isn't as big an issue as the author seems to suggest; otherwise it would have been a well-known issue in the days of shooting slide film, much like reciprocity failure.

[Ash]

Cameras bumping ISO up secretly? I'm not so sure. My Tamron 70-200 is optically quite a complex lens. I shoot with it at f/2.8 and 4 often at ISO 1600 on my K20D. Results from these shots show no sign of them being any noisier than any other ISO 1600 shot I've gotten from a simpler FA 50 or 77.

Not a believer.

[DAZ]

There was a time when a magazine did a revue of a lens not only was the T stop difference part of the revue but also the variation in T per aperture stop. The aperture is a mechanical devise so there can be some variations. The lens construction is what makes the overall T stop difference.

With TTL metering and film with more latitude this seems to have fallen out of vogue. Then again with the rise of the supper zooms maybe the manufactures just discouraged the magazine revues from discussing a less then luminous point about their product. Maybe it is time it should come back.

As others have pointed out if the camera is only using new lenses that report the information to the camera this is an easy way to get consistent looking photos. With a camera that uses older lenses that do not report this info it is probably best the camera stay as close to the correct ISO as possible. Then at least the photographer can correct on a lens by lens bases.

When ISOs were all low this was probably not much of an issue but when you start looking at high ISO numbers the under reporting will make the camera staying closer to the correct ISO look much worse then it really is.

The issues of Pentax inconsistent and under metering being do to T stop differences and variation in T per aperture stops for various lenses has been brought up by others but so far has fallen on mostly deaf ears.

DAZ

[Lowell Goudge]

I find this argument hard to believe

I have however seen a 1 stop difference between two lenses at the same aperture. The problem is they are not new complicated designs but old and equivalent designs

I tested an XR Rikenon 135/2.8 side by side against an SMC Yakumar 135/3.5

FromF5.6 and smaller the rikenon always shot at half the shutter speed to the tak on the same subject with the same light

My point is

are lenses different? Yes

But I don't believe the Manufacturers play any games with the iso

[JonPB]

I'm curious as to whether people are (a) not reading or (b) not understanding the article.

Look at the third graph on that page. It shows that by f/2.8, the secret ISO increase is less than about 0.03 EVs--effectively negligible. But at f/1.4, the difference is about 0.3 to 0.5 EVs--which may matter, but we're talking about the difference between ISO 100 and 140 here. Personally, I'm not going to sweat that difference, but I'd like to know it is there if my camera is jumping up to it, especially if I'm choosing to shoot at 1250 to avoid the noise my camera makes at 1600; I'd rather take the time to shoot at ISO 1000 (really 1250) in that case.

This is related to T-stops, but it isn't about how complex lenses affect light transmission regardless of f-stop. And by "slow" lenses, we're talking about f/1.4 compared to f/2.8, rather than f/2.8 to f/4.

Fast lenses (e.g., f/1.4) cause light to hit the sensor from the side. If light hits a CMOS sensor at too large an angle--too much from the side--then the light will hit the well wall rather than the sensor and it will not be recorded. This, in a way, reduces the T-stop of the camera system: the sensor isn't recording light transmitted by the lens. And so the camera boosts the ISO to compensate for the fact that it knows it isn't recording as much light as it being transmitted.

Note that this is not a problem (so far as I know; I'm deviating from the article here) of f/2 or faster lenses. Rather, this is a problem of f/2 or faster lenses being too close to the sensor. To have such a wide aperture too close to the sensor requires some of the light to hit the sensor at a large angle, even too large of an angle to be recorded. The further away the lens is from the sensor, the less this will be a problem. I'd expect the M9, if it had a conventional sensor, to suffer from 1 stop (more or less) greater difficulty with this problem than an SLR camera.

Furthermore, the smaller depth of field caused by large aperture lenses is largely caused by high-angle light; the wells on a CMOS sensor effectively increase DOF at large apertures.

Which raises two questions in my mind: is this why Pentax is leaning towards f/2.4+ lenses? Perhaps. If so, then is the 55/1.4 designed with this in mind?

But, no, we're not better of with slower lenses. We are, however, better off knowing what our cameras are doing. I didn't buy a DSLR to be kept in the dark about how my exposure is being created.

[Lowell Goudge]

in readig the article one could draw another conclusion also, specifically that the canon F1.2 lens and the nikon F1.4 lens are not really F1.2 and F1.4

While there has been discussion in the past about old technology lenses and having light hit the sensor at oblique angles, this was most at issue at the corners and caused vignetting.

Also, the discussion here is about the problem of CMOS sensors and the deep well design associated with CMOS, and not CCD sensors. Although I have not checked every model of nikon camera in his chart, I am quite certain most of the cameras are CCD and not CMOS therefore the argument may not hold true.

A long time ago there was discussion about camera makers cheating on ISO settings, and not having the ISO settings set as indicated, in order to give brighter and brighter images. I believe canon was at the root of this also, in terms of being the worst offender.

It would be interesting here to see the performance of the systems discussed AT ALL APERTUTES, what is presented is a single datapoint, that only really shows IMO exposure error of lenses wide open, and this is NOT the best judge of the lenses or the sensors/systems.

With respect to the movie industry and T-Stops, the reason they moved to T-stops was because the medium had such a small exposure lattitude that they had no other option if the exposure was to be spot on. It is not because all lenses are so far off the mark, it is because when making a movie on film, with minimal lattitude in the first place, the 1/3 stop made a difference, an big difference.

As an aside, while I am skeptical about camera makers fooling substantially with the ISO ratings, I have no reservations AT ALL about camera makers fooling with metering and the issues associated perhaps with bright focusing screens and fast lenses. Just look at how the K10D meters with fast manual lenses, or lenses taken out of the A mode, yet they meter perfectly with the same lens in the A position. Clearly the camera has the metering inaccuracy "calibrated out" for known maximum apertures.