[BigMackCam]

Steve, this is an excellent and very useful post. Many thanks for taking the time to write and share it.

As suggested, I've made the post "sticky"

[marcusBMG]

Well done Steve.
I too have gone over to using "M" mode a lot more with my MF lenses to avoid exposure inconsistencies.

Might be worth mentioning in your piece that mount conductivity/non-conductivity can also factor into wrong exposure results, as for example I wrote about re my K-r here.
(just noticed I need to re-post the pics).
Lowell Goudge has in the past posted some graphical results for exposure with several pentax dslr bodies..

[stevebrot]

I can't even remotely see how the presented deviations of 0-1 EV are of any significance to anyone outside theory.

Good point! The posted graphs were used as illustration because they are the ones I have for "A" lenses where "expected" was the actual metered value for that lens. They were part of a larger test comparing the stock screen with the Katz Eye product. I did not want to get bogged down with too much detail in the post, but with real world subjects the situation with the K10D was much more alarming with some lenses. The worst example might be my LZOS MC Jupiter-9 85/2. Wide open with a real world subject, the stop-down reading would yield 2.5 - 3.0 stop underexposure with readings at f/5.6 being much closer to expected. As noted in another comment, a quick test yesterday with an XR Rikenon 50/2 was grossly overexposed at f/5.6 for a green-button reading of my well-lit breakfast nook (a common test subject).*

Unfortunately, most reports are unavoidably anecdotal and in the real world, many lenses manage just fine. Here are a few general notes:

• Current generation bodies are much better than those from a decade ago, at least for M-mode metering
• Complaints/questions about stop-down metering continue even with today's cameras. The general form is, "My Sony meters my Taks just fine, but metering on my K-S2 is all over the place".
• Faster lenses tend to be more problematic
• Moderate maximum aperture lenses tend toward more predictable metering. I don't even chimp with my Tamron 70-150/3.5 on the K-3
• Most lenses (the above Rikenon excepted) are fairly predictable at moderate stop-down apertures
• It is difficult to separate the mechanical accuracy/precision of the aperture mechanism from meter problems
• Stop-down metering at narrower apertures is complicated by lack of linearity at the limits of meter sensitivity

It is that last point that I forgot to include in the original post. The manuals for stop-down cameras such as the Spotmatic typically include a table indicating a so-called meter coupling range, with combinations of ASA(ISO)/shutter settings outside the approved values on the table being outside the range of meter sensitivity and/or linearity. Moving forward to the present, the RGB meter in my K-3 may be sensitive to -3 EV (100) for stop-down readings with a f/1.4 lens wide open, but not linear in the same light with the same lens stopped down to f/2. Under those conditions, the camera will set too high a shutter speed and underexpose.

It is a complex subject.


Steve

* Yes, the diaphragm is good on the Rikenon. Aperture performance (based on histogram centering) is linear. Why stop-down metering is flaky is beyond me.

[stevebrot]

Somewhere inside the camera's software is a numerical conversion factor that relates the light flux measured by the TTL sensor to the estimated light flux of the scene as modulated by the effects of the half-silvered reflex mirror and the efficiency of the focusing screen to capture the light from the lens.

Yep...thanks for the concise description

How that works and the failover strategy for non-A lenses is a matter of speculation, unfortunately.


Steve

[stevebrot]

An easier fix would be to stick a little paper label inside the lens cap with the corrections to use for that lens written on it.

I considered doing something like that. For my Jupiter-9 it would read something like:

f/2, add 2.5 stops EC
f/4, add 1 stop EC
f/5.6 and narrower, no correction


Steve

[stevebrot]

On the K-5 with my "S-type" screen I get fairly consistent results (which I got mainly for the ease of manual focusing fast manual lenses, but improved metering was a welcome bonus)

The S-type screen is a special case worth noting. It is optimized for enhanced focus sensitivity with fast lenses and brightness is fairly linear by stopped-down aperture down to a "cliff" at about f/4. I did not test mine extensively when I had it (long story), but wondered if it might work better with stop-down metering than other screen options.


Steve

[stevebrot]

Might be worth mentioning in your piece that mount conductivity/non-conductivity can also factor into wrong exposure results

Yes, things get complex with some bodies when coupled with the non-conductive bases on most M42 lenses. Thanks for the reminder.


Steve

[UncleVanya]

Thank you for this. Very helpful.

[vonBaloney]

The S-type screen is a special case worth noting. It is optimized for enhanced focus sensitivity with fast lenses and brightness is fairly linear by stopped-down aperture down to a "cliff" at about f/4. I did not test mine extensively when I had it (long story), but wondered if it might work better with stop-down metering than other screen options.

I haven't done any rigorous testing, but when these subjects come up I usually grab a lens or two and do some comparisons, which I did just before posting last, but f/2 or f/8 or even f/22, difference is usually pretty slight (CDAF vs PDAF). Sometimes it is up, sometimes it is down, sometimes it is nothing -- total range probably a half-stop, but let's call it 2/3rds. And again, I do have to "center" some lenses differently overall anywhere from -0.7 to +0.7. But no big 2-3 stop aberrations from one aperture to another that I can think of. (I do think I had slightly more trouble with the framed screen, so I will stick with the plain one.) And the S-screen is a dream to focus with on fast lenses -- the stock screen just couldn't do it at all faster than f/2. I can get pinpoint focus on this screen with my f/1.2. Love it. I see focusingscreen.com is still making these for all models. I'll put another $100 in my KP budget I think.

[Ontarian50]

I always figured that despite claims stop down metering could be "more accurate" because you were reading the actual brightness of the lens, I considered that reasoning flawed - and more just marketing hype.

I think a big issue with reading image brightness off the focusing screen is that there's always a tiny amount of light that creeps past your eye through the eyepiece, and adds itself to the reading. It's probably miniscule - let's say it's only 5% of what the meter sees, and it can be compensated for in the meter calibration. If you're outdoors in bright sunlight, more light gets into the eyepiece - but the image brightness from the lens is correspondingly much brighter too - so it all works out.

But, if you stop that lens down to f16 for extra depth of field, your focusing screen image is now 0.78% as bright as it was when the lens was wide open at f1.4. And if you're still leaking that same amount of light past your eye into the finder, it's now more than six times brighter than your focusing screen image. That can't be helpful.

I always figured that was a good argument for using a tight fitting rubber eyecup when using stop down metering.

In practice, it's likely eyepiece leakage wasn't as much as 5%, but even if one percent, it would explain underexposure by a stop or so when using smaller apertures.

So, add that to the inconvenience of dim viewing, it's not hard to see why full aperture metering took over, especially when autoexposure became a must-have.

Of course, with no eyepiece to leak light, mirrorless cameras are immune to this issue, I would think.

[stevebrot]

I think a big issue with reading image brightness off the focusing screen is that there's always a tiny amount of light that creeps past your eye through the eyepiece, and adds itself to the reading.

Good point and especially pertinent for stop-down metering. I was doing some tripod work away from the eyepiece and wondered why the stop-down exposures were off way more than usual...duh!

So, add that to the inconvenience of dim viewing, it's not hard to see why full aperture metering took over, especially when autoexposure became a must-have.

Eyepiece leakage is a big concern for AE as well. A blind was provided on my Ricoh XR-2s and several other models of the time. My XR7 and Super Program were intended for use with an included cover. The same is true for modern Pentax dSLRs. I use my thumb.

As for dim viewing...I learned with a stop-down meter body. The trick is to meter before shooting. The automatic aperture actuation allows focus and framing full open without fiddling in the dark. Meter once and shoot freely until either subject or light changes. Unfortunately, that auto feature is lost with adapted M42 lenses


Steve

[stevebrot]

I don't even chimp with my Tamron 70-150/3.5 on the K-3

the RGB meter in my K-3 may be sensitive to -3 EV (100) for stop-down readings with a f/1.4 lens wide open, but not linear in the same light with the same lens stopped down to f/2. Under those conditions, the camera will set too high a shutter speed and underexpose.

Below is the graph for the above-mentioned Tamron on the K10D. Stop-down performance on the K-3 is similarly good . The point at f/22 shows where the light value was below the meter's lower level of linearity. The camera indicated 1/3s at both f/16 and f/22.

Steve

[Eyewanders]

The S-type screen is a special case worth noting. It is optimized for enhanced focus sensitivity with fast lenses and brightness is fairly linear by stopped-down aperture down to a "cliff" at about f/4. I did not test mine extensively when I had it (long story), but wondered if it might work better with stop-down metering than other screen options.


Steve

This is interesting to me (I'm late and I just noticed this thread - fantastic opener, Steve) and quite honestly didn't realize folks were still struggling to an extent with stop-down metering. I remember well (I'd all but forgotten), now that you've posted this, the troubles with the K100D Super and K10D I used back then, but that was my first foray into dSLR and using legacy glass for the first time on digital I chimped and chimped and dealt with it. I was still figuring it all out.

I moved to the K-7 immediately on realease and nearly as quickly went with a Katzeye screen which improved things drastically... to the point I didn't think on it anymore. With the K-5 and K-3 I used S-type - even better. I all but forgot about discrepancies of past and I still don't notice any significant issues. I use M with Green button almost exclusively on k-mount M lenses, and I use Av with Takumars about 75% of the time, and M+Green the other 25%. Haven't even thought about it in all this time if I'm honest, and perplexed/interested that others are struggling occasionally. I've not had a stock screen in a DSLR in a decade however...

[rogerstg]

An easier fix would be to stick a little paper label inside the lens cap with the corrections to use for that lens written on it.

I love this idea for brief lens notes, such as best aperture, etc. It would not have helped with my K10D/manual lens metering though because I needed a table for each lens and aperture setting.
Still, a simple idea worth noting. FWIW, Post-it correction tape works well for making small labels in tight spots.

[bhbrake]

Thanks for taking the time to post this as I was not aware of the history and many of the challenges. Even with modern A lenses, The metering is noticeably off quite often when I stop down on the K3. The K50 metering goes completely bonkers when stopped down, but that was not always the case and is defect I have made due with. Typically I shoot with an EV adjustment of .3 under exposed as I feel many DLSRs are more prone to overexpose- my old Nikon D40 and Canon Rebel xti were the same way. Very easy to lose highlights when stopping down.