[Noam]

Got myself a K-5II and it’s very nice so far. I’m relieved that the AF seems far more accurate and invariant to light-color. Finally!

However, P-TTL shooting with bounced (tilted) flash is severely over-exposing, which is very disappointing, especially since it’s an old bug inherited from previous models. To verify this, I conducted a controlled test in an effort to isolate the problem and dismiss other factors. My flash is the Sigma 610 Super.

The camera was put on a tripod and set to mode X (1/180s). The lens was FA35/2. Unless noted otherwise, all shots were taken with zero compensation on both camera and flash, and automatic flash zoom (50mm as expected). Ambient light was negligibly very low.

I performed multiple sets of shots, each with a different combination of aperture, ISO, and object distance:
1. ISO400 f/8, distance about 3.5m.
2. ISO800 f/4, distance about 3.5m.
3. ISO800 f/4, distance about 2m.
4. ISO80 f/8, distance about 3.5m.
3. ISO400 f/4, distance about 2.5m.
These combinations represent different power requirements from the flash. Some require almost full power while others require minimal power, thus testing the (over-)exposure behavior at various points on the power scale.

In each set I took the following shots:
1. P-TTL with flash pointing forward (distance scale shows on flash LCD)
2. P-TTL with flash pointing forward and flash swivel lock button pressed (distance scale disappears, and tilt icon shows on flash LCD)
3. P-TTL with flash tilted slightly downwards (both distance scale and tilt icon show on flash LCD) - this is the position designed for macro shots.
4. P-TTL with flash tilted very slightly upwards - just enough to make the distance scale disappear and the tilt icon appear.
5. Seven shots at all available manual power levels (1/1 to 1/64) with flash pointing forward.

The observed behavior was basically the same in all sets:
1) P-TTL with flash pointing forward was generally good, with a mid-gray dominant histogram, far away from clipping.
3) Tilting down slightly gave similar (good) exposure, sometimes maybe a little weaker (about 1/3-2/3ev).
2. and 4) Tilting slightly upwards or merely pushing the swivel lock button had a dramatic adverse effect - causing significant overexposure anywhere from 1.33ev to 2.66ev!! Note that the actual flash direction changed very very slightly - only as much needed to make the flash “think” it is tilted.

BTW, I measured the exposure differences by adjusting exposure of the RAW files in LR4.3RC, and equalizing their histograms (and visual previews).

So there you have it. No significant change in flash direction (or any other condition) but strong overexposure when the flash thinks it is tilted. It smells like a big fat silly bug to me.

I also took some shots with “proper” tilting 60 degrees upwards and got similar overexposure behavior (unless the flash was too weak to overexpose, e.g. in the ISO80 f/8 case).

I also tried some tips I gathered from forum posts - enable Highlight Correction, set flash zoom manually to either 24mm or 105mm, and set flash compensation either in camera or in flash, and the behavior was always the same - whenever the flash was in “tilt mode” (distance scale not appearing on LCD) it fired significantly brighter than in “straight mode” (distance scale appearing on LCD), given enough available power.

Finally, I also took some shots in Wireless mode (using the built-in flash set as non-contributing controller) and the Sigma off-shoe. The exposure was good, with no overexposure. On a side note, even though the built-in flash was set as a controller, a significant amount of light still leaked to the image, which is unfortunate.

So, in summary, something goes bad in the flash exposure whenever it is tilted and on-shoe. Shooting straight forward or as a wireless slave, exposure is fine.

Now, here’s an intriguing additional observation that may or may not shed some light on the bug “mechanism”: When I shot the flash with manual power, I observed the proper object distance as calculated and displayed by the flash at the current power setting (ISO and aperture are communicated to the flash). Interestingly, the distance suggestion was wrong, resulting in significant over-exposure if shot from the suggested distance. So maybe this reflects the miscalculation done in P-TTL mode and tilted flash. But then I’d be puzzled howcome exposure is different (and better) when the flash is tilted or off-shoe.

Of course, these problems could be the Sigma’s fault. After all, it’s a single unit of a single model from a third-party manufacturer. Still, I suspect it is actually the camera’s fault, with the symptoms being very similar to those encountered by many people with the classic K-5 (and possibly K-7) and various flash models, including the Pentax 540. It is therefore important that other people who own different flash models, preferably Pentax, perform similar controlled tests and share their results.

It’s getting really late and I’m going to bed, so no photo uploads for now. Sorry. I’ll upload some samples from the test later on.

[Lowell Goudge]

What is the guide number of the flash?

Note that at ISO 400 the AF540FTZ has a guide number in excess of 12 meters at F8, therefore at your ISO you may be simply too close to the subject, even with bounce to control exposure with the camera.

Learn to do manual flash calculations to start your tests. Try again at ISO 100

I have no issues with my AF 540 on the K5. There must be something you are doing wrong, somewhere.

What is the metering, spot or matrix, and what is the subject to background ratio

[LeDave]

Just did a project in my class with this flash. Take Guide number / Distance = F-stop. Not sure how ISO works into play with this case because I used a light meter to fire the manual moded 540 FGZ into the light meter with it's setting at 400 ISO.

[dosdan]

The off-axis test should be repeated by someone with a Pentax flash. Same thing happens with bounce flash with K-5 & Sigma 530 Super (GN 53). I usually end up using the K20D with the Sigma if I want to use flash much.

Dan

---

Last edited by dosdan; 12-11-2012 at 11:50 PM.

[Anvh]

Just did a project in my class with this flash. Take Guide number / Distance = F-stop. Not sure how ISO works into play with this case because I used a light meter to fire the manual moded 540 FGZ into the light meter with it's setting at 400 ISO.

for GN the ISO is often 100, when you go to ISO200 you double the GN.
So with ISO400 you've 4 times the GN.

[kiwi_jono]

for GN the ISO is often 100, when you go to ISO200 you double the GN.
So with ISO400 you've 4 times the GN.

Not quite. There are some good examples in the K-5 (and K-5II) manuals page 185 for the K-5 one. The guide number doubles from ISO 100 to ISO 400 (not 4 x).

Max flash distance = Guide number / Aperture value
Min flash distance = Max flash distance / 5

[LeDave]

ISO 100 to 200 is a 2x difference, so 200 to 400 is a total of 3x difference and not 4x, iso 100 to 800 is a 4x difference.

[dosdan]

for GN the ISO is often 100, when you go to ISO200 you double the GN.
So with ISO400 you've 4 times the GN.

I don't like using the terming "double the GN". At a matching FL, the GN is constant, although if it's assisted by a zooming flash reflector, the GN can increase with FL, but this assistance (tightening the main angle of the flash "cone" as you increase the FL) will cease after a certain FL.

The GN is specified at:

• f/1
• ISO100
• a certain lens FL, e.g. 50mm. For example, with a 50mm lens the Angle-of-View and Angle-of-Illumination of the lens and flash reflector may match, so the max. amount of flash light illuminates the captured scene. But with a 25mm lens, much of the scene will not receive the flash illumination, so the GN will fall.
• 25% subject reflectivity
• It is possible that the locale may affect the GN. I've been advised that the GN rating is determined in a light coloured room. If so, and reflections off the walls are affecting the GN rating, then shooting in a large hall, outdoors or in a dark-walled room may lead to the stated GN not being achieved.

While varying the f-number and ISO will change the effective distance, they aren't really changing the GN rating itself.


See: https://www.pentaxforums.com/forums/accessory-memory-articles/81651-understan...de-number.html


Dan.

[dosdan]

ISO 100 to 200 is a 2x difference, so 200 to 400 is a total of 3x difference and not 4x, iso 100 to 800 is a 4x difference.

No, it's not. ISO100->ISO200 is achieved through 2x pre-ADC gain, as is ISO200->ISO400 and ISO400->ISO800. ISO100->ISO400 is 4x gain.

Distance should follow the inverse square law. So, with a non-zooming flash of fixed intensity, going 10m->14.14m means 1/2 the level of flash illumination and 10m->20m = 1/4 the illumination. But, according to the article in Wikipedia, due to the light beaming effect of the reflector, this may not be completely true.

See: Guide number - Wikipedia, the free encyclopedia

Dan.

[calsan]

[pinholecam]

I do get better results that yours with a Metz48, though I can't say that its always good.

Anyway, its a good evaluation and perhaps lets you know what to avoid with your current setup.

[Noam]

What is the guide number of the flash?

Note that at ISO 400 the AF540FTZ has a guide number in excess of 12 meters at F8, therefore at your ISO you may be simply too close to the subject, even with bounce to control exposure with the camera.

Learn to do manual flash calculations to start your tests. Try again at ISO 100

I have no issues with my AF 540 on the K5. There must be something you are doing wrong, somewhere.

What is the metering, spot or matrix, and what is the subject to background ratio

Lowell, I apologize for writing such a long and information-heavy post (and no images) that certain important details are easily missed on first read. Please allow me to re-explain the core idea of the test methodology:

Using exactly the same conditions (ISO, aperture, distance), I took several shots where the only difference is that the flash is pointed forward in one shot and very slightly tilted in the other shots - so slightly that the flash is practically pointing to the same direction. The purpose of this is to provide an apple-vs-apple comparison with absolutely identical conditions.

Please notice that I did in fact repeat the test using several very different combinations of ISO and aperture, including one set at ISO80.

In all tests, the forward shot was exposed nicely with a hefty highlight-clipping margin. This proves that the flash was indeed capable of exposing correctly in the given conditions. I also took shots at all available manual power levels to assess the power used in the P-TTL shots and to verify that the flash was indeed within working range (BTW, these shots also demonstrated that P-TTL mode is capable of firing at a weaker level than the lowest available manual level of 1/64, something around 1/128 ~ 1/256).

Given these fixed identical conditions, one would expect a similar (good) result in the other (i.e. slightly tilted) shots. However, huge overexposure was observed, which indicates a bug somewhere in the camera/flash logic. More precisely, whenever the flash thinks it is bouncing (the distance scale disappears) it behaves badly. Whenever the distance scale is displayed (forward shot and slight downward “macro-tilt”), the exposure is nice.

Suspecting that the slight upward tilt may be “illegal” in some way, I also did shoot at a normal 60 degrees tilt and still got overexposure. In fact, the tilt and off-ceiling bounce should have made it even harder to over-expose (requiring more power to illuminate the scene), thus further amplifying the indication that there is a bug here. Furthermore, as noted, I also shot wireless with the flash off-shoe held in my left hand, close to the camera and pointing forward, and it exposed nicely.

So, to summarize the essence, we have identical shots in identical conditions - except one is forward and one slightly tilted - that demonstrate a terrible gap in exposure (overexposure in the latter) instead of similar results. And this is repeated in all ISO, aperture and distance combinations tested. I call this a bug.

It would still be interesting to know the flash parameters, and I apologize for not including them in the first place.
According to the flash user manual, the guide numbers of the flash, with flash zoom at 50mm, are (@ISO100):
1/1 46m
1/2 32.5m
1/4 23m
1/8 16.3m
1/16 11.5m
1/32 8.1m
1/64 5.8m
(as expected, each stop gains sqrt(2) more distance)

Here are again the sets I shot, along with the appropriate guide numbers and the corresponding flash power:
1. ISO400 f/8, 3.5m: 3.5 * 8 / sqrt(4) = 15.2 ~1/8
2. ISO800 f/4, 3.5m: 3.5 * 4 / sqrt(8) = 5 ~1/64
3. ISO800 f/4, 2m: 2 * 4 / sqrt(8) = 2.8 ~1/256
4. ISO80 f/8, 3.5m: 3.5 * 8 / sqrt(0.8) = 31.3 ~1/2
3. ISO400 f/4, 2.5m: 2.5 * 4 / sqrt(4) = 5 ~1/64

The conditions for series 3 are possibly a little off, but even here the same (mis-)behavior was observed.

Lowell (and others), if you own a 540 and a K-5, then by all means please repeat my controlled test procedure and contribute your results - Shoot the same stuff on a tripod with fixed identical parameters - forward and tilted - and report your findings.

Metering mode was Matrix (not that it matters in P-TTL, and anyway I shot on manual (X) and the exposure was pitch black without flash).

I shot a green-painted wall with a canvas print hung on it and a matte-finished brown upright piano along the wall, so there is no subject-background ratio - It’s all a pretty flat subject. No shiny parts either.

Sorry again for not posting images yet. I’ll do it back home.

[Lowell Goudge]

just a note here but metering does matter.

P-TTL flash uses the camera;s metering to measure the combination of flash and ambient light. it uses the same sensors as the camera therefore if you for example meter in spot mode it measures the flash intensity in spot mode also.

[Noam]

just a note here but metering does matter.

P-TTL flash uses the camera;s metering to measure the combination of flash and ambient light. it uses the same sensors as the camera therefore if you for example meter in spot mode it measures the flash intensity in spot mode also.

AFAIK, P-TTL always meters the entire frame in a Matrix-like fashion, regardless of camera metering mode, but I could be wrong. this is an interesting question but please let's leave it to another thread, so that we don't lose focus here...

thx!

[Anvh]

I don't like using the terming "double the GN". At a matching FL, the GN is constant, although if it's assisted by a zooming flash reflector, the GN can increase with FL, but this assistance (tightening the main angle of the flash "cone" as you increase the FL) will cease after a certain FL.

Wasnt talking about the focal length or anything else, just the ISO, with ISO200 you double the GN, how would you say it then?