Is PTTL worthless?

I do know that reflective strips on kids coats are a real, 100% certified nightmare for P-TTL and TTL flash
But normally when you have reflective surfaces you can see that something has reflected the flash and skewed the reading. I can't see any great evidence from the supplied images.
What about other subjects that are definitely not reflective?

(Having seen how my AF360 and AF540 respond on both the *ist-D and K10D I do wonder whether you've got some real problem though).
Matt

Ah thank you Matt, you've just explained why some shots I took a while back were seriously and mysteriously underexposed.

And wouldn't you know it, as I try to reproduce the problem using matte surfaces everything is consistent. A little darker than I would like but I just cannot reproduce the wildly varying results seen earlier.

I've had similar problems with people... I wonder now if even the tiniest spec of highlight such as from eyes, teeth or jewelry can throw off the metering that much. I would think not, but it would explain the problems I've seen.

Honestly, I don't buy that myself, so I'll have to play a bit more and see what I can do. At least it is a starting point.

The other thing that's always bothered me a little...

One thing that's often forgotten is the wavelength of the light. I wonder if the sensors used for the metering (esp. the flash metering) are particularly sensitive to one wavelength - more so than the CCD?

(AFAICT, even the RGB histogram is affected by the colour-space, so you can't take it as a concrete guide )

Just a thought for the melting pot...
Matt

I think that what people overlook is that the light from a flashgun is extremely intense. You wouldn't dream of trying to take a picture with the flashgun pointing at the lens, but that is effectively what happens when you have a good reflective surface.

In the picture, light from these reflections appears as a blown-out highlight. Try stopping down the lens until the highlights are not blown out. You'll probably find you are on f256 at 1/20,000 sec shutter speed.

(That's assuming you have a K1D, with one of the new landscape lenses.)

G

I think the specular highlight has a lot to do with the issue. Although I'm not exactly sure how. From another very quick and non-scientific test it seems that the camera handles spot metering just fine... UNLESS a specular highlight falls within the metered area.
Even with a HUGE flash spot nearby the metering is still acceptable, but with the tiniest little spot in the center of the frame seems to spoil the results.

The following images are of a semi glossy photo (again of my dog) with some shiny objects nearby.

One good thing is that no matter how I tried to angle the flash or cover the sensors, the exposure seems to be similar.

A little depressing, but I think I know what types of things to look out for. It is just very hard to see a microdot of reflection off something like the ring on the dog's collar or in the darker case below, probably the glint on the edge of one of the trophies.


http://moringello-east.no-ip.com:8080/test/sample_02.jpg

Edit: I took out the inline image, as it ruined the formatting of the entire page. Ugh...

Are you using spot-metering all the time, or just in this instance? If so, try the 'matrix' metering (or whatever it is on a Pentax ). It may then give less weighting to the highlight than would happen if it happens to be in the spot area.

Dan

I seem to get the same results on spot, center and matrix. Yet for photos without these extreme highlights, the metering modes do seem to make a difference. So the system is working, it just isn't working anywhere near what I would expect.

But it is in fact working properly. The meter can do do more than measure the light coming through the lens.

Diffusing the flash would probably help.

G

Do you get better results if the flash is used off the camera? In that case reflections might not be bounced directly back to the camera? Might help in the reflective jacket situation.

Yes, it seems to be working "properly" in most cases. I am just surprised by the results. If there is a huge glaring bright hole in the image, it is metered just fine. Yet if there is just a tiny spec of reflection, the shot is blown to hell.
I'm not sure this is truly expected behavior. If it is not just my system and it is "expected", I'll learn to live with it and I'll try to work around it.

Oddly, I have a few pictures with reflective strips on clothing in them and they come out just fine. Results similar to the example above where a large portions of the image is overly bright, but the majority of the image is properly exposed.

Anyway, one such workaround might be to use the flash off-camera. I may give that a try... although holding the flash and trying to aim it correctly doesn't always work well. Guess I'll need to buy a bracket. The wifey will be so pleased that I want to buy another accessory.
I'm just glad at the moment that I have a little time to think up some reason for "our need" to afford the new D* lenses.

For use off-camera, I just place the flash on the back of a chair or a table, or anywhere. Holding it doesn't usually get it far enough away from the camera.

G

When I have a studio type environment I put them on stands. But when on the go and walking around, I usually just attach the flash to the camera.

That being the case, I rarely have issues with the flash off to the side when diffused through brolleys.

I guess the alternative to a bracket is to make that hat thingymajig... there was a link on ePhotozince a while back where a guy attached a small reflective umbrella and the flash to a hat. Apparently it gave pretty decent results, if looking a bit bizarre at the same time.