Completely Dejected (auto focus K5)

I what Algernon says is true then it is impossible is it not?

If I shoot a portrait from a metre or so away and want the iris sharp I have a margin of error of 2cms. I am saying that without luck you're not going to nail it consistently without a focusing aid.

I would live to be wrong! It would solve a lot of problems.

If you get a rail similar to the one in the link I posted previously http://www.slrgear.com/articles/focus/focus.htm allowing back and forth movement of about 300mm each way and take several shots say 20mm apart.... one of them should be in focus.... if you set the focus at mid-rail first.... Ha! Ha!

Note: slrgear also have perfect lighting with about four 400w daylight units costing hundreds of pounds similar to this:
http://www.imaging-resource.com/ARTS/TESTS/HMI.HTM

Also note they have a small focusing rail on the rig for left/right movement. Unfortunately the focus errors are normally so large that something bigger than a small focusing rail was needed by them in the front/back direction.

I've tried with a Novoflex focusing rail which allows +/- 30mm movement and no matter what method of focusing I used. It still indicated in focus for the length of the rail. I wanted to get the two points where it goes in and out of focus so that I could take a point halfway between the two. I might try again with one focusing rail on top of another.

Even if you get it in focus, you still have curvature of field throwing it out at the edges and focus shift changing the point of focus as the lens is stopped down.

If you focus just by looking at the screen (and I've always done that with 6x6 cameras) you can only say it's acceptably in focus within the DOF because that's the definition of the DOF theory.

The best way is to practice focusing on small boxes until you can read the lettering clearly.

I reckon about 1 shot in 4 will be almost exact. I'm not sure if I was just lucky with this test MF with a 50mm f/1.4 exact to the mm, notice the change in colour around the 0 point




Falk Lumo has done extensive tests about the low light focus problem and printed a graph which shows points in daylight and bright indoor lighting as well, he's not explained what all the colours mean, but very few points fall on top of others link

My solution with a 24-35mm lens is use f/8 and infinity.

I can just see that at the next Asian wedding he tries to shoot. (Maybe he can commandeer a desert trolley?)

What is the point of that question?

Skill is skill and has to be recognised as such. Tools that are used to help make that task easier are available.

Best regards

The one thing that hasn't been mentioned is practice. Manually focusing well needs lots of practice to gain the skill to judge the correct point of focus. It will be considerably slower than using AF and the amount of error can be surprisingly large.

I have a magazine article on this somewhere in an olf Photo Technique showing the amount of error with a 50mm lens when we focus several times. AF is incredibly accurate by comparison.

Manual focusing is very useful in macro work though.

I've just done some more tests, all shot at 1.4, tripod with remote. The lens was the Pentax FA 50mm

This one was EV 0.3 so pretty dim, not expecting much here, Auto Focus (single point on the large letter T of the word "The" on the left of the page.)




This one was the same as above but Live View




I added some more light and brought it up to EV3.6, this shot was AF single point.




This was Liveview.




For the sake of the test I just left the camera on Av and let it figure out its own exposures.

I then tried manually focusing and I reckon I have sharp eyesight, my efforts were nowhere near as close as any of the above. In fact bizarrely when I focused I could clearly read the word I was focusing on (in this case the word 'village' just underneath the large T) but when I looked at the pic it is barely readable and the part that IS in focus was totally unreadable when looking through the view finder.

I adjusted the diopter from fully to the right to almost all the way to the left and this effect still happened i.e. text I could read through the lens was unreadable in the result.

This is the manual focus pic




Clearly something is amiss. Are there any forum users in the West Yorks area that could take a quick look at my cam and verify my results...?

That last test was rough-and-ready but more meaningful. For more usual apertures the DOF would mean focus was close enough, but for precision focusing you'll need more light by the look of things.

Your manual focusing was terrible, so adjust the dioptre properly as detailed previously. If you can't tell the difference from one extreme to the other then you're probably trying to set it whilst looking at an approximately in focus image. Defocus completely and concentrate on getting the lines etched in the viewfinder sharp.

Please could you explain how you calculated the EV as 0.3?

Your EXIF says 1/80 at F1.4, ISO 400.

I may be misreading the table here, but if it's right (and based on ISO 100), your exposure comes out between EV 5-6.

(1/80 at ISO 400 is equivalent to 1/20 at ISO 100, then reading from the F1.4 column in the table).

Not sure I understand this EV thing yet

An EV table, taken from Carl Shipman's Pentax SLR Cameras book. These tables are based on ISO100 and an f1.4 lens.
[IMG]http://i240.photobucket.com/albums/ff250/johnriley1uk/001EVTable.jpg[/IMG]
Any combination of aperture and shutter speed that is equivalent to the ones given will give the same EV number.

Thanks John.

So if I've understood this right, F1.4 and 1 second would be EV -1 at ISO 400?

So EV 5 would be 1/64 at F1.4 and ISO 400?

That corresponds with the wiki table - so I'm still unclear where the EV 0.3 comes from.

Of course, EV 5 isn't particularly low light at all. If that level of front focus at EV 5 is typical, it will be a real problem for low light portraits under tungsten, where a wide aperture is desired to focus very specifically on an eye, for instance.

Not a trivial or rarely-encountered problem with the camera at all.

Why can't they put the EV value on the LCD panel?

A Pentax Spotmeter gives it directly as it's reading to 0.3 EV

I'm not sure if it's displayed on the camera if you set it to spot?

Edit: My Gossen Sixtomat Digital gives them as well
currently about EV 5 in the living room... not very bright.

A typical tungsten exposure might be more like 1/60sec at f5.6 using moderately powerful photoflood lighting, but I suspect that we're thinking here more about wanting to use ordinary room lighting.

To be honest, that's dreadful lighting for a portrait, if indeed that's what we are talking about. Images that appear to be natural room lighting are probably extensively augmented by additional hidden lights, thus giving reasonable exposures.

Seriously mate did you read my post, I said I adjusted the diopter from one extreme to the other and the the error was still consistent.

A portrait was just an example. So what if the picture doesn't meet some arbitrary aesthetic standard of lighting and composition!!

Seriously, you're not really suggesting that it's unreasonable to take a picture at 1/80 at ISO 400, at quite short range, and expect that the camera focuses where it's supposed to (and claims to), rather than an inch or more in front?

I'm sure he did read your post. Adjusting it from one extreme to the other is not the same as adjusting it correctly for your eye.