G2 autofocus distance on LCD vs reality

[stepien]

Hello everyone,

I'm having troubles with AF in my G2 in a way that distance on the top LCD is not as in reality. For instance, for 2.0 m LCD says 2.1 m, for 3.0 m LCD says 3.2 m. Error is progressive so for ~6 m I'm getting ~8 m. Could someone please verify what distances on LCD you get when the camera is placed in:

- 2.0 m
- 3.0 m
- 4. 0 m

?

I'm wondering if it's typical for G2 that there are some distance discrepancies.

Some more examples of my G2 (real distance / LCD distance):
1.0 m / 1.01 m
1.30 m / 1.32 m
1.50 m / 1.54 m
1.60 m / 1.65 m
1.80 m / 1.87 m
1.90 m / 1.97 m

 

[TimVermont]

Bardzo mi miło. Which lens are you using? Are you focusing with your thumb on the rear button? (see the thread below & other G2 threads) What is your focusing target?

 

[stepien]

Hi Tim, For the test above I used 28 mm lens and shutter release button for focusing. The target was a typical chessboard with black and white fields. With regards to the lens - the G2 can also operate w/o to determine distance.

PS do you know Polish?

 

[TimVermont]

My G2 shows little variance (1-2cm) between display and measured distances in landscape orientation; but slightly more variance (3.06-3.08m at 3m) in portrait orientation. The camera was in an L-Bracket and on a tripod, so the change from vertical to horizontal did not change the physical distance. However, when I lock the shutter open on "B" and check focus with a piece of ground glass it is on-target at f2 (45mm/2 lens) regardless of the displayed distance. So the displayed distance has a certain tolerance to it.

You are dealing with a much larger variance. I suggest setting up a ruler at a 45 degree angle to the edge of your target (Google "LensAlign") and use either ground glass or film to see where it is really focusing.

I grew up in a neighborhood of refugees and immigrants and had the good fortune to be exposed to many languages.

 

[stepien]

Thank you very much Tim, your post proves that my variances are not acceptable. I've already done a test where I shot a target with 45 deg aligned ruler in 2 configurations AF and MF (I measured distance with tape measure and it was input to G2). The AF ones were out of focus, whereas MF ones were OK. So apparently once the body has a proper distance on the LCD it can shoot sharp photos. I would like to use ground glass for further tests but don't have any at the moment. Was it difficult to see if the image was sharp on the ground glass?

 

[TimVermont]

I used a very low-tech approach and taped a small piece of ground glass (a piece of Gepe 35mm slide cover glass ground on a fine sharpening stone) to an 8x loupe and used a hat as my shade. My target was an X-Rite color checker, and if I could read the type I considered it sharp.

 

[Leigh B]

You're failing to take into account manufacturing tolerances.

The displayed distance is based on how far the lens mount has moved from its infinity position.

But that __assumes__ the lens' focal length equals its nominal value.
That's seldom the case.
Actual focus is determined by the true focal length.

There are also tolerances in many other dimensions in the camera.
What you're viewing on the LCD is the __ideal__ distance to the subject.
That can only be achieved if all of the various errors are zero, which is impossible.

- Leigh

 

[stepien]

I used a very low-tech approach and taped a small piece of ground glass (a piece of Gepe 35mm slide cover glass ground on a fine sharpening stone) to an 8x loupe and used a hat as my shade. My target was an X-Rite color checker, and if I could read the type I considered it sharp.

Thanks, I wanted to get a loupe anyway.

You're failing to take into account manufacturing tolerances.

The displayed distance is based on how far the lens mount has moved from its infinity position.

But that __assumes__ the lens' focal length equals its nominal value.
That's seldom the case.
Actual focus is determined by the true focal length.

There are also tolerances in many other dimensions in the camera.
What you're viewing on the LCD is the __ideal__ distance to the subject.
That can only be achieved if all of the various errors are zero, which is impossible.

- Leigh

Thanks Leigh. What do you mean by 'how far the lens mount has moved from its infinity position' ? I've heard the G2 lenses might need calibration...Also I noticed that in my G2 the actual AF patch is moved to the left w.r.t. the AF bracket in the center of the viewfinder. In other words, approx. half of the central AF bracket is inactive, and instead focus can be tracked outside it. I'm wondering if this can lead to the distance errors I'm getting.

 

[Leigh B]

Hi,

My comments just reflect the fact that tolerances can cause correlation errors.

This is true of any manufactured product... not unique to photography.

Any algorithm (like for calculating distance) defaults to using nominal dimensions.
The idea of "calibrating" a lens is to use the real dimensions rather than nominal.

- Leigh