[jcsnyc]

So I have embarked on the journey of adjusting all my lens (15+) across two cameras (K3 II and K1). I have referenced a great post by Class A and other references to GhettoCAL (which I built today.)

I would like some opinions on the following:

On GhettCAL and other more costly solutions, the target to focus is about 3.5" to the left of the ruler with numbers to help with calibration. While this distance sounds small, according to Pythagoras, there is an impact.

By my calcs, if the target is 60" (5 feet) from my lens AND the number on the ruler are 3.5" to the right of the target, the distance to the number 0 on the ruler is actually 1/10" of an inch further.

Sqrt(60^2 + 3.5^2) = 60.101997 - so .101997 further.

Since we are trying to get a really accurate focus, does the 1/10th of an inch matter?

There is a chance I am being a BIT TOO precise, but this is stumping me a bit.

---------- Post added 11-29-18 at 03:22 PM ----------

NOTE: I could just make a mark 1/10" close on the ruler to compensate. Does that make sense?

Attached Images

 

[SSGGeezer]

You are letting your nerd loose! At anything other than f/0.95 I don't suspect you are going to see any effect in the real world. DOF and all of that jazz!
Showed your Rokinon 24 shot to my doc today after he commented on my Pentax forums shirt, he loved it!

[Docrwm]

The answer is - it depends on the f stop used and the focal length of the lens. My focal point when I do those things is ALWAYS the"0" on the right. Works well for me.

[BarryE]

There is a chance I am being a BIT TOO precise ...

I'm inclined to think you might be - in the real world

Many moons ago I worked through my lenses; set them; used them ... and then out of curiosity, a little while later, sampled a few micro adjustments. Some where the same. Some were +/- 2 different. The camera test settings were the same, but deliberately the test environment (light, distance, test chart etc) were different. I concluded micro-adjustment had got me close to what was required, but that there were too many real world variables to bother fiddling any more. Now if it's critical I bracket the focus or manual focus or use LV or use a loupe etc. AF is only ever a close approximation and some lenses' AF vary from shot to shot by a surprising amount.

[derekkite]

You are focusing on a flat plane, the focus target parallel to the body sensor. If the focus point in the body is out of adjustment and focusing on point an inch back of the target, the sharp section of the ruler image will be offset an inch back from zero.

I don't see how the calculation is relevant.

[jcsnyc]

You are letting your nerd loose!

Yes - I think my nerd is in overdrive. Good advice from all - close is close enough!

(Thanks, SSGGeezer on the photo shout out...with your doctor! I was reading a Porsche magazine while visiting my urologist - and ended up accompanying him to pick up his new Boxster a few weeks later....)

[swanlefitte]

The focal plane should be pretty flat. If the focal plane was severely curved it would matter. Think how hard photographing a building would be without focus stacking. The distance to the sides of the building can be meters farther away.

[MarkJerling]

So I have embarked on the journey of adjusting all my lens (15+) across two cameras (K3 II and K1). I have referenced a great post by Class A and other references to GhettoCAL (which I built today.)

I would like some opinions on the following:

On GhettCAL and other more costly solutions, the target to focus is about 3.5" to the left of the ruler with numbers to help with calibration. While this distance sounds small, according to Pythagoras, there is an impact.

By my calcs, if the target is 60" (5 feet) from my lens AND the number on the ruler are 3.5" to the right of the target, the distance to the number 0 on the ruler is actually 1/10" of an inch further.

Sqrt(60^2 + 3.5^2) = 60.101997 - so .101997 further.

Since we are trying to get a really accurate focus, does the 1/10th of an inch matter?

There is a chance I am being a BIT TOO precise, but this is stumping me a bit.

---------- Post added 11-29-18 at 03:22 PM ----------

NOTE: I could just make a mark 1/10" close on the ruler to compensate. Does that make sense?

Interesting. I can't do math in inches and feet so, I checked the math in metric!
You are correct, the difference in distance is 2.591mm which is, indeed, very close to 1/10". Theoretically, therefore, we could improve the target by moving the vertical target some 2.5mm back in relation to the sloping ruler.

[SSGGeezer]

Interesting. I can't do math in inches and feet so, I checked the math in metric!
You are correct, the difference in distance is 2.591mm which is, indeed, very close to 1/10". Theoretically, therefore, we could improve the target by moving the vertical target some 2.5mm back in relation to the sloping ruler.

Are you trying to out-nerd a guy who is a big nerd mark?

[Docrwm]

Interesting. I can't do math in inches and feet so, I checked the math in metric!
You are correct, the difference in distance is 2.591mm which is, indeed, very close to 1/10". Theoretically, therefore, we could improve the target by moving the vertical target some 2.5mm back in relation to the sloping ruler.

It comes back to my original response - it depends on the f-stop and focal length......but, yes it could possibly make a difference.

[MarkJerling]

Are you trying to out-nerd a guy who is a big nerd mark?

I can do nerd without trying!

[Snafu711]

As swanlefitte said: it‘s about the focal plane. If you took a picture of a person in front of you (pointing the camera horizontally) and focused on the person‘s eye, you would expect to have the person’s feet in focus, too. Although trigonometry will tell you that the are significantly further away.

Same with the target.

[Class A]

Theoretically, therefore, we could improve the target by moving the vertical target some 2.5mm back in relation to the sloping ruler.

This would only apply if the lens' field curvature weren't flat.
The field curvature refers to the surface in the object space which is rendered in focus in the image space.

Any decent lens, however, is designed to minimise field curvature. In other words, in order for objects to be rendered in focus, they must all be in a flat plane in the object space that is parallel to the sensor.

Some lenses deliberately allow some minor field curvature in order to improve other lens properties, but with the distances occurring in a micro adjustment focus setting, they all should be fine.

[MarkJerling]

This would only apply if the lens' field curvature weren't flat.
The field curvature refers to the surface in the object space which is rendered in focus in the image space.

Any decent lens, however, is designed to minimise field curvature. In other words, in order for objects to be rendered in focus, they must all be in a flat plane in the object space that is parallel to the sensor.

Some lenses deliberately allow some minor field curvature in order to improve other lens properties, but with the distances occurring in a micro adjustment focus setting, they all should be fine.

Aha! Thanks for explaining that.

[swanlefitte]

This is a post from Pentaxuser
johnriley
Link Posted 08/01/2018 - 10:16
Lenses project a curved image onto a flat surface, and good ones do it very well, managing to keep centre and edges in sharp focus at the same time. This is especially true of macro lenses, which are often used to make flat copies.

A cheap lens will show field curvature and at wider apertures the edges may not be in focus when the centre is. This is easy enough to discover using a newspaper as a target. One solution is to curve the film/sensor so that everything is in focus.

The other thing is deliberate field curvature and Minolta made a VC lens where the amount of curvature could be dialled in. This made semi-circular groups of people easy to photograph without distortion and whilst keeping them all sharp. It's not been a mainstream idea that's caught on in any great way, except probably within the secrets inside inexpensive compact cameras.
Best regards, John