Sigma 18-35mm infinity?

[Abitconfused]

When calibrating the Sigma 18-35mm zoom lens using Sigma's Lens Dock, values for infinity at four focal lengths are required. However, it is impractical to build a target the size of Manhattan to use for this setting. As a practical matter, what distance can be used for the infinity input given that I can build a suitable target up to .6 X .6 meters?

[NorthernFocus]

Can't you find a brick wall somewhere to use as a target? Then use the calculated hyperfocal distance plus a skoshe.

[Abitconfused]

The target (or a brick wall) is of modest value absent an attached scale. The scale extends behind and before the target to capture the location of front or back focusing. Additionally, lines on the scale also betray chromatic aberration. A line with minimal fringing is often a good prospect for point of focus adjustment. So the target must be large enough for easy focusing using one point when the lens is set to maximum aperture and the scale must also be large enough to be resolved by the lens, in this case at a mimimal infinity distance whatever that may be.

[NorthernFocus]

Sounds like you've got a challenge. Not sure which Sigma lens you're working on but hyperfocal distance for 35mm at f2.8 is 47ft and at f1.8 it's 75ft. Then again I have no idea what Sigma means when they say "infinity". Might be worth clarifying.

I've not used their lens dock. Do you enter data indicating at what range focus is sharp or is it determined with software?

Glad I only calibrate my long lenses and never bother with the short stuff

[Abitconfused]

I am calibrating the Sigma 18-35mm. Data is entered through visual inspection of the scale image captured at the required focal length and distance. I use Photoshop Curves white eyedropper to aid contrast. As mentioned above, green fringing and magenta fringing on target scale lines serve as a guide because they fall beyond and before the focal plane (as I understand it). My calibration is 3/4 done as only the 4 infinity inputs remain. I have observed a significant improvement in image quality thus far. I am tickled spitless. This is a truly sharp lens. I also enjoy the process. It's like a puzzle.

[xpatUSA]

The target (or a brick wall) is of modest value absent an attached scale. The scale extends behind and before the target to capture the location of front or back focusing. Additionally, lines on the scale also betray chromatic aberration. A line with minimal fringing is often a good prospect for point of focus adjustment. So the target must be large enough for easy focusing using one point when the lens is set to maximum aperture and the scale must also be large enough to be resolved by the lens, in this case at a mimimal infinity distance whatever that may be.

As Dan said, focusing on the hyperfocal distance plus a tad should do it.

At f/1.8 and a CoC of 0.0188mm, where CoC = sensor diagonal / 1500, I calculate the said distances for four focal lengths as:

18mm: 9.6m
24mm: 17.1m
28mm: 23.2m
35mm: 36.4m

Good luck ...

[Abitconfused]

I am beginning to think that adjusting at infinity may not be needed. If the adjustments are additive from closest to furthest then the first three may be sufficient. An illustration of the Datacolor Lenscal target showing color fringing on scale lines...

[george013]

I don't know what Sigma is doing with that calibration. But if I understand it good, there're now 2 captains on 1 ship: the camera AF and correction and the Sigma corrections.
Further I don't understand how distance calibrating can influence CA. I must be mising something.

George

[Abitconfused]

The lens has firmware that interfaces somehow with the camera’s autofocus. Firmware is rewritten as you update calibration data using the Lens Dock. I don’t know that focus can actually minimize CA but it seems to. I adjust AF to a point where CA is minimally observed in a photo of lines from the target scale.



After calibration sharpness improved sufficiently to assuage a sharpness neurotic such as myself.

[Stagecoach]

Here's an article showing 'Longitudinal Chromatic Aberration' giving similar as shown in post 7.

https://photographylife.com/what-is-...tic-aberration

[Abitconfused]

Yes, and at some point, I cannot tell if any change is needed. There is a word for this, done.

[Abitconfused]

A lower ISO reveals even greater detail this lens can deliver when calibrated.



I added a warming adjustment layer with a clipped brightness adjustment to this image,

[Abitconfused]

View of an image of the Lenscal target. The lens is backfocusing.

[Abitconfused]

I achieve very good results adjusting the distance of the camera to the target using autofocus while being certain the distance scale on the lens matches the image provided by the Lens Dock (red bars). I don't use rulers at all.



For the Sigma 135mm, I didn't need to make any calibration changes at all.

[Abitconfused]

If you focus a lens at a distant wire, you may see evidence of chromatic aberration, even in lenses with specialized lens elements designed to reduce CA. I think no lens can completely eliminate CA. Or could the lens be capturing the color of light diffusing around the small point?



Is this similar to the line in the Lenscal target at infinity? And does the spacing of the fringing suggest good lens calibration? I think so...

[Stagecoach]

And does the spacing of the fringing suggest good lens calibration? I think so...

By 'spacing' I assume you are referring to the equal amount of different colour fringing each side of centre?

If so,

All you need to do is make a target using something like a similar piece of wire with something 6'' in front and something 6" behind.

Now focus at each depth (near middle behind) and see if the fringing spacing remains equal with the centre target when you know it is certainly not accurately focused.

That will confirm if your assumption is right or wrong Ed

[Abitconfused]

That's a great idea!

[Stagecoach]

Here's the answer..........................

Three wire nails spaced 20mm apart pushed through a bit of cardboard. Nails were placed roughly at 45 deg to the sensor plane giving a rough distance of 10mm difference between each.

Camera on tripod with remote but unfortunately pointed up slightly to get the cloudy threatening sky as background. I could not re-shoot due to heavy rain but it clearly demonstrates things.

Focused on each nail in turn at approx 3/4 up its height using AF with a Nikon 50mm at f/1.4

No 1 - Focus on front nail


No 2 - Focus on centre nail


No 3 - Focus on rear nail


I undertook the test first off with a Tamron 28-75 2.8 but CA is almost non existent.

So yes, if you have a lens that exhibits lots of CA it appears you could use the 'colours' to assist in determining focus plane.

Edit : What I also noted whilst doing this and using Live View enlarged fully whilst playing was that the CA colours could very easily be seen.

[Abitconfused]

Thank you for this worthy experiment. When we calibrate a lens we are really calibrating lans & Camera and the most significant variable must be sensor placement. A tiny bit back or forth, I presume, is a significant factor or is it variables within the focusing sensor? Inquiring minds want to know...

[Stagecoach]

Thank you for this worthy experiment. When we calibrate a lens we are really calibrating lans & Camera and the most significant variable must be sensor placement. A tiny bit back or forth, I presume, is a significant factor

Yes, that's why we should calibrate a lens to each camera we use it on. I have not come across a lens of mine yet that has given me cause to calibrate it other than when using certain ones with a 1.4TX.

or is it variables within the focusing sensor? Inquiring minds want to know...

I suspect the greatest error with absolute 'focus accuracy' is repeatability due to dead band.