Possible to recompose without messing up focus plane?

[bhop]

Here's a little diagram I just made.. red line is the focus, yellow is the lens field of view. I think it's pretty self explanatory other than that. Note the red dot at the end of the focus line, how it moves away from the focal plane after recomposing causing the subject to be slightly oof but usually this isn't that noticeable unless you're shooting wide open.

 

[Josh66]

Nice diagram. You should make the green line (plane of focus) move with (and stay perpendicular to) the red line in the second frame though, just for clarity.

edit
Or maybe do that, but add another line (different color) where the green line used to be. Green line = actual plane of focus, other line = intended plane of focus.
Or keep the green line where it is and make the 'actual' plane a red line. The colors don't matter - you know what I mean. (I hope.)

 

[dumeril7]

Nice diagram. You should make the green line (plane of focus) move with (and stay perpendicular to) the red line in the second frame though, just for clarity.

That seems like the critical part (at least relative to the way I was misunderstanding plane of focus). If I understand this right, the re-composing would put the plane behind the original subject.

d7

 

[bhop]

Ah.. I just put the green line there to make it easier to see how the focus spot moves away from the subject, didn't even think about that being the plane too.. heh, heh.. but i guess it'd be something like this. I could make a cleaner version, but I closed that illustrator file already and didn't save it.. gotta get back to work anyway.

 

[Josh66]

That's better. :thumbup:

 

[usayit]

This is probably my most posted link. Hopefully it will enlighten a few people:

Why Focus-Recompose Sucks

Its exactly the same notion as I calculated in the thread I linked earlier.... if you actually run through the calculations, the person who wrote that is making a bif deal out of nothing.

 

[Josh66]

This is probably my most posted link. Hopefully it will enlighten a few people:

Why Focus-Recompose Sucks

Its exactly the same notion as I calculated in the thread I linked earlier.... if you actually run through the calculations, the person who wrote that is making a bif deal out of nothing.

It all depends on the specific shooting circumstances... Landscapes - yeah, you can pretty much ignore that.

Wide open, up close with a 50mm 1.4 - well, we get threads weekly from people wondering why their shots aren't in focus...


The link doesn't apply to every situation, but it does apply to some. And I think it's a good idea that people are aware of it, for those times when they are in one of those situations where it actually matters.

 

[usayit]

since no one seems to read links:

Here's what I posted several years ago:

Here's the calculations from the same thread:

=======

I'm no math wiz.. but.. something to consider.

Assumptions:
* 35mm frame
* recompose to 1/2 the Field of view at a specified focal length
* 10 meter subject distance
* Focal length of 24mm
* Aperture of f/1.4

Fov = 2*arctan(d/2*f)

Fov = Field of view
d = diagonal dimension (35mm for full frame)
f = focal length.

Let's just start at a 24mm focal length

Fov = 2 arctan(35/2*24)

Fov = 72.2 degrees @ 24mm. As per assumption, Frecompose = 72.2/2 = 36.1 degrees

Next... Focus plane error when camera is rotated during recomposition (Frecompose).

E = a - a * COS (t)

E = focus error (behind subject)
a = distance to subject
t = degrees rotation.

Let's just take a subject distance of 10 meters with a rotation of 36.1 degrees

E = 10 - 10 * COS(36.1)
E = 1.9 meters


Now let's lookup the DOF for the same parameters (I use dofmaster.com online calculator)

DOF in front of subject = 4.2 meters
DOF behind subject = 27.7 meters

If my calculations are correct, 1.9 meter focus error should still be within DOF for a 24mm focal length at f/1.4 at a subject distance of 10meters.


I ran through the calculations for 50mm at f/1.4 as well (same 10meter subject distance)
E = 0.15 meters
DOF in front of subject = 1.44 meters

yup still ok...

How about a super fast telephoto 200 f/1.4
E = 0.038 meters
DOF in front of subject = 0.1meters

yup still ok...


Ok ... so 10 meters might be too far.. Lets try a subject distance of 5 meters

24mm Error = 0.95 meters
DOF in front of subject = 1.34 meters.

50mm Error = 0.28 meters
DOF in front of subject = 0.39 meters

200mm Error = 0.019 meters
DOF in front of subject = 0.03 meters.


The way I see it, your are fine until you enter in to the subject distance of 3meters or less. Of course doing this on paper is different than real world use since there are so many other factors involved (CoC for different formats for example).

These are just numbers... whether or not it matters enough to worry about is a personal decision. (I'll keep mine to myself)


Then again.. I suck at math.. so if I made a mistake please point it out.

=========


IMO, making a big deal out of something pretty small..... don't trust everything you read online... think things through.

 

[usayit]

Wide open, up close with a 50mm 1.4 - well, we get threads weekly from people wondering why their shots aren't in focus...

As I said, I shoot with a 50mm f/1... I do see the shift but its so small I can easily correct it by leaning back. The way I see it, the link takes a corner case and generalizes it without actually making any real experimentations or calculations to prove or disprove. 90% of the time, mis-focus is the person not the focus shift due to recomposition. This is especially true with novice shooters with multi-area AF... the camera latched onto something and lighted the big green dot without the person realizing the "something" is not intended.

 

[Josh66]

10 meters is pretty far from "up close".

 

[usayit]

10 meters is pretty far from "up close".

fine.. how about 5 meters:

=====

Ok ... so 10 meters might be too far.. Lets try a subject distance of 5 meters

24mm Error = 0.95 meters
DOF in front of subject = 1.34 meters.

50mm Error = 0.28 meters
DOF in front of subject = 0.39 meters

200mm Error = 0.019 meters
DOF in front of subject = 0.03 meters.

The way I see it, your are fine until you enter in to the subject distance of 3meters or less. Of course doing this on paper is different than real world use since there are so many other factors involved (CoC for different formats for example).

===



Or did you not read?????



Do the calculations.. determine yourself.. stop being argumentative for absolutely no reason... The math is right there. I know at some point the DOF shrinks past shift error but at that point, its a corner case and not as big of deal as the link makes it out.

 

[Josh66]

The way I see it, your are fine until you enter in to the subject distance of 3meters or less.

I didn't see this when I made my last post...

See, I'm not talking about 5 or 10 meters... I'm talking about "up close" - at or near the minimum focusing distance. 3 meters or less. Like I said - it depends on the circumstances. All I'm saying is that you need to know when focus-recompose will cause problems, which (like I've said a few times already in this thread) is rare.

 

[AperturePriority]

Ah.. I just put the green line there to make it easier to see how the focus spot moves away from the subject, didn't even think about that being the plane too.. heh, heh.. but i guess it'd be something like this. I could make a cleaner version, but I closed that illustrator file already and didn't save it.. gotta get back to work anyway.

View attachment 7112

I think what you bhop miss here is the the plane of focus is considered as flat but itn't. It's a sphere equidistant from the sensor. You can only consider it is a flat plane in cases of the long tele focal lenses which have narrow FoV and closing the apertures down.

For the landscape photos, mostly people use normal or wide-angle lens which have wide FoV. In such a case, the plane of focus is a REAL SPHERE EQUIDISTANT FROM THE SENSOR, it isn't a flat plane anymore. Specially, when one wide-opens a normal or a wide-angle lens, her/his photos only get sharp at the center but not at the edges nor the corners.

You may have a plane of focus is close to flat at wide-open aperture on some special lenses like 58mm 1.2 Noct or the long focal lenses, but not for the other normal or wide-angle lenses.

 

[Josh66]

Ah.. I just put the green line there to make it easier to see how the focus spot moves away from the subject, didn't even think about that being the plane too.. heh, heh.. but i guess it'd be something like this. I could make a cleaner version, but I closed that illustrator file already and didn't save it.. gotta get back to work anyway.

View attachment 7112

I think what you bhop miss here is the the plane of focus is considered as flat but itn't. It's a sphere equidistant from the sensor. You can only consider it is a flat plane in cases of the long tele focal lenses which have narrow FoV and closing the apertures down.

For the landscape photos, mostly people use normal or wide-angle lens which have wide FoV. In such a case, the plane of focus is a REAL SPHERE EQUIDISTANT FROM THE SENSOR, it isn't a flat plane anymore. Specially, when one wide-opens a normal or a wide-angle lens, her/his photos only get sharp at the center but not at the edges nor the corners.

You may have a plane of focus is close to flat at wide-open aperture on some special lenses like 58mm 1.2 Noct or the long focal lenses, but not for the other normal or wide-angle lenses.

Sorry, but you are wrong. Especially the part you put in all caps.

 

[usayit]

like I've said a few times already in this thread) is rare.

Do the calculations at 3 meters or less on a 50mm f/1.4 wide open. You've got to be even closer than you think. The key here is "rare" as you said.. something that most online articles about the topic fail to make obvious. For most reasonable situations.... its not something to worry about too much. In my case, I was wondering because I was shooting 50mm at f/1 at minimum distance the lens is capable of... 1m. I would consider this a "corner case". Here's the test photo I took along with the thread to show the razor thin DOF i was dealing with:

In my case, a simple lean back of the head and shoulders was enough to correct... something that I practiced since then. btw.. those toy cars are about 1 inch in length each.