In his excellent post on comparing focus stacking with a rail or by turning the focusing ring, Rik notes that at medium magnification (say a flower), it is better to use the ring. Using the rail will result in changes in perspective that the stacking program may have trouble dealing with.
It would seem to me that using a longer focal length lens would mitigate changes in perspective, since the traverse of the focus stack would be proportionally less as compared to the object distance. Would this be meaningfully significant?
Bill
Focus Stacking with Rail: focal length and perspective
Moderators: rjlittlefield, ChrisR, Chris S., Pau
-
- Posts: 91
- Joined: Fri Dec 30, 2016 1:59 pm
- Location: Lake Forest, IL, USA
- rjlittlefield
- Site Admin
- Posts: 23608
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: Focus Stacking with Rail: focal length and perspective
Yes. A good model is that what matters is the ratio between depth of stack and distance to entrance pupil. Sometimes there are significant differences even between two lenses of the same focal length, when one lens has its entrance pupil farther back than the other. There can also be significant differences between lenses in how far the entrance pupil moves as the focus ring is turned. I don't know any way to predict these things from published lens specs, but they can be easily seen by just looking into the front of the lens as it is stopped down and focused.billjanes1 wrote:It would seem to me that using a longer focal length lens would mitigate changes in perspective, since the traverse of the focus stack would be proportionally less as compared to the object distance. Would this be meaningfully significant?
--Rik
-
- Posts: 91
- Joined: Fri Dec 30, 2016 1:59 pm
- Location: Lake Forest, IL, USA
Re: Focus Stacking with Rail: focal length and perspective
Rik,rjlittlefield wrote:Yes. A good model is that what matters is the ratio between depth of stack and distance to entrance pupil. Sometimes there are significant differences even between two lenses of the same focal length, when one lens has its entrance pupil farther back than the other. There can also be significant differences between lenses in how far the entrance pupil moves as the focus ring is turned. I don't know any way to predict these things from published lens specs, but they can be easily seen by just looking into the front of the lens as it is stopped down and focused.billjanes1 wrote:It would seem to me that using a longer focal length lens would mitigate changes in perspective, since the traverse of the focus stack would be proportionally less as compared to the object distance. Would this be meaningfully significant?
--Rik
Thank you for your thoughtful reply. My long macro lens is the Sigma 180 mm f/2.8, which is asymmetrical and internal focusing. Looking into the front of the lens with its focus set to a magnification of 1:4 reveals that the entrance pupil is about 4 inches to the rear of the mount of the lens.
To get a more accurate measurement, I made a simple test bed as shown below. A focusing target was fixed to the mount of the macro lens and secured in place by wedging it in place with an extension tube.
I moved the focusing camera forward and took a picture of the aperture diaphragm. Then without changing the focus of the lens, I moved the camera backwards until the focusing target at the rear of the lens came into focus and measured the distance of the adjustment, which represented the distance of the aperture from the mount of the Sigma lens.
These are the results:
The entrance pupil moves backwards rather markedly with increasing magnification. At a magnification of 1:10 (corresponding roughly to a bouquet of flowers) the distance is 8.4 cm, and at a magnification of 1:4 (0.25x, corresponding approximately to a single rose) the distance is 15 cm as read from the graph. In your table comparing focusing by ring vs rail, you state that the rail is awful for a bouquet of flowers and mediocre for a single rose. Would this change with the Sigma macro?
Note: Edited 16 Jan 2017 @ 20:20 GMT to replace the table and add a graph.
Thanks,
Bill
Last edited by billjanes1 on Mon Jan 16, 2017 1:22 pm, edited 3 times in total.
- rjlittlefield
- Site Admin
- Posts: 23608
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Bill, thanks for the excellent test.
I don't understand exactly how the last 4 columns differ from each other, but the values are all close enough that I'll take them as equivalent.
I suspect a data entry error someplace in the first column, because although magnification increases monotonically, the Distance numbers first decrease and then oscillate: 0.70, 0.55, 0.60, 0.55.
In any case, let me "think out loud" about the cases that you've asked about.
Case 1, the bouquet. Let's suppose for a moment that the bouquet of flowers has its closest point at 1.10 m, mag 0.10, so about 23 cm wide on an APS-C sensor. Supposing also that the bouquet is round, so 23 cm front-to-back, then stacking by rail would require 23 cm of camera movement, versus what I get as something well under 1 cm of entrance pupil movement if you focus by ring. The ratio between 23 cm and 1 cm indicates that focus by ring will give a lot less change in perspective, and the ratio between 23 cm and 110 cm (=1.1m) indicates that focus by rail is liable to give problems. The situation by rail is not nearly as bad as if you were trying to shoot that bouquet with a 50 mm lens, but I still wouldn't do it.
Case 2, the single rose. I don't see 1:4 (mag=0.25) in your table, so let me use instead the whole range from 0.20 to 0.33. You're showing a difference in focus distances of 0.08 m (=0.78-0.70) = 8 cm, which would be the distance that a rail would have to move. In comparison, focus by ring would move the entrance pupil by about 8.8 cm (=22.3-13.5). The difference between 8 cm and 8.8 cm is too close to call, and both of them are around 10% of the distance to subject. Both of these strike me as a little troubling ("mediocre"), so I would test to be sure. If your numbers are correct, I might well end up deciding to use the rail because that looks to give a lot less variation in magnification at this focus distance.
At higher magnifications, it looks like with this lens the argument is strong for using a rail instead of the focus ring. As I'm reading your numbers, at mag=0.56, the entrance pupil is more than 1 meter away from the subject, which would imply hardly any change in perspective with movement, while the change in magnification versus focus is significant.
Does this make sense, or have I misinterpreted something?
--Rik
I don't understand exactly how the last 4 columns differ from each other, but the values are all close enough that I'll take them as equivalent.
I suspect a data entry error someplace in the first column, because although magnification increases monotonically, the Distance numbers first decrease and then oscillate: 0.70, 0.55, 0.60, 0.55.
In any case, let me "think out loud" about the cases that you've asked about.
Case 1, the bouquet. Let's suppose for a moment that the bouquet of flowers has its closest point at 1.10 m, mag 0.10, so about 23 cm wide on an APS-C sensor. Supposing also that the bouquet is round, so 23 cm front-to-back, then stacking by rail would require 23 cm of camera movement, versus what I get as something well under 1 cm of entrance pupil movement if you focus by ring. The ratio between 23 cm and 1 cm indicates that focus by ring will give a lot less change in perspective, and the ratio between 23 cm and 110 cm (=1.1m) indicates that focus by rail is liable to give problems. The situation by rail is not nearly as bad as if you were trying to shoot that bouquet with a 50 mm lens, but I still wouldn't do it.
Case 2, the single rose. I don't see 1:4 (mag=0.25) in your table, so let me use instead the whole range from 0.20 to 0.33. You're showing a difference in focus distances of 0.08 m (=0.78-0.70) = 8 cm, which would be the distance that a rail would have to move. In comparison, focus by ring would move the entrance pupil by about 8.8 cm (=22.3-13.5). The difference between 8 cm and 8.8 cm is too close to call, and both of them are around 10% of the distance to subject. Both of these strike me as a little troubling ("mediocre"), so I would test to be sure. If your numbers are correct, I might well end up deciding to use the rail because that looks to give a lot less variation in magnification at this focus distance.
At higher magnifications, it looks like with this lens the argument is strong for using a rail instead of the focus ring. As I'm reading your numbers, at mag=0.56, the entrance pupil is more than 1 meter away from the subject, which would imply hardly any change in perspective with movement, while the change in magnification versus focus is significant.
Does this make sense, or have I misinterpreted something?
--Rik
-
- Posts: 91
- Joined: Fri Dec 30, 2016 1:59 pm
- Location: Lake Forest, IL, USA
Rik,rjlittlefield wrote:Bill, thanks for the excellent test.
I don't understand exactly how the last 4 columns differ from each other, but the values are all close enough that I'll take them as equivalent.
I suspect a data entry error someplace in the first column, because although magnification increases monotonically, the Distance numbers first decrease and then oscillate: 0.70, 0.55, 0.60, 0.55.
Does this make sense, or have I misinterpreted something?
--Rik
Thanks again. Your analysis makes sense and has increased my understanding of the issues.
There was indeed a data entry error. I redid the experiment and posted the corrected results in my original post. The three columns are data in triplicate, which were averaged to give the final result.
Regards,
Bill