rdeloe
Well-known member
I often see posts where people are wondering how much they can shift using a particular lens and sensor.
There are some nice tools out there already for well-known combinations of lenses and sensors, e.g., the DT image circle visualization tool. However, I've found it's easier to use a spreadsheet approach. I use all kinds of lenses that aren't included in DT's tool, and sometimes I'm interested in whether I can do combinations of shift in X and Y. I'm also interested in knowing things like aspect ratio and shifted resolution when flat stitching.
If you want to see how this works, I've posted a stripped down version of the spreadsheet I use. If you find it useful, enjoy.
My camera is a Fuji GFX 100S, so I'll explain how it works using that system. In a nutshell, you just plug in the amount of shift you are using in X (landscape) and/or Y (portrait). I use the yellow cells for my 100S.
In this example I want to know the image circle I need to shift 15mm in X (landscape). On my GFX 100S, I need 80.8mm.
If I'm not flat stitching, then I can ignore everything to the right of the "Shifted IC mm" column because I'm using a single frame. However, if I'm actually shifting left and right 15mm to flat stitch a panorama, then I can read off the new aspect ratio (2.24 to 1 ), effective sensor size (73.8 mm x 32.9 mm) and resolution (19,626 x 8,736 pixels).
You can use the spreadsheet to calculate the image circle for shift in two directions. If I want to shift 20mm in landscape and rise 15mm, I need 104.8 mm. Note this is effectively a diagonal shift. I think in terms of view camera shifts (separate rise/fall and left/right shift). Therefore, you'll have to customize the spreadsheet if you're using a tilt-shift lens that shifts diagonally.
The information to the right of "Shifted IC mm" will be useful if you're building a multi-row panorama with these movements. For example, in this case if I wanted a huge 4:3 aspect ratio image, I'd need to make a multi-row panorama based on flat stitching a series of images that include 20mm of shift left and right, and 15mm of rise and fall.
The "Film" section at the bottom is there because I'm using film lenses as shift lenses. If you're using film lenses that have image circles designed to allow shift (e.g., lenses for 4x5 film, or shift lenses for medium format cameras), you will have to add a row for them with their actual image circles if you want to keep that information handy.
To illustrate, I've been using a SMC Pentax-A 645 35mm f/3.5 lens for my primary "wide" lens. The circle of good definition of that lens is 70.8mm in my spreadsheet. As film photographers will know, film formats do not use the full area in the name of the format. So a 645 lens does not actually create an image on film that is 60mm x 45mm. It varies a bit by system. In my spreadsheet I've simply taken 3mm off each dimension for each format.
Assuming a 70.8mm image circle, I should only be able to shift 9.5mm in X (landscape) on my GFX 100S sensor with my SMC Pentax-A 645 35mm f/3.5 lens (shifted IC is 70.9mm and nominal IC for 645 is 70.8mm). This is conservative. In practice, I can easily shift 12mm in landscape. I routinely shift 12.5mm in Y (portrait) to create an effective field of view of a 26mm lens, and image quality at the edges is good at f/8. If there's nothing going on in the far-shifted zone (e.g., sky above a building), I can rise (shift) 15mm in landscape with my SMC Pentax-A 645 35mm f/3.5 lens before I reach the edge of the circle of illumination (defined by the point at which hard mechanical vignetting appears).
The take-away is that the image circle sizes in my spreadsheet for the film formats is conservative. This may be true of published image circles for technical camera lenses too (witness the many debates about how big the image circle of a lens actually is).
There are a lot of variables in play around image circles, including these (some of which will apply to all lenses, whether for film or digital):
There are some nice tools out there already for well-known combinations of lenses and sensors, e.g., the DT image circle visualization tool. However, I've found it's easier to use a spreadsheet approach. I use all kinds of lenses that aren't included in DT's tool, and sometimes I'm interested in whether I can do combinations of shift in X and Y. I'm also interested in knowing things like aspect ratio and shifted resolution when flat stitching.
If you want to see how this works, I've posted a stripped down version of the spreadsheet I use. If you find it useful, enjoy.
My camera is a Fuji GFX 100S, so I'll explain how it works using that system. In a nutshell, you just plug in the amount of shift you are using in X (landscape) and/or Y (portrait). I use the yellow cells for my 100S.
In this example I want to know the image circle I need to shift 15mm in X (landscape). On my GFX 100S, I need 80.8mm.
If I'm not flat stitching, then I can ignore everything to the right of the "Shifted IC mm" column because I'm using a single frame. However, if I'm actually shifting left and right 15mm to flat stitch a panorama, then I can read off the new aspect ratio (2.24 to 1 ), effective sensor size (73.8 mm x 32.9 mm) and resolution (19,626 x 8,736 pixels).
You can use the spreadsheet to calculate the image circle for shift in two directions. If I want to shift 20mm in landscape and rise 15mm, I need 104.8 mm. Note this is effectively a diagonal shift. I think in terms of view camera shifts (separate rise/fall and left/right shift). Therefore, you'll have to customize the spreadsheet if you're using a tilt-shift lens that shifts diagonally.
The information to the right of "Shifted IC mm" will be useful if you're building a multi-row panorama with these movements. For example, in this case if I wanted a huge 4:3 aspect ratio image, I'd need to make a multi-row panorama based on flat stitching a series of images that include 20mm of shift left and right, and 15mm of rise and fall.
The "Film" section at the bottom is there because I'm using film lenses as shift lenses. If you're using film lenses that have image circles designed to allow shift (e.g., lenses for 4x5 film, or shift lenses for medium format cameras), you will have to add a row for them with their actual image circles if you want to keep that information handy.
To illustrate, I've been using a SMC Pentax-A 645 35mm f/3.5 lens for my primary "wide" lens. The circle of good definition of that lens is 70.8mm in my spreadsheet. As film photographers will know, film formats do not use the full area in the name of the format. So a 645 lens does not actually create an image on film that is 60mm x 45mm. It varies a bit by system. In my spreadsheet I've simply taken 3mm off each dimension for each format.
Assuming a 70.8mm image circle, I should only be able to shift 9.5mm in X (landscape) on my GFX 100S sensor with my SMC Pentax-A 645 35mm f/3.5 lens (shifted IC is 70.9mm and nominal IC for 645 is 70.8mm). This is conservative. In practice, I can easily shift 12mm in landscape. I routinely shift 12.5mm in Y (portrait) to create an effective field of view of a 26mm lens, and image quality at the edges is good at f/8. If there's nothing going on in the far-shifted zone (e.g., sky above a building), I can rise (shift) 15mm in landscape with my SMC Pentax-A 645 35mm f/3.5 lens before I reach the edge of the circle of illumination (defined by the point at which hard mechanical vignetting appears).
The take-away is that the image circle sizes in my spreadsheet for the film formats is conservative. This may be true of published image circles for technical camera lenses too (witness the many debates about how big the image circle of a lens actually is).
There are a lot of variables in play around image circles, including these (some of which will apply to all lenses, whether for film or digital):
- Image circle size increases with magnification for all lenses; at 1:1 it's twice as large as it is at infinity.
- "Image circle" is a sloppy term we all use for convenience. Strictly speaking, we need to distinguish between the circle of illumination (the whole illuminated area, bounded by the edges of mechnical vignetting) and the circle of good definition (the part of the circle of illumination where image quality is "good"). The difference can be important because the zone between the circle of good definition and the edge of the circle of illumination is only usuable if that part of the image has no detail, e.g., featureless sky during day or night. Distortion that is well behaved in the circle of good definition can also be wildly bad in this zone.
- Lenses within the same system may have different sized image circles, e.g., my Mamiya N 80mm f/4 L has a larger circle of illumination and circle of good definition than other lenses in the Mamiya N lineup; similarly, the image circle of the Pentax-A 645 45-85mm f/4.5 zoom gets larger as you approach the middle of the zoom lens, and shrinks back to "normal" at the long end.
- Some sensors don't play nice with some lenses when shifting. We all know how wide symmetrical lenses can have ugly lens cast and very strong light falloff with older non-BSI sensors. I've also seen examples where strange artifacts appear on some sensors but not others using the exact same lens and shift amount.
Last edited: