My 35 Digitar possible shifts diagram

[tashley]

Now this is really anal but my excuse is that it's part of my ongoing process of getting all the details tied down so when I shoot in the field I get what I was expecting!

This diagram adapts Schneider's own for this lens at F11. As they state, if you are shifting ONLY vertically then you have 19mm up and 19mm down to play with. Similarly you have 17mm for horizontal.

But if you want to shoot for maximum field of coverage, in other words shifting horizontally AND vertically to get the largest 'virtual sensor' possible, then you have to avoid the vigged corners.

I made the following diagram and measured what maximum COMBINED shifts I could use at F11 to get both 4/3 and 3/2 frame ratios.

The results are:

4/3
Vertical 7mm in each direction
Horizontal 12mm in each direction

3/2
Vertical 5mm in each direction
Horizontal 14mm in each direction

So for example if you want a classic 3:2 ratio frame you can do four frames of

Top Left 14mm left 5mm up
Bottom Left 14mm left 5mm down
Top Right 14mm right 5mm up
Bottom Right 14mm right 5mm down

You might in fact choose to use greater movements and more frames so as to get the sweet spots placed optimally but the advantage of my calculations is that if you are shooting without a viewfinder or focus screen, you can use these numbers to accurately gauge your framing by shoot-and-chimp.

I'm sure someone with better math than me could have worked this out mathematically but it only took five minutes! Please let me know if there are any obvious mistakes!

Best

Tim
View attachment 14627

ps for a 4:3 image this gives a total virtual sensor of about 80 mp
for a 3:2 image this gives a total virtual sensor of about 78 mp

 

[carstenw]

Have you taken the corners into account as you calculated the radius of the circle? With this I mean that if you calculate the shift along the axis, the corners hit the real radius 1/2 sensor height above and below the axis, for horizontal, for example.

 

[Lars]

Some of the Digitars illuminate a much wider IC than specs indicate, so if you can get away with less resolving power in the corners there might be more to use.

 

[tashley]

Not following you there Carsten: I just made pretty boxes and moved them around in a scale diagram of the image circle, then measured the shifts.

 

[carstenw]

I was just wondering how you made the big circle. It now sounds like you moved a sensor-sized box 19mm left and then drew the circle. I was wondering if you had made the circle 19mm+1/2 sensor width, which is too small.

 

[Jack]

Why not just shift fully and crop inside the vignette after the fact? You end up with the same total image coverage and this way is a lot easier

 

[tashley]

I was just wondering how you made the big circle. It now sounds like you moved a sensor-sized box 19mm left and then drew the circle. I was wondering if you had made the circle 19mm+1/2 sensor width, which is too small.

No, I made a 'as quoted' (i.e. 90mm) circle for the image circle, then made to scale a sensor sized box of 49x37mm (the same size as my P45+ sensor) which I placed dead centre in the image circle. Then I made the largest 4:3 box that would fit in the circle and the largest 3:2 box that would fit in the circle, placed them, and measured the number of millimetres from teh relevant edges of the main sensor box the the respective 'virtual sensor' boxes to determine those shifts which created the larger virtual sensor.

It seems to work too!

;-)

 

[tashley]

Why not just shift fully and crop inside the vignette after the fact? You end up with the same total image coverage and this way is a lot easier

Yup, that's what I always have done. But with this method you get an accurate idea, in the field, of exactly what your larger 'virtual sensor' framing is and you don't therefore risk missing off a bit of the subject that you later have to crop because you didn't use the patented tashley method!

There is method (just) in my madness! For those with finders that indicate shift (I don't use one) using my numbers will show you the frame quite well too!

 

[tashley]

Some of the Digitars illuminate a much wider IC than specs indicate, so if you can get away with less resolving power in the corners there might be more to use.

I agree but I am after sharp edge to edge so using Schneider's image circle estimate means I can be pretty sure of getting that. Of course with some types of subject matter you can relax that assumption.

 

[carstenw]

Possibly you can relax the image circle upwards quite a bit... If there is sky there. Vignetting can be fixed if there is no detail

 

[Bob]

Yup, that's what I always have done. But with this method you get an accurate idea, in the field, of exactly what your larger 'virtual sensor' framing is and you don't therefore risk missing off a bit of the subject that you later have to crop because you didn't use the patented tashley method!

There is method (just) in my madness! For those with finders that indicate shift (I don't use one) using my numbers will show you the frame quite well too!

You PATENTED squaring the circle?
-bob

 

[tashley]

You PATENTED squaring the circle?
-bob

Yes™