The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Circle of Confusion and current MFDBs

GrahamWelland

Subscriber & Workshop Member
I was wondering what CoC's people were using for their digital backs? I did a search but didn't find any consistent numbers. Ihave some from various sources but I was interested what folks had settled on for the backs.

Looking for numbers for P40+, P65+/IQ160 & Aptus II 12/IQ180. I just want to calibrate my DoF calculator which suggests 0.0354, 0.0435 for P40+/P65+ but I'm not confident with these numbers for critical hyper-focal setting. (I know that I should go out and measure the lenses/acceptable DoF at each f-stop but I was hoping to find a good number to plug in to my calculator).
 

Shashin

Well-known member
Why not start with the DoF calculator numbers which seem right and then adjust for personal preference. DoF should be judged with prints at correct viewing distance--equal to the print area diagonal. DoF will not appear correct at 100% on a monitor.
 

Shashin

Well-known member
Have you determined an acceptable CoC for another format that you like? It should be relatively easy to then convert those numbers to your larger sensors.
 

Bob

Administrator
Staff member
I use twice the pixel pitch.
Maybe sort(2) * pixel pitch is more correct for max resolution.


However,
if all you are printing at a specific maximum size then you can base it on that.
-bob
 

GrahamWelland

Subscriber & Workshop Member
In the end I ran the numbers via the Alpa worksheets and for my typical 16x20 print @ 360dpi, and standard viewing distance (diagonal) and 5lpm eyesight & 7lpm for the print it calculates out to 0.025 CoC for a 6um sensor. I can live with that.

The "Zeiss Formula" suggests 0.032 & 0.39 for P40+ & P65+ ...

For pixel peeping I can see that Bob's 1.414 x 6um would be 0.0085 which of course explains why pixel peeping DoF with digital backs is a frustrating exercise!

For my longer lenses the focus stacking continues ... :rolleyes:
 

greygrad

Member
This is a useful resource. The diffraction limit calculator returns the CoC based on the given variables (print dimensions, viewing distance, eyesight, resolution, camera type and aperture). Unfortunately, the 'camera type' list is a bit limited.
 

dick

New member
The theory is that you should always get the picture as sharp as you can, because someone might want a 3m * 4m print of it (so you ignore print size)... but when that means that you will have to stack 200 300Mpx shots... you need a £10k Mac!

For web work I leave the H4D-60 at home and take my GH2!
 

Woody Campbell

Workshop Member
In the end I ran the numbers via the Alpa worksheets and for my typical 16x20 print @ 360dpi, and standard viewing distance (diagonal) and 5lpm eyesight & 7lpm for the print it calculates out to 0.025 CoC for a 6um sensor. I can live with that.

The "Zeiss Formula" suggests 0.032 & 0.39 for P40+ & P65+ ...

For pixel peeping I can see that Bob's 1.414 x 6um would be 0.0085 which of course explains why pixel peeping DoF with digital backs is a frustrating exercise!

For my longer lenses the focus stacking continues ... :rolleyes:
Graham - .025 for a 6 micron sensor suggest .021 or .022 for my IQ 180. Using the Alpa calculator this gives a hyperfocal distance with my 35 xl at f 12.5 of 16 feet or so, which is consistent with my experience with this lens.
 

f8orbust

Active member
I'm not so sure about the numbers the Alpa spreadsheet returns for CoC. For my P45+ it gives a baseline CoC (i.e. ignoring print size, viewing distnace etc.) of 0.042. This is the usual number thrown out when basing the CoC on a single pixel. However this is only valid (as far as I understand it) for scanning backs, foveon sensors and achromatic backs. For those DBs using a bayer filter, the CoC has to be close to 2 pixel widths - which in the case of my P45+ makes the CoC 0.0136.

If you use a PC, there's a great utility called Barnack which has all the Phase, Leaf and Hasselblad data in it. It's free and you can download it here.
 

Jack

Sr. Administrator
Staff member
I have discussed this ad-nauseum with a group of other photo tech types. The general answer is always something like, "Well, the second we go to 100% view to assess file quality, we are effectively adopting single-pixel CoC." And of course that's correct. A few years ago many of us started using 50% view to replicate (very large) print resolution. I find this to be a pretty valid estimator, therefore I adhere to Bob's assessment of 2x pixel pitch -- but only for very large prints. Bottom line is you can get away with a lot larger CoC for smaller prints. Basically I think we've come full circle to where in order to determine an appropriate CoC we first need to consider our final print size and viewing distance...

My .02, YMMV.
 

dick

New member
Yes... if you want the best possible pixel quality for a 1Mpx file for the web, you need to down res by 10* if your camera has an Anti-Aliasing filter, and by a further 4* if your camera uses Bayer interpolation, so DSLRs with less than 40Mpx are not ideal, even for web work.

I'm not so sure about the numbers the Alpa spreadsheet returns for CoC. For my P45+ it gives a baseline CoC (i.e. ignoring print size, viewing distnace etc.) of 0.042. This is the usual number thrown out when basing the CoC on a single pixel. However this is only valid (as far as I understand it) for scanning backs, foveon sensors and achromatic backs. For those DBs using a bayer filter, the CoC has to be close to 2 pixel widths - which in the case of my P45+ makes the CoC 0.0136.

If you use a PC, there's a great utility called Barnack which has all the Phase, Leaf and Hasselblad data in it. It's free and you can download it here.
 

dick

New member
Yes, great, Barnak says that at f1.7, 20mm @ 2m, a GH1 has one and a half as much DOF as my H4D-60 has at f8 at the same field of view! ...and at f11 the Hasselblad has the same dof.

I have a GH2, I do not know how different it would be. I do use both cameras for the same job - dancing girls and gymnasts, pictures at rosalindcaplisdanceacademy soon.

If you use a PC, there's a great utility called Barnack which has all the Phase, Leaf and Hasselblad data in it. It's free and you can download it here.
 

f8orbust

Active member
The Alpa spreadsheet is calculating the CoC for a DB based on the circle of confusion of a 35mm film camera, in this case using a reference of 0.030mm as the starting point, and multiplying it by the 'enlargement factor' (for want of a better term) of the DB.

My P45 has a diagonal of 61.36mm, and 35mm a diagonal of 43.27mm.

The spreadsheet calculates the CoC thus: (61.36/43.27)*0.030 = 0.0423mm or, to put it another way, over 6.3 pixels.

This is a very generous CoC. If I used it, then the hyperfocal distance of my 28mm lens would be 2.3m, which it clearly isn't (that's more the hyperfocal distance of that lens on a much smaller format).

In the field I've found that a CoC of 2 x pixel pitch is pretty much on the money, which is a shame because it's so small. In the case of my P45, that's 0.014mm or 3x less than the Alpa spreadsheet. This means that the hyperfocal distance of my 28mm is at 7.2m, which is closer to what I've found.

Yes, great, Barnak says that at f1.7, 20mm @ 2m, a GH1 has one and a half as much DOF as my H4D-60 has at f8 at the same field of view! ...and at f11 the Hasselblad has the same dof.
It does? Using the 35mm film CoC reference and these numbers I get DoF for the GH1 = 0.5m and for the H4D = 7.3m. Using the 2 pixel width reference, I get DoF for the GH1 = 0.3m and for the H4D = 0.8m.
 

Jack

Sr. Administrator
Staff member
Agreed, most calculators are a bit too generous. I like using 2 or 3 pixels. It does limit DoF and aperture freedom somewhat (and why I went to a tech cam for tilts), but seems to align nicely with big prints viewed close.
 

GrahamWelland

Subscriber & Workshop Member
You need to work backwards from your print dimensions in the spreadsheet to get reasonable numbers. I agree that the tables of CoC are too generous which is what drove me to start this thread in the first place. It's not just Alpa but just about every other CoC source I came across.
 

Jack

Sr. Administrator
Staff member
Correct Graham, ALL CoC tables are generous -- my belief is because they were built around film which at the time, had no finite resolution limit compared the to equipment they were reproducing it with and intended use; most prints were not intended to be both very large AND viewed very closely at the same time, so the formulas worked for that.

Enter digital, and IMHO we need a new convention; we do now view big prints close up, even with magnified reading glasses on. It is impressive to me that I can now put a loupe to a 24x32 print and see more detail than I can see with my naked eye -- I like that :D
 
Top