Warren T.
Well-known
Actually, this is one of the most divisive and confusing topics, and is really a "huge can of worms" :bang:.
However, as Roland mentioned, I'm sure you're going to run some tests for yourself to see how your adapted lenses work out. It will be fun, and then you'll tell us all about it.
--Warren
mfunnell
Shaken, so blurred
No. If past experience is any guide, we won't.
For example (but please, I'm not picking on anyone, just giving an example):
Is a comment which misses a key factor, namely the bit not in red in the quoted comment ("same print size and viewing distance, etc."). When taking that into account, the acceptable Circle of Confusion size is smaller for the smaller-size sensor (or, conversely, the CoC value to calculate "acceptable" sharpness needs to be smaller as the enlargement factor is greater for the same size print).
That smaller CoC value has the effect of making the apparent DOF more shallow for a smaller sensor using a lens of the same focal length than the (again, "apparent") DOF with a larger format. (The rule of thumb calculation - and that's all it is - for an "acceptable" CoC size, just BTW, is [Diagonal of Sensor in mm]/1400.)
There are lots of things that open and shut here. People, um, focus on the particular aspects that concern them and neglect things that don't. That's fine, as long as people don't start getting pedantic without fully thinking things through.
But this is the internet. People (including me; though I try not to) can get pedantic about things they don't understand as well as they think they do.
Which is why these DOF / angle-of-view / equivalent-lens-length discussions seem both interminable and futile. I've probably just contributed to that but I'll try to make it my last "contribution" as well.
...Mike
Last edited:
ssmc
Well-known
Here's how it works in bullet-form (thanks, Ansel!) :
- depth of field is proportional to the square of the subject distance (so moving twice as far away will give you 4x the DOF)
- depth of field is inversely proportional to the square of the focal length (so using a 25mm lens will give you 4x the DOF of a 50mm lens)
- depth of field is directly proportional to numerical f-stop (so f/4 will give twice the DOF of f/2, etc.)
Here's where the whole "unavoidable extra DOF" comes from: for example, say you want to take the "same" photo using a 135-format camera with a 50mm lens and a 4/3 camera with a 25mm lens. They give the same FOV from the same distance, so the framing and perspective can be the same. But precisely because the subject distance is unchanged, term 2 (above) comes into effect and the DOF is 4x greater, assuming you are shooting at the same ISO/shutter speed/f-stop, to get the same exposure. To get the same DOF in each image you would need an f-stop (not aperture) 4x smaller (numerically) on 4/3, e.g. f/1.4 on 4/3 and f/5.6 on 135, or f/2 and f/8, etc. You can see the problem in trying to duplicate a scene shot in different formats... to get extremely shallow DOF on 4/3 you either have to have ultra-fast lenses (impossible beyond some point) or settle for a different perspective (see below)
Now, consider using a "legacy" 50mm lens on 4/3; because the edges of the image are cropped off, it has the FOV of a 100mm lens. To get the same framing as a 50mm, you would need to move twice as far away. In this case, the first 2 terms (above) cancel out and you are getting the "real" 50mm DOF - so far so good. But... (and it's a huge BUT) since you've moved twice as far away, the perspective has changed (probably a lot) so the photo will no longer look like either 50mm on 135 or 25mm on 4/3. For some purposes this may not matter, for many others it will.
The light gathering power is a function of f-stop and is the same for any format (i.e. numerical exposure values are unchanged whether it's an 8"x10" view camera or a 1/1.8" compact digicam.
As for bokeh, this is determined by the lens in question only (I am ignoring weird interactions with the microlens array which may or may not occur). The degree of blur and it's character are fixed with respect to each individual lens - but in the case of 4/3, you may find yourself using (much) larger apertures to try and get a shallow DOF (if that's what you're trying to achieve) and sometimes odd things happen (especially toward the edges of the frame) at wide apertures.
These issues also exist for APS-C but they are a lot less severe because the ^2 terms have much more effect on a 2x crop than a 1.5x crop (2.25x vs. 4x)
Hope this helps
Scott
:- depth of field is proportional to the square of the subject distance (so moving twice as far away will give you 4x the DOF)
- depth of field is inversely proportional to the square of the focal length (so using a 25mm lens will give you 4x the DOF of a 50mm lens)
- depth of field is directly proportional to numerical f-stop (so f/4 will give twice the DOF of f/2, etc.)
Here's where the whole "unavoidable extra DOF" comes from: for example, say you want to take the "same" photo using a 135-format camera with a 50mm lens and a 4/3 camera with a 25mm lens. They give the same FOV from the same distance, so the framing and perspective can be the same. But precisely because the subject distance is unchanged, term 2 (above) comes into effect and the DOF is 4x greater, assuming you are shooting at the same ISO/shutter speed/f-stop, to get the same exposure. To get the same DOF in each image you would need an f-stop (not aperture) 4x smaller (numerically) on 4/3, e.g. f/1.4 on 4/3 and f/5.6 on 135, or f/2 and f/8, etc. You can see the problem in trying to duplicate a scene shot in different formats... to get extremely shallow DOF on 4/3 you either have to have ultra-fast lenses (impossible beyond some point) or settle for a different perspective (see below)
Now, consider using a "legacy" 50mm lens on 4/3; because the edges of the image are cropped off, it has the FOV of a 100mm lens. To get the same framing as a 50mm, you would need to move twice as far away. In this case, the first 2 terms (above) cancel out and you are getting the "real" 50mm DOF - so far so good. But... (and it's a huge BUT) since you've moved twice as far away, the perspective has changed (probably a lot) so the photo will no longer look like either 50mm on 135 or 25mm on 4/3. For some purposes this may not matter, for many others it will.
The light gathering power is a function of f-stop and is the same for any format (i.e. numerical exposure values are unchanged whether it's an 8"x10" view camera or a 1/1.8" compact digicam.
As for bokeh, this is determined by the lens in question only (I am ignoring weird interactions with the microlens array which may or may not occur). The degree of blur and it's character are fixed with respect to each individual lens - but in the case of 4/3, you may find yourself using (much) larger apertures to try and get a shallow DOF (if that's what you're trying to achieve) and sometimes odd things happen (especially toward the edges of the frame) at wide apertures.
These issues also exist for APS-C but they are a lot less severe because the ^2 terms have much more effect on a 2x crop than a 1.5x crop (2.25x vs. 4x)
Hope this helps
Scott
PentHassyKon
Established
Unfortunately it isn't that simple. It is complex because everytime I read about this topic it is stated w/o the additional statement in bold italic.
This is the crux of my confusion - one must add the additional statement in bold italic to "DOF is greater" because.....
So.... using the original picture posted by berlincontemporary in post # 6. To get the same result as the full frame, one has to step back hence effectively changing the DOF since one is now at a different distance. But then, now the perspective is different and...
oh hell!....
ferider
Veteran
Thanks for jumping in, Mike. I will refrain from further comments on DOF and refer to dofmaster.com.
I don't think "weird interactions" should be ignored. Film is much thicker, and just as the M8/M9 are more sensitive to RF mis-alignment than, say, an M6, I predict bokeh will look different too. When you think about a cone of light being projected on the medium, film cuts a much thicker slice out of that cone, than a thin sensor does, which should, qualitatively look different.
Raid, again, I would be really interested in seeing quality of bokeh on film vs. digital. Maybe it's the same, maybe not. We'll see.
Roland.
I don't think "weird interactions" should be ignored. Film is much thicker, and just as the M8/M9 are more sensitive to RF mis-alignment than, say, an M6, I predict bokeh will look different too. When you think about a cone of light being projected on the medium, film cuts a much thicker slice out of that cone, than a thin sensor does, which should, qualitatively look different.
Raid, again, I would be really interested in seeing quality of bokeh on film vs. digital. Maybe it's the same, maybe not. We'll see.
Roland.
Last edited:
raid
Dad Photographer
I appreciate such wealth of information being poured into this thread.
Roland: I will look into it.
Roland: I will look into it.
cosmonaut
Well-known
Brad Buszard
Established
Resource on a complicated issue
Resource on a complicated issue
Lots of information flying around here; most of it good, but none of it complete. I do not mean to critique: this is a complex issue, and I don't believe that I could adequately address it here either (even were my knowledge extensive, which it is not).
The least discussed aspect in the debates I have read is enlargement and cropping of the image. The acceptable circle of confusion (often stated as 1/167 inch) depends on a nominal viewing distance of 10 inches from the final image, which then invokes issues of standard viewing distance (which is usually reckoned as approximately equal to the diagonal of the final print). The circle calculations naturally must take into consideration the enlargement of the negative or sensor required to make the final print, but these may (or may not) be offset by the presumed step or three back that a viewer would take when viewing a larger print.
So analyses of DOF that are solely dependent on f-stop and object distance, while technically correct, ignore an additional host of issues.
We can't hope to cover the issue satisfactorily in such short space. My only contribution will be to ask that we share references. The best resource I've yet found on this subject is the 7th chapter of Stroebel's View Camera Technique. The LF forum site has some detailed analyses here:
http://www.largeformatphotography.info/
I find them less clear than Stoebel, but they go into greater depth on some details. Has anyone else found another good discussion by a trustworthy authority?
Resource on a complicated issue
Lots of information flying around here; most of it good, but none of it complete. I do not mean to critique: this is a complex issue, and I don't believe that I could adequately address it here either (even were my knowledge extensive, which it is not).
The least discussed aspect in the debates I have read is enlargement and cropping of the image. The acceptable circle of confusion (often stated as 1/167 inch) depends on a nominal viewing distance of 10 inches from the final image, which then invokes issues of standard viewing distance (which is usually reckoned as approximately equal to the diagonal of the final print). The circle calculations naturally must take into consideration the enlargement of the negative or sensor required to make the final print, but these may (or may not) be offset by the presumed step or three back that a viewer would take when viewing a larger print.
So analyses of DOF that are solely dependent on f-stop and object distance, while technically correct, ignore an additional host of issues.
We can't hope to cover the issue satisfactorily in such short space. My only contribution will be to ask that we share references. The best resource I've yet found on this subject is the 7th chapter of Stroebel's View Camera Technique. The LF forum site has some detailed analyses here:
http://www.largeformatphotography.info/
I find them less clear than Stoebel, but they go into greater depth on some details. Has anyone else found another good discussion by a trustworthy authority?
lynnb
Veteran
ssmc, you mentioned
4 stops is huge. From sample images I've seen that didn't look quite right, so I plugged the figures into DOFmaster. Here's the results:
5DII, 50mm, f8, subject distance 10'
near limit 7.77'
far limit 14'
total 6.28'
Panasonic GH2, 25mm, f4, subject distance 10'
near limit 7.75'
far limit 14.1'
total 6.34'
So according to those calculations, it's 2 stops, not four. A two stop difference admits 4x the amount of light. I assume that's what you meant?
4 stops is huge. From sample images I've seen that didn't look quite right, so I plugged the figures into DOFmaster. Here's the results:
5DII, 50mm, f8, subject distance 10'
near limit 7.77'
far limit 14'
total 6.28'
Panasonic GH2, 25mm, f4, subject distance 10'
near limit 7.75'
far limit 14.1'
total 6.34'
So according to those calculations, it's 2 stops, not four. A two stop difference admits 4x the amount of light. I assume that's what you meant?
dave lackey
Veteran
One more time...what is a "legacy" lens?
tlitody
Well-known
One more time...what is a "legacy" lens?
It's pimp speak for old and out of date. But its fundamentally wrong as some us know that contemporary (the opposite of legacy) isn't necessarily better.
raid
Dad Photographer
People refer to legacy lenses as if they mean the [often older] lenses from film cameras.
Dave: I asked the same question a while ago. I had no clue at all what was meant by this term.
Today, I used the 35mm/2 Summicron (8 elements) on the EP-2. It is a legacy lens, but it is a fine lens. The crop shows what you would get from a 70mm lens.
Dave: I asked the same question a while ago. I had no clue at all what was meant by this term.
Today, I used the 35mm/2 Summicron (8 elements) on the EP-2. It is a legacy lens, but it is a fine lens. The crop shows what you would get from a 70mm lens.
Last edited:
tlitody
Well-known
Legacy is term that came to the fore in the IT industry around 10 to 15 years ago. It is used to mean old software systems that are still in use. Needless to say the term is now creeping into all sorts of other arenas.
dee
Well-known
I just use my little CV 35 f2.5 and , various Leica screw lenses on the G1 ... and adapt to what I see ... this follows on from a Leica Dig 3 [ 4/3rds ] which adapts to my Minolta Rokkors ... and my M8 adapted to Sonnar clones all 50mm .
Can't get blinded by science LOL .
Incidentally , should you wish to minimise the crop factor , the Samsung range , developed from the Pentax tie up , are all APSc , though adapters are a touch more expensive .
Can't get blinded by science LOL .
Incidentally , should you wish to minimise the crop factor , the Samsung range , developed from the Pentax tie up , are all APSc , though adapters are a touch more expensive .
rxmd
May contain traces of nut
If by "look" you mean the relative sizes of background and foregound objects, it will indeed give the same "look".
Try it one day. Take your favourite FF or film camera, take a picture with a 50 and a 100, crop the 50 picture to the angle of view of the 100 picture and enlarge it accordingly, and hold the two pictures next to each other. We've had a number of examples here (I remember posting examples from a 15 and a 50).
biomed
Veteran
One more time...what is a "legacy" lens?
I think "alternative lens" is a better term. I always understood "legacy lens" when used in context with the M4/3 system to be a lens not native or designed for the M4/3 mount that is mounted to a M4/3 camera body by means of a lens adapter. For example, mounting a Nikon F mount lens on an EP1 body using a Nikon F to M4/3 aadapter.
isorgb
Well-known
I did a little test with the Fuji S5100 (6.5x crop) and the Olympus E-1 (2x crop).
Real parameters for Fuji: 51mm f3.1 (equivalent ~300mm)
and Olympus 150mm f10 (300mm equivalent).
How do you think? What are the differences between the images?
I also think about bokeh. Is it different?
First theory: No, there is no difference, they are equivalent.
Second theory: Yes, there is a big difference, they are totally different focal length.
Real parameters for Fuji: 51mm f3.1 (equivalent ~300mm)
and Olympus 150mm f10 (300mm equivalent).
How do you think? What are the differences between the images?
I also think about bokeh. Is it different?
First theory: No, there is no difference, they are equivalent.
Second theory: Yes, there is a big difference, they are totally different focal length.
Last edited:
ssmc
Well-known
Good pick-up - except that DOFMaster uses a "circle of confusion" of 0.03mm for the 5DII and 0.015mm for the GH2, in an attempt to account for the smaller format of 4/3 compared to 135 (this would be a good way to look at it for film, assuming the CoC rules were adhered to), and the extra factor of 2x is where you would get 4 stops if you were using the same CoC for both formats (partly why I ignored CoC in my original post, for the rest of the reasons, read on...)If that were the case, you would indeed have to shoot the 25mm lens at f/2 to get the same DOF as the 50mm lens at f/8 for the same subject distance. This scenario would be appropriate if both cameras had the same number of pixels and therefore the same (nominal) spatial resolution in lines per picture height (e.g. a 12MP D700 vs. a 12MP MFT like an Oly Pen)
But... CoC is a concept based on a "standard print size and viewing distance" which might have been okay for film in the mid-20th century, and in the case of DOFMaster (for example) it only takes into account format, not the number of photosites, which provides a hard limit on resolution - and one that can vary drastically between cameras with the same sensor size, which would not be the case with film (Tri-X in an M3 is the same as Tri-X in an F6). For example, DOFMaster gives a CoC for all 1.5x APS-C cameras of 0.02mm (0.03/1.5), regardless of MP count, so the 6MP D70 has the same CoC as the 16MP D7000! The 5DII is given a CoC of 0.03mm (more or less the old stadard for 135 film), yet it has 6.4um pixels, while the D70 has a CoC of 0.02mm but 7.9um pixels! By this reasoning, resolution differences on any given format (and between different fomats) are ignored, or worse.
I have become convinced that CoC is of dubious use for digital unless it's being related to prints, and even then, only in a very specific way. If you're only ever going to print at the "standard size" (at say 300dpi), anything over 8MP would be superfluous. But nowadays we can print much larger, even with film, since emulsions have improved tremendously.
Happy shooting!
Scott
isorgb
Well-known
Good pick-up - except that DOFMaster uses a "circle of confusion" of 0.03mm for the 5DII and 0.015mm for the GH2, in an attempt to account for the smaller format of 4/3 compared to 135 (this would be a good way to look at it for film, assuming the CoC rules were adhered to), and the extra factor of 2x is where you would get 4 stops if you were using the same CoC for both formats (partly why I ignored CoC in my original post, for the rest of the reasons, read on...)If that were the case, you would indeed have to shoot the 25mm lens at f/2 to get the same DOF as the 50mm lens at f/8 for the same subject distance. This scenario would be appropriate if both cameras had the same number of pixels and therefore the same (nominal) spatial resolution in lines per picture height (e.g. a 12MP D700 vs. a 12MP MFT like an Oly Pen)
But... CoC is a concept based on a "standard print size and viewing distance" which might have been okay for film in the mid-20th century, and in the case of DOFMaster (for example) it only takes into account format, not the number of photosites, which provides a hard limit on resolution - and one that can vary drastically between cameras with the same sensor size, which would not be the case with film (Tri-X in an M3 is the same as Tri-X in an F6). For example, DOFMaster gives a CoC for all 1.5x APS-C cameras of 0.02mm (0.03/1.5), regardless of MP count, so the 6MP D70 has the same CoC as the 16MP D7000! The 5DII is given a CoC of 0.03mm (more or less the old stadard for 135 film), yet it has 6.4um pixels, while the D70 has a CoC of 0.02mm but 7.9um pixels! By this reasoning, resolution differences on any given format (and between different fomats) are ignored, or worse.
I have become convinced that CoC is of dubious use for digital unless it's being related to prints, and even then, only in a very specific way. If you're only ever going to print at the "standard size" (at say 300dpi), anything over 8MP would be superfluous. But nowadays we can print much larger, even with film, since emulsions have improved tremendously.
Happy shooting!
Scott
OK, but where is difference?
CoC, film, digital sensor, resolution - what does this mean for the appearance of pictures?
Another resolution of the sensor affects the DOF effect?
rxmd
May contain traces of nut
Simply put - if your sensor is smaller, you have to enlarge the pictures more, which is why you then have to take different resolutions into account.
If you look at two pictures at the same size, and the sensors were the same size too, the resolution of the sensor makes at most a minimal difference DOF-wise, if at all.
Last edited: