Simply Sensor Size

[rxmd]

Hi Fitzi,

the angle of view on a crop vs. a longer focal length is the same, but the depth of field should still be different.

For example, on 35mm film, a 50mm lens set at 1 m and f/4 will deliver a DOF of 9.1 cm. With an APS-C size sensor, a 50mm-equivalent lens at the same settings will deliver a DOF of 12-something cm. A 75/1.4 on a M7 wide open at 3 m distance delivers a shallow DOF of 13 cm; a 50/1.4 on an R-D1 already delivers 16 cm, and in order to get the same DOF as on the larger format, you need at least a 50/1.1, which can get expensive if you've just paid for a digital body.

This is I think a well-grounded reason why many DSLR people are complaining about small sensors. The problem isn't so much that you need new wideangles, after all; the problem is you need even faster lenses for shallow DOF, which either aren't available at all (there is no 40/0.9 to replace the Noctilux) or are prohibitively expensive. The only thing you can do is to make the crop factor small enough and hope that people will adapt.

Philipp

 

[varjag]

Digital camera sensors are unusual because the size is a constant but the number of transistors is flexible. Here are some comparison numbers for context: Intel's current top of the line Pentium chip (3.7 ghz "Presler" core) costs about $1050 and has a die size of 162 mm2. The Core Duo chip that's been all over Apple's recent machines has a die size of 90.3 mm2. The die size of a 24x36mm sensor is 864 mm2. That's monstrously huge! Economies of scale can go a certain distance, but you can't get around the fact that the chip fabrication plant has simply got to burn more silicon to produce full-frame photo sensors. Worse, the overall yield of useable chips decreases as sensor size increases due to flaws in the etching process.

However, photo sensors have very regular structure, and much greater tolerance for error than logic circuits of CPUs. Dead pixels are allowed. All this leads to order of magnitude better yields for the same square.

 

[jaapv]

Hi Fitzi,

the angle of view on a crop vs. a longer focal length is the same, but the depth of field should still be different.

For example, on 35mm film, a 50mm lens set at 1 m and f/4 will deliver a DOF of 9.1 cm. With an APS-C size sensor, a 50mm-equivalent lens at the same settings will deliver a DOF of 12-something cm. A 75/1.4 on a M7 wide open at 3 m distance delivers a shallow DOF of 13 cm; a 50/1.4 on an R-D1 already delivers 16 cm, and in order to get the same DOF as on the larger format, you need at least a 50/1.1, which can get expensive if you've just paid for a digital body.

This is I think a well-grounded reason why many DSLR people are complaining about small sensors. The problem isn't so much that you need new wideangles, after all; the problem is you need even faster lenses for shallow DOF, which either aren't available at all (there is no 40/0.9 to replace the Noctilux) or are prohibitively expensive. The only thing you can do is to make the crop factor small enough and hope that people will adapt.

Philipp

This is real, but as I explained in my original post, on a 1.33 sensor the real difference in the DOF illusion is about 1/2 a stop compared to 35 mm frame, not enough to get knickers twisted. Focussing tolerances, camera shake, cropping, even dust in the air will account for more. Funny thing, one never hears these arguments in 645 vs 35 mm.....

 

[fitzihardwurshd]

Hi Fitzi,

the angle of view on a crop vs. a longer focal length is the same, but the depth of field should still be different.

Philipp

This is true of course and if DOF was what Willie meant saying it may look very different he is right.

I meant rather the compressed perspective which you can get from a cropped 50mm shot as well as from a tele lens. Which is hard to believe until one watches a demo.

Technically the APS sensor should be enough for the most purposes, tho noise is still a bad issue. But I think the crop factor is the most important reason why people want a full frame sensor, who wants to mess his worthful lenses this way ? An APS chip needs it's own lens design.

Fitzi

 

[jaapv]

I rated this thread "excellent" because of the positive way this potentially inflammatory subject was responded to. 🙂

 

[photogdave]

You are comparing apples to oranges. Of course the lens in your compact Leica is going to be designed for minimal distortion, otherwise it would be useless and no one would buy it!
My point is that in the average 35mm SLR lens group, ultra-wide lenses distort more than regular wide. (ie 17mm vs 28mm). Obviously there are exceptions to the rule but not everyone can the afford more expensive corrected lenses.
I know the 17-40 L is a better lens but it wasn't available to me at the time. However, it still gives you a FOV of 27mm on the APS-C sensor and wouldn't have the same characteristics as a 28mm on full-frame.
Can we talk about rangefinders now please?

 

[John]

originally posted by jaapv
I rated this thread "excellent" because of the positive way this potentially inflammatory subject was responded to.

I agree and enjoyed reading through this thread and believe I have learned a thing or two.
However, could a poll be devised? Here's my take.
Suppose Leica or another manufacturer plans on introducing a new digital rangefinder camera. There are three identical models to choose from for the exact same price. The only difference being sensor size choices of 1.5, 1.33, or full frame. Choose which one you would want. 🙂

 

[jaapv]

You are comparing apples to oranges. Of course the lens in your compact Leica is going to be designed for minimal distortion, otherwise it would be useless and no one would buy it!
My point is that in the average 35mm SLR lens group, ultra-wide lenses distort more than regular wide. (ie 17mm vs 28mm). Obviously there are exceptions to the rule but not everyone can the afford more expensive corrected lenses.
I know the 17-40 L is a better lens but it wasn't available to me at the time. However, it still gives you a FOV of 27mm on the APS-C sensor and wouldn't have the same characteristics as a 28mm on full-frame.
Can we talk about rangefinders now please?

I thought the EF-S series was designed for APS sensors?

 

[photogdave]

They are. That has nothing to do with my point:
"Obviously there are exceptions to the rule but not everyone can the afford more expensive corrected lenses."
I'm sure (or I would hope) the new EF-S 17-55 f2.8 is less distorted than the 17-85 but it's twice the cost. I would rather put my money into a full-frame digital camera and have lenses that perform well on a film body AND the APS-C sensor digitals, rather than limit myself to mid-range digital body and lenses that will only work with it and its future upgrades. Regular EF lenses will work on FOUR types of Canon EOS bodies.
But going way back to my original comment on your original post, most of what I love about photography is the act of capturing the moment. Having shot with both sensor formats, I simply LIKE full-frame more and that's enough for me!

 

[HenningW]

Funny, how you take a crack at marketing executives, yet your comments on full frame sound like they come straight from Nikon marketing!😀
I will add only two things:
First, one aspect of photography often overlooked by the general public but cherished on this forum, is the joy in the actual process of capturing the image. Bringing the viewfinder to the eye, focusing etc. The world looks a lot brighter through a full-frame viewfinder compared to the little tunnels you have to squint through on smaller-chipped DSLRs. That alone has sold more than a few 5Ds.
Second, for the working professional, it can be imperative that a lens behaves the way it was designed for. Although you mentioned FOV and DOP, you didn't mention distortion. Wide angle lenses distort much more than standard, even if your not using the whole image circle. Example: shooting a wedding with the Canon 20D and EF-S 17-85 (so in this case I was using the whole image circle). Despite being a digital design, at 17mm the lens still give you a FOV of almost 28mm. It looks like 28mm through the viewfinder. From past experience of shooting with 28mm lenses I am confident that I can frame people close to the edges without their heads going too wobbly. But wait! IT IS STILL A 17mm LENS! When I finally view these group shots on a full-size monitor the people at the edges have oval heads. Not too cool.
Rather than change my technique and style, I would buy a 5D if was to shoot weddings on a regular basis.
Actually after the whole experience I decided to return to film for weddings anyway (IF I ever shoot another!)

Sorry, I came upon this post rather late as I don't read the list all that often.

The above conclusions re: 17mm on a 20D and 28mm on a 5D are incorrect.

The 17mm lens will produce exactly the same amount of 'egg-headedness' or 'going wobbly' on a 20D as a 28mm lens will on a 5D, as long as the people are not really close to the camera (less than 5ft or so).

This presupposes that linear distortion in the lenses is negligible or corrected. In fact, if barrel distortion is present, the 'egg-headedness' will be less than with a better corrected lens.

Distortion of 3 dimensional objects, when using a well-corrected lens with minimal linear distortion (2 dimensional distortion) is purely a function of angle of view, and the self same lack of linear distortion. Except for linear distortion correction (and remember, better corrected lenses do a worse job with 3-D objects in corners than 'poor' lenses with barrel distortion) lens design has no effect on 3-D distortion.

It matters not whether I use a 4mm lens on a tiny P&S sensor to give me a 75degree angle of view, or a 17 on a 10D, a 28mm on a 5D or a film Leica or Canon, or a (approx) 55mm on a 6x7 or (approx.) 100mm on 4x5. The three dimensional distortion (and angle of view) will be identical. And yes, I have all those cameras and lenses.

There are good reasons to use a 5D, or a 20D, but this 3-D distortion thing which I read here and in other places is a non-starter.

Another thing that has been posted here is that wide angle performance is better on the 5D than on the cropped cameras. This is not true. Canon's wideangles are generally not as good as some of the competitions, and the 5D only shows that better. The 16-35 is not as good on the 5D as the 10-22 is on the 20D. The 20/2.8 is really bad, the 14/2.8 is poor and while I enjoy and often use the 24/1.4, it's performance on the 20D is a lot more satisfying than on the 5D. The 24TSE is also not great. The only really good wideangle, and the widest really good lens is the 35/1.4. BTW, I use a 12-24 Sigma because it has noticeably better performance than the 20 or 16-35. At present I do not have the 14 and 20 of the above, because of the Sigma. I should also sell the 16-35.

With respect to the Leica M8, the thing that annoys me about not having it be full frame is that I won't have a lens with an approximate angle of view of 35mm that has a maximum aperture of 1.4. Other than that, I'm fine with it.

Henning

 

[photogdave]

All that math makes my head hurt. I only speak from real world shooting experience.
Not to say that you don't. I'm quite sure you and jaapv are both smarter than me.

 

[willie_901]

Try Number 2

Try Number 2

Ever since this thread began I've been trying to come up with a scenario to explain how the statement – "So what if my sensor has a 1.5 crop factor? All I have to do is use a 35 mm lens to achieve the results I'm used to with a 55 mm lens" – is somewhat misleading.



A few years from now two photographers with digital rangefinder cameras decide to take a walk in the park. Photographer A's camera has a full-frame sensor and photographer B's camera has a smaller 1.5 crop-factor sensor.

They come upon a group of large dogs sitting on the grass about 15 feet from the sidewalk. "A" notices there is a "Keep Off the Grass" sign next to the sidewalk and to the left of the dogs. "A" also sees a one-story flat-roofed building about 100 feet behind the dogs. "Park Police" is painted along the top of the building in large letters.

"A" fits a 55 mm prime lens and moves about until the composition in the finder seems right. He takes a picture.

The photographers view the result on A's LCD screen. The sign is on the far left side of the frame and is half as tall as the dogs. The dogs are in the middle. The Police building is at the very top of the frame on the far right side.

"B" like the image so much he decides to take one too. He thinks, "if A used a 55 mm lens I'll use a 35 mm lens." "B" stands exactly where A stood and frames the image in his finder to match what he saw on A's LCD and takes a picture.

B is surprised that his image is different than A's. The angle of view in both images is identical. But B notices the sign is the same height as the dogs and the building is smaller.

The relative size of objects in an image from a 35mm focal-length lens is different than a 55 mm lens.

"B" moves further away from the dogs and takes another picture. "B" thinks, "with my 48 MP 64 bit sensor I'll just crop the image and no one will notice the difference". "B" takes photos until the ratio of the sign's height to the dogs' height matches A's image. Then he crops the image right there on the camera's LCD screen. "B" is surprised again. Since B moved further away from the Police building, it appears to be even smaller than it did in his first image.

Finally B mounts a 55 lens on his camera. "B" moves away from the dogs and takes pictures until the angle of view is similar to A's image. He frames the scene and takes a picture. Now the relative heights of the sign, the dogs and the building are the same as in A's image.

The only purpose of this long-winded story is to illustrate how the relative size of objects in an image depends on the len's focal length. Even when the angle of view and DOF are adjusted to be identical, a 35mm lens on a 1.5 crop camera will not always give the same image as a 55mm lens on a full-frame camera.

willie

 

[Dougg]

"B" like the image so much he decides to take one too. He thinks, "if A used a 55 mm lens I'll use a 35 mm lens." "B" stands exactly where A stood and frames the image in his finder to match what he saw on A's LCD and takes a picture.

B is surprised that his image is different than A's. The angle of view in both images is identical. But B notices the sign is the same height as the dogs and the building is smaller.

The relative size of objects in an image from a 35mm focal-length lens is different than a 55 mm lens.
...

Finally B mounts a 55 lens on his camera. "B" moves away from the dogs and takes pictures until the angle of view is similar to A's image. He frames the scene and takes a picture. Now the relative heights of the sign, the dogs and the building are the same as in A's image.

The only purpose of this long-winded story is to illustrate how the relative size of objects in an image depends on the len's focal length. Even when the angle of view and DOF are adjusted to be identical, a 35mm lens on a 1.5 crop camera will not always give the same image as a 55mm lens on a full-frame camera.

Greetings, willie, and with respect I suggest you actually perform the above experiment. In doing so you will discover you are incorrect. The relative sizes of objects in the pictures are entirely and solely dependent on camera location. The angle of view of the lens determines how much stuff is included in the frame. Crop the photo to include the same stuff as the narrower-view lens and the relative sizes of near and far objects will be exactly the same. Try it! 🙂

 

[Avotius]

despite all this crop frame vs full frame mumbo jumbo, in the end all I care about is my 35mm lens being a 35mm lens. I didnt buy a 50, thats not what I want to use.

 

[VinceC]

There's not too much I can add here. I've been saying for a long time that the 35mm format became so popular because it WAS the digital of its era -- compared to middle and large formats of the 1930s, 40s, 50s, 35mm's smaller "sensor" size allowed for smaller lenses, more portability and greater depth of field. Calculations aside, a 35mm format photo has much more depth of field than a larger format photo made with an 80mm "normal" lens.

What's happening with digital today is simply another evolution of formats.

The reason we're using 35mm film is fairly accidental ... it became the standard film for the motion-picture industry in the early 20th century and so was made in large batches with consistent emulsions, catching the attention of camera-enthusiast Oskar Barnack as he experimented with a way to make his beloved photography hobby more portable. But we aren't using the same format as motion pictures anymore. They have moved on to a 70mm "wide-screen" format, and not many people are lamenting the loss of the old-fashioned TV-shaped screens.

In addition, 35mm cameras don't use the film the same way as motion-picture cameras did. Movie cameras fed the film roll vertically across the shutter, giving the equivalent of a half-frame image. Film emulsions of the 1920s were quite inferior to what we have today, if you've ever looked at stills from silent movies, even restored ones, they were grainy and prone to blown highlights. It's not such a big deal because silent movies were projected at 16-18 frames per second, resulting in a 16-fold increase in the amount of "pixels/grain" per second.

Barnack saw that the motion-picture format didn't translate well to still photography, so he opted to use more of the film surface area by running the roll horizontally through the camera while keeping a horizontal frame ... instead of the 18x24mm movie-film standard, he used 24x36mm.

As most of us know, this isn't all that standard a dimension. It's wider than a US-standard 8x10inch print -- using just 24x33.3mm of the negative. So, for decades, photographers and photolabs have been slicing off 9 to 10 percent of the image on the negative. My scanning software still tries to do this if I don't change its default settings.

Yes, for about 35 or 40 years, the 35mm format was THE standard of photography for most users, an era that ended a few years ago -- running roughly from 1960 to 2000. Before 1960, most consumers and many preofessionals used a larger format. And throughout the 1960s, 70s and 80s, many consumers were quite happy with smaller formats such as the Kodak Instamatic. But we are now adapting to new standards imposed not by the motion-picture industry but by the computer/electronics industry. The nature of the free market is such that, if a 24x36 sensor were economically feasible, someone would have introduced it. In the meantime, we're using smaller sensors because they can be economically made with consistent quality, much like motion-picture film 80 years ago.

We here at RFF are, by definition, throwbacks. Most of use are using stuff that was fantasitcally high-priced cutting-edge gear when new, but is now affordable because technology has moved on. A few forum members adopted Leicas because of their reputation for excellence. But most of us came to the rangefinder world through hand-me-downs, cheap used equipment or a stubborn Soviet refusal to stop making 1930s Industrial-Age cameras. Even my Nikon rangefinders -- considered exotic and pricy -- were and are considerably cheaper today on the used market than they were when introduced in the 1950s where, adjusted for inflaton, each body cost over $2,000 and each lens was at least $1,000. In fact, one of the reasons I bought them in the late 1980s was because they produced excellent photojournalism results for a lot less money than equivalent Leicas. A big portion of us -- me included -- could only afford the RF equipment because it was considered obsolete. And we constantly trade anecdotes about lenses that were once hugely expensive investments for professionals and serious hobbyists, which can now be bought for the price of a not too snazzy dinner.

That's probably too long-winded a way of saying: The format is changing. We need to buy new wide-angles. We'll get used to it.

 

[rvaubel]

With respect to the Leica M8, the thing that annoys me about not having it be full frame is that I won't have a lens with an approximate angle of view of 35mm that has a maximum aperture of 1.4. Other than that, I'm fine with it.

Henning

I agree, a 28mm f1.4 would be a great lens for the M8 as it would be equivelent to the 35mm f1.4 .... but the cost to cover full frame would be very high. The other lens missing in the line-up is a full frame equivent to a 21mm f2.8. Leica is supposed to introduce such a lens (15mm??) but I tremble at the cost.

Rex

 

[LCT]

What about compression?

What about compression?

What about compression in all this?
Each focal length has its own way of compressing / expanding backgrounds and foregrounds.
Due to the crop factor, a wide angle lens like a 28mm may become a 'normal' one as far as FoV is concerned but it remains a WA as far as DoF and compression.
It may give nice results by expanding foregrounds for instance but also ugly ones like the brick walls in the pic below (R-D1 & 28/2) which would not bend that way with a 40mm lens on a 35mm camera.
Best,
LCT

 

[Bob Ross]

Hi LCT,
I think that you do have the compression factor involved in changing formats. I have been using 4/3rds format for several years and find myself drifting back to my favorite 90mm lenses, that I used with 35mm film. I noticed the same can hold true for 35mm compared to 6 X 7 MF where 90mm is a normal. The image compression gradient significance may have to do with our individual eye sight or mind's eye vision, the way we want the world to stack up. Changing formats might be a time to discover whether you want to "get it all in' or "have what you got stack up the way you imagine it should", when you make the picture.

 

[Mark Norton]

VinceC talks a huge amount of sense. Good job!

 

[James Brannan]

I do take 645 negative, 35 mm negatives and 6Mp digitals, and enlarged to A3 the 645 clearly look best, the digital comes second and the 35 mm last by a fair margin. I won't open the whole argument here, but after all factors are considered a 6Mp image has enough resolution for any print, provided that print is viewed at the correct distance. You would, I am sure, not print a billboard at 300 Dpi....

Which 6mp body? Just curious. Adding to the point, I had a 8mp digicam that was abysmal at A3. I have 3 6mp DSLR bodies that easily do A3 and when I get everything right even make the occasional decent poster size print for me. I have higher resolution bodies but see 6mp as the turning point for consistently outdoing 35mm film.