Photography has been either blessed or cursed in the digital era in terms of sensor size. For one, you’ve got options, and there are certain ‘benefits’ to cropped or full frame sensors, but the behavioral differences between the two are the catalyst for much discussion, debate, and divergent opinions. We all know they’re different, but in what ways and to what degree is usually where the controversy arises. The two most contested talking points when comparing tend to be resolving capability, and then, how DOF and focal length behaviors differ from one size sensor to another. It’s a bit nerdy, but it’s useful.
Understanding the differences can help you in the purchasing process, and in very practical use. For instance, if we accept the fact that compression of a 50mm will be the same on a crop or FX sensored camera, but that the DOF will differ, we can understand what the limitations are, and then choose the right gear for the type of shot we’re going for. Conceptually, it’s not the simplest to grasp or extract from text, but from a visual demonstration, the concept becomes clear and easily understood. The problem is doing the set-up yourself and having all the equipment necessary to conduct such a demonstration isn’t practical for most, but as luck would have it, there’s an online DOF simulator that will help.
Polish Photographer Michael Bemowski has released a bokeh and DOF simulator which can help you to understand the concept of DOF on a whole, and allow you to ‘see’ how an image is affected by various changes in aperture settings, focal length, lens choice, sensor type, and how it all works depending on the distance from the subject and background.
Thoughts
For the record, let’s just be clear that those of you of a DPReview and DxO Mark persuasion will likely find this heinous, and may try to poke holes in it. But it’s not really intended for you if you’ve already got camera theory mapped on the brain. That said, it’s highly customizable and you just may be surprised at one or two things you may find.
[REWIND: Why You Should Multiply Aperture By Crop Factor When Comparing Lenses]
For others, this can really be helpful in a myriad of ways. From the changes that you see visually in the preview image, you may be able to understand how to configure your gear to get certain types of shots, and that may highlight for you what type of gear you may be missing, and what you may need to achieve what you want. It’s also a good way to make sense of the numbers and ‘see’ them rather than ‘read’ them.
You can find the web app here, and if you’re interested in the differences that come with a change of sensor size, the following video is sure to be very illuminating.
Source: Reddit
17 Comments
Leave a Reply
You must be logged in to post a comment.
Fun and helpful!
I think he could give the exact same talk, pretty much, comparing a 4×5 camera, a MF camera, and a 35mm camera. Despite all of the “70 year old standard” and “made for film” protests, nothing he’s discussed here would be any different in a discussion of different film sizes.
He’s also being just as misleading on the aperture stuff. Yes, his correction, going f/2.8 down to f/5.6, is going to give you “depth of field” equivalents, just as rating the 12-35mm lens at 28-70mm is going to give you the 35mm equivalent crop. But that’s it. You’re getting the same light per mm^2, which is all that aperture ever promised anyone. It’s not deception, they’re telling the truth. Again, back in the film days, I don’t think you expect an f/2.8 on a 35mm to collect as much light as an f/2.8 on a Hasselblad — it’s providing just as much light per mm^2, but there’s not as much area to cover.
Oh boy… he’s usually pretty good. But clearly not an Engineer.
He’s wrong about ISO. Ok, digital cameras really don’t have the ISO analog to film cameras. In a film camera, the ISO is based on the ISO 5800:2001 standard. Yup, 2001 is the current revision, all films currently sold are rated based on this standard. The original, derived from ASA and DIN, was published in 1979. It’s not really that old. The ISO standard rates the photosensitivity of an emulsion.
Digital cameras don’t have that kind of photosensitivity… the photodiodes in your camera have a dynamic range somewhere in the 100dB-120dB range, maybe more these days. That sensitivity never changes. However, your analog to digital converter is typically only 12 to 14 bits… 72dB to 84dB. So there’s a gain stage. On a camcorder, it’s called… GAIN. But since that gain setting is different for every camcorder (eg, no correspondence to a specific EV), the International Organization for Standards decided to make an equivalent ISO rating for digital photography. That’s known as ISO 12232:2006, the current version… the original was created in 1998. There are technically five methods available to a manufacturer for scaling ISO in a digital camera. However, by Japanese Law — which applies to many modern cameras — only the two new ones, Recommended Exposure Index (REI) or Standard Output Sensitivity (SOS) may be used currently.
So every sensor, regardless of size, is rated the same way. Just like film. Didn’t matter then, doesn’t matter now.
He’s also way off on the SNR of a sensor. There are four contributing factors. There’s the collection of photons, of course… the more photons available, the more electrons flow from each photodiode, the more charge collected, the better the sensor will be against noise. But in low enough light, you can have enough variation between the photon counts at each sensor to get a grainy photo — that’s photon noise. Next is dark current — every photodiode produces a current in total darkness. Modern cameras measure this before every exposure, so this is accounted for. Then there’s random thermal noise — every analog system has a thermal noise floor based on the random activity of electronics from heat. This accumulates with longer exposures — the reason astrophotographers cool their sensors. And finally, read noise — the noise generated by the sensor in the process of reading. That’s also random, and the main thing that’s seen generational improvements over the years. In short, it’s pretty complex, and all sensors have these same issues. Light sensitivity is entirely per pixel, and based on pixel size, not sensor size. Naturally, a large 50Mpixel sensor with 4um pixels won’t show as much noise at the same print size as a 16Mpixel sensor with 4um pixels, all things being equal. But it’ll have exactly the same per-pixel noise.
More later…
Great way of learning DOF.
nice app to play with it. although if youve been shooting for a few good years, when you come to shoot a subject or a place, you already know what lens and settings to use to get the image you already envisioned in your mind.
but it was fun to play with. and I liked how the chnages to the settings were immediate to see.
bravo Michael Bemowski!
meaning , you have an image in your mind of the final result and know what lens youll need and what aperture/iso/shutter you want to go with. a WA lens for exaggerated perspective, or a tele for compression etc
Of course experienced photographers will know what settings they want. It’s more for the ones starting off. =)
Sent this link to my girlfriend. I’m trying to get her into photography, lol.
but I enjoyed using it nonetheless and appreciate Michaels effort. it will help a lot of people.
Sorry…I don’t buy the ISO bit!
I don’t believe that total amount of light is relevant
If I were to cut my sensor in half….or cover half with tape…
Will the image recorded by the remaining half all of a sudden become noisier?
This is a good learning opportunity:
If you “crop” half of the sensor, then you will have a field of view that is one half the size it was before, despite using the same focal length lens.
In order to return to the same subject framing, you would need to back up to twice the distance to the subject, or you would need to use a lens with 1/2 the focal length.
If you focus, say, a 50mm lens to 10′ away at f/2, the DOF will be roughly the same in any format – any differences in the DOF will be caused by a different circle of confusion. Where the difference comes in is that the 50mm lens will give you entirely different perspectives when using different formats. So on a 35mm sensor it would give you a `40 degree field of view. But if you used a 50mm lens on an m4/3 camera, it would give you a 20 degree field of view with the same depth of field.
So, here’s where the ISO equivalence comes in. The real advantage of larger formats, other than pure image quality capability, lies in their ability to control depth of field. So if you’re comparing an m4/3 camera lens to a 35mm lens, for equivalent results, you are going to be using a focal length twice as long on the 35mm camera for the same subject framing with equivalent lenses.
Because the m4/3 system uses a shorter focal length, it happens to be that you need to double the aperture (As you double the focal length of the lens) compared to a 35mm camera, so you can calculate the depth of field for that.
If you’re using a full frame camera with an 85mm f1.2 lens to take a headshot of someone, to match it on an m4/3 camera, you would need a 42.5mm lens (to match the field of view), and an aperture of f/0.6 to match the depth of field. If such a combo existed, then you would find that the images would be equivalent in depth of field.
So if you want to use less depth of field, a larger sensor size will give you the results you want.
But you don’t always want to use less depth of field, do you? Sometimes you want to use more.
As it happens, m4/3 systems will give you more depth of field. A lens at f/2.8 on an m4/3 camera will give you the same depth of field as a 35mm camera at f/5.6.
So if you’re shooting and need more depth of field, the m4/3 system will do that for you at a larger aperture.
The larger aperture will allow you to use an ISO that is 2 stops lower. So if you’re at F5.6, ISO 3200 on a full frame camera, to get the same image quality, with a micro 4/3 camera, you’d shoot at f/2.8, ISO 800. As it happens, the 35mm camera has a sensor 4x as large, so it collects 4x as many photons across it, but with an aperture of f/5.6 versus f/2.8, the light is 4x less bright.
If you go to a hypothetical sensor that is 1/4 the size of a m4/3 sensor, then to compare it to full frame, you would do the same math again. f/16 on the full frame sensor at ISO 1600 will be reduced by 4 stops. It would be equivalent in image quality to f/4 at ISO 100 on the small sensor camera.
By equivalents, the goal is to say “What would it take to match the larger format in EVERY way?” And you factor that in by considering noise and depth of field. So if you made a print of both of them and looked at them, you wouldn’t be able to guess which camera the image came from because it looks essentially the same.
Get it now?
Very well written Jeff.
Of course not. He seems to be trying to say technical things in a non-technical way, and that makes it a little confusing if you know what’s going on.
So the actual low-level noise is based on the design of the sensor, the amount of light captured per pixel, which is of course based on the size of the pixel, all else being equal. So a FF sensor with 4um pixels is going to have the same levels of noise as a m43 sensor with 4um pixels. Cropping out half of your photo doesn’t change that, and everyone know it.
But what does happen is that, when you magnify the image by 2x, you magnify the visual appearance of that noise by 2x. You also change the effective focal length by 2x. So in thinking of the whole picture, which seems to be what Tony is after here, cropping will absolutely make your image noisier, when brought to same full photo size as your view of the full frame. If there is absolutely no captured noise, that’s not the case, but as long as there’s some, you’re blowing it up.
This is really neat. I’ll be revisiting this many times.
I can remember having some serious, serious bokeh envy with my D7000. I could never get it to look “professional”. I did everything I could do to get that out of focus background. It drove me crazy.
Eventually, I gave up and figured that I would get one of the better full frame cameras. At the same time, I realized that all the photographers and images that I liked, the backgrounds were more in-focus than out of focus. They seemed to be shot around f/5.6 to f/11. The backgrounds were lit, and elements in the background were included as part of the composition.
However, Tony’s video shed a lot of light on why I couldn’t get the same result with my setup. It immediately clicked for me. I’m definitely going to play around with this bokeh simulator.
It really honestly comes down to glass I think, in my experience anyways. The first time I used a 24-70 f2.8 there was no going back. Or the 70-200’s out there.. amazing bokeh.
That’s the thing. I had the glass, but still couldn’t get the look. All my photog friends told me to save on the money on the body, and spend money on the glass. So, a Tamron 70-200 f/2.8, and a 28-75 f/2.8. I would also rent lenses from the local camera shop, and still no dice.
That’s why Tony’s controversial video about applying the crop factor to the focal length and the aperture, to get an idea of what the bokeh will look like, made total sense.