Camera Resolution (Digital and Analog)

I'd argue that...

1. Bit-depth is relevant to resolution because it's a part of "information recorded into the media/medium". In fact, it's not just bit-depth, but how the camera, as well as your computer display, or printer, or printed medium, represents this bit-depth that's an issue. It's not just the gradations of colors, but it's also how much info you can compress into a certain "range" of color. For instance, if you over-expose a shot, most of the information will be stored in higher bits. When you try to pull those bits down, you'll find that not even 10 bits per color channel will be enough to recover information. The same is true of the other extreme... where you'll lose information in the dark because the camera sensor compresses information into lower bits.

2. Dynamic range is also relevant because most current standards are unable to display this high dynamic range. Anything above or below a certain range will simply be discarded. This is especially true in digital because in digital terms, there's no "accepted standard" on how to deal with high bit-depth information. Consider how something like... Adobe Lightroom deals with a RAW file, for instance. It'll simply compress information and makes it look like the RAW file has super high contrast but low saturation. This is the default behavior. You have to reduce highlight and push up shadows, essentially "compress dynamic range" just in order for those info to be visible again in a conventional display. If you took your photo with a JPEG file format only, then it's even worse because there won't be anything left. It'll have discarded the higher and lower bit info in the process of compression because the current standard "thinks" people won't be able to perceive those higher or lower bits on their 5-bit or 6-bit displays.

It's not just a coincidence that there's always this "shoot in RAW" thing among professionals these days.

 

@TRex I think part of that is because they also refer to a separate (equally ugly) document:

http://videopreservation.conservation-us.org/library/film_grain_resolution_and_perception_v24.pdf

(Now we're going blissfully off-piste..)

 

1. Bit-depth & resolution are part of information, but they are different. Resolution determines sensor ability to capture detail, meanwhile bitdepth determines sensor ability to capture color. For example, a sunset in Golden Gate bridge (pick this just because it's gorgeous). High resolution/low bit depth sensor is capable to capture fine details of bridge, seabirds, cars, pedestrians but the transition from water surface (hazy) to the sun (intense orange) isn't smooth (color banding). Low res/high bit-depth sensor isn't capable to capture fine details but the color transition of sunset is smooth.

Over-exposed region isn't stored in higher bit-depth. Over-exposed simply means higher luminance (brightness) value thus it needs high range. I think you're confused between bit-depth and color/dynamic range. Bit-depth represents the number of gradation betwen a range, meanwhile range represents the width of two opposites (how much info can be stored) -see #2.

2. DR is important, but it is different than resolution or bit-depth.

A good resource explaining resolution, bit-depth, DR (since I'm very sucky on explaining things): http://www.peachpit.com/articles/article.aspx?p=1709190&seqNum=2

@Kattefjaes so they counted the number of silver halide particles on films? That's neat, but I don't know a company which is capable to produce films with same, consistent number of silver halide particles. Digitals sensors, in other way, have certain number of pixels.

Edit: RAW vs JPEG. Yes, JPEG is a compressed-and-edited-everything version of RAW, but every JPEG is created based on RAW file, so RAW vs JPEG is about post processing not sensor readout.

 

Well, you'd obviously be sampling and using statistical methods to estimate your variance etc.

 

Nice thread! Some thoughts:

- To my understanding film has a bit depth of 1. A halide crystal either develops into silver, or it doesn't. The resolution is crazy high (for any ISO film) but some of that is traded off via dither for a higher bit depth (smooth tonal gradations). The difficulty in measuring the effective resolution and bit depth is the same difficulty in measuring the effective resolution and bit depth of DSD.

- As mentioned, film's response curve is non-linear so there is lots of dynamic range in the upwards direction. In fact I would say that it is effectively infinite in the upwards direction. Exhibit 1: here is a photo I took on negative film that is approx 16 stops over-exposed. The negative looks completely black, and a really bad solid kind of black. But there is enough information to recover it.



- I scan 35mm film to 24megapixel files, and that seems to be about the correct resolution for 100 ISO negative film. For slide film it seems there is probably more information available. Step up the film size and there is plenty more information available: I scan 6x6 to 81 megapixel files. At this point the film is so resolving that, as already mentioned, other physical factors are the limiting ones: sharpness of the lens (a bit); diffraction effects (a lot); vibration (most of all).

- Direct photographic prints will always look sharper than digital prints because, even from 35mm film, you can reasonably get 1000dpi in a 6x4" print. Digital prints are limited to 300dpi (or 600 if you are lucky). If you contact print medium or large format then the difference in sharpness is just ridiculous, you can see it from across the room.

- I think for 35mm, digital is just about caught up with film in terms of technicalities. But medium and large format analogue just shit over anything digital can do, absolutely destroy it in every way possible. Go look at some large format portraiture, it's amazing.

 

@Garns
- Films have more than 1 bit depth. Not 100% of halide converts to silver (BW films). Chromogenic & color films mechanisms are different though. When you see through transparencies, you'll notice semi-transparent areas -that's the remaining halide.

- Films do have limited range, even in the upper region. Your photo has nice shadow detail, but the sky and even part of grass are overblown (no detail, no information so just the color of the paper). 16-stop overexposure (not range) is jut too much for any film.

 

Ah good, thanks! Didn't know that.

Yes, I agree. But to reach this point one would have (as I did) to shoot with a fast film, on a sunny day, with wide-open aperture, on a tripod, for an extended exposure. So by "effectively infinite", I really mean, "infinite, if you are not deliberately trying to do something idiotic".

 

NB I also suspect the dynamic range of my scanner could be more of a limiting factor than the film itself. Scanning a totally black negative is pretty hard.

 

The shittiest thing about the SDR to HDR conversion is that all of our gear (cameras, displays, video cards,.....) already supports it. Not all will support 10-bit accuracy, but the backlighting required to be bright enough for HDR is pretty much here already (I can only use 15/100 brightness on my cheap $100 monitor before it burns my eyes out), and we always have high quality dithering and debanding to make up for 8-bit panels. Feels kinda weird knowing that some have been spending thousands and thousands on cameras only not to be able to see what it could fully do because standards hadn't caught up with technology.

EDIT: And then there's the UHD Blu ray DRM,HDMI 2.0b, DisplayPort 1.4, VGA 6.9z and all that bullshit......Schiit was right about catering to whatever's most widely used (PCM 44.1khz/16bit). The $400 PS4 Pro is meant to be connected to HDR TVs that cost $2k minimum and can't play UHD blu ray. Necessary upgrades are being locked out from consumers. I just need a monitor below $500 that supports HDR to appear. Is that so hard? Thank god for companies like Schiit working to drive a consumer-friendly market.

 

@TRex if you have used your camera in extreme conditions, you'd know... that even the best current digital camera on the market (say, for full frame 35mm format, that's the Sony A7Rii) cannot recover a shot that's over-exposed by 16 stops. The most you can recover is about 5-6 stops, and the end-result will still look messier than what @Garns has shown. It is typically around 1/2-stop for phone and compact cameras, 1 - 1+1/2 stop for micro4/3 or APS-C, and 2-3 stops for full frame 35mm.

Medium format digital camera (that's basically cameras that have a 1.5 to 1.6 crop factor compared to 645 film) on the market at best can do 9-10 stops, and that's already pushing it.

For digital, you need a sensor that's basically 4 times bigger than full frame 35mm, and with the same sensitivity, in order to get just barely under what was shown as possible. Considering that technology still hasn't progressed enough in the last decade that we can push for that, it is safe to say that even now, film still has a dynamic range edge over digital sensors, simply because, as said, they have a non-linear curve. And in fact, nature itself has a non-linear curve when it comes to light.

For more info, refer to the ultraviolet catastrophe experiment and results to understand why natural light has a curve.

But digital sensors have a linear curve, mostly because... it's faster to process in real time. And this is what makes them unnatural. Some companies try to "correct" for this using post-processing techniques in-camera (Leica and Fuji are infamous for pulling this off), but most of the time, that's why we shoot in RAW and then try to correct the curves of inidividual color channels in Lightroom or Photoshop or Capture One or something similar.

@Garns, I think one thing to take note is that for 35mm films, the reason why 24MP seems enough is because most lenses for film cameras resolve only up to about 20MP or so of sharpness. But dynamic range will be more of an issue than MP count. This is why some buy new digital camera bodies as they come out. It's not for the extra sharpness, but for the extra dynamic range. I wish Sony would make the A7Rii sensor tech more... mainstream and trickle it down already. An A7iii with more compact body, with none of the blink, but with a 24MP sensor that is back-side illuminated would be an instant-purchase for me.

Otherwise, if you're using a lens that resolves more than 20MP of sharpness, then it's best to increase scan resolution. Modern Zeiss lenses (when they are stopped down) are the first things that come to mind.

 

What we need right now are cheaper consumer HDR displays/monitors. Dynamic range is good for correcting exposure of course but only because we're seeing 8 on our displays out of the 14 stops that current cameras with good sensors can achieve at low ISO. 2-3 stops of recovery is already good enough for most proficient photographers, and besides, overexposure/underexposure is not really a thing with mirrorless cameras anyway. I want to actually experience everything that the sensor can capture, everything piece of information for which I paid a ton of money and time to be able to capture. Even current HDR is only 10 bits but I'll take what I can get

I'm not very knowledgable about Lightroom so I don't know how it converts a 14 bit RAW to an 8 bit bitmap for normal use (JPEGs are 8 bit) or to display it on our outdated screens (most are 6-8 bit). Does it compress the dynamic range down or simply cut the extremes off to fit 8 bit?

 

I agree that 35mm should be able to offer more than 24MP. I am limited to that by my scanner (Nikon 9000) and the only upgrade path seems to be an Imacon or a drum scanner, both of which are completely unreasonable for a hobbyist. But for negative film at ISO 200 and up the Nikon can already resolve the grain, so I feel like for such films it's a reasonable approximation to the available resolution. Something like Velvia 50 is another question. I've never seen any grain in scanning that. So with super sharp lenses and an excellent tripod I'm sure you could get plenty more than 24MP out of it, and I'd be interested to know how much more exactly.

All of this is why I like medium format - no need for extraordinarily sharp lenses or exotic scanners to get way more resolution and way better tonality...

 

http://arstechnica.com/gadgets/2016/10/nikon-d5500-peltier-cooler-astronomy-less-noise/

Cool device.

 

Most of the discussion I have seen on the web seems to revolve around film being scanned in order to compare it to digital. I am curious if there is any scientific way to compare an analogue print to a digital print? I know you can digitize analogue pictures to a reasonable level with a consumer scanner, but it seems to me like making a high quality print would be easier and cheaper with film than digital.

 

Isn't that a bit like auditioning speakers on YouTube? Isn't what you get the result of the scanner's capability and resolution?

 

Bleh semantics. Experience matters more. Film looks better and always will. I could give a shit what words are used.

do any of you have experience shooting and analog printing medium and large format film?

 

Yes, up to 4x5 enlarged and 8x10 contact print. All B&W.

On my best day in the darkroom, I could produce prints that are 1/2 as nice to look at as my worst day in front of the computer with files from my X100S.

A great analog print is a thing of beauty. So is a great digital print.

 

You really think that? You must have been a shitty darkroom printer

 

I mainly do landscape photography and pretty amateur when it comes to film. From a completely non-technical standpoint, I've found film to render highlights much better and have smoother gradations of tones than digital. However, shadow details I think digital can handle a lot better.

Large and medium format portraits are something else that even the nicest digital hassys can't quite match. I really don't know what it is specifically but they just have 'a look'

All that said, I still lust for the Leica M10 monochrom

 

Standing and watching

Used to take original artworks to the studio where 10*8 colour transparencies would be made, from which we would get colour separation and CYMK positives would be made for printing.

Although I was only involved with greeting cards, the company also made fine prints, and the colour accuracy of the transparency was important. The colour aspects (even if they got messed up later in the separation and printing) had to be predictable). Well, professionals will control colour through the digital process today, but the vast majority of us won't. But then... I never took a colour-critical eye to my holiday prints.

The studio cameras were the most mechanically simple I have ever seen. Possibly also the most expensive! I'll never forget that the "shutter" was the lens cap, and the correct exposure was down to metering and the correct timing of the photographer.

There is zero point in the digital/film argument. It's worse than digital v analogue in audio! On second thoughts: there is certainly a point in the discussion: it's been interesting to review this thread.
Some people will always enjoy what they can do in one more than what they can do in the other. Or prefer the process. Or just opt for the convenience. Or, even, have the ability to replicate the film look in digital.

What I Miss... Having 36 exposures and that's it. As a youngster, that might have been it for a whole week of holiday. It's a matter of discipline. Now I go to a concert and take a hundred photos; which wouldn't be too bad if 60 of them missed focus, and only ten of the rest were actually good. Modern tech means that if I point the camera in the right direction at the right time, with no mistakes in basic settings, I will get at least a half-decent pic. Sorting wheat from chaff was easy: sorting decent wheat from half-decent wheat is hard work.

I do not lust for film. I loved my Olympus OM1n, but it has been in the cupboard for far longer than has my record deck.