Size of Film Grain?

[thegman]

I like this article as a simple comparison:

http://www.twinlenslife.com/2011/01/digital-vs-film-canon-5d-mark-ii-vs.html

The Ektar seem to outresolve the 5D Mk. II on the "North Face" badge. This 35mm film, so obviously 6x6 is going to be very dramatically better.

 

[andrew.roos]

Zeiss published an article titled "Resolving power of photographic films" in Camera Lens News volume 19 (March 2003).

It gives the measured resolving power in line pairs per mm of 13 films. Resolving power is more relevant to the intent behind the original question, since grains are not pixels and grain size does not determine image resolution.

The Zeiss article states that T-Max 100 resolves 180 lp/mm. For a digicam to resolve a line pair requires two pixels, one for the dark line and one for the light line. So 180 lp/mm is equivalent to 360 pixels per mm. On a 56 x 56 mm (medium format 6x6) negative, this would be about 406 MP.

However there is a caveat that with a digicam that uses Bayer filtration the image resolution is only at best about half the sensor resolution, in megapixels. This is because the sensor does not capture separate R, G and B values at each sensor pixel, but rather interpolates values from adjacent pixels in the demosaicing algorithm. So to a digicam with 406 MP image resolution would require at least 812 MP sensor resolution. Of course T-Max is B&W; for a fairer comparison, Velvia gives 160 lp/mm in colour which on a 56 x 56 mm negative is equivalent to a 640 MP Bayer sensor.

Of course in reality this figure will not be reached due to other limitations on the resolution of the system including subject contrast, lens resolution, diffraction, etc.

 

[stavrosk]

I like this article as a simple comparison:

http://www.twinlenslife.com/2011/01/digital-vs-film-canon-5d-mark-ii-vs.html

The Ektar seem to outresolve the 5D Mk.

I will say it again.
Seriously when looking at this comparison, is it resolution that wins your attention?
Am I the only one who couldnt care less about that compared to the blown highlights?

Do you all print posters for covering buildings?

 

[pekelnik]

Shooting regularily 5dm2 and 1v with the same glass, it's rather obvious that resolution is absolutely not the reason to shoot film (of the same size). MF film is more comparable to FF digital, but still, there are many reasons to shoot film, resolution is just not one of them. (Even compared to large format, I can just stitch 50 20mpixel photos into a 500 mpixel one and print wall sized with no effort.)

When asked why I shoot film I say that I like the tones which are hard (for me) to reproduce in digital, the grain, smoothness of rendering, larger dynamic range, the whole process which makes every photograph more valuable in my eyes. There is no reason to shoot film if it doesn't motivate you and move you forward in my opinion. It is certainly not cheaper.

 

[Athiril]

andrew roos:

The article you are referring to is http://www.zeiss.com/C12567A8003B8B6F/EmbedTitelIntern/CLN_19_en/$File/CLN19_en.pdf

That isn't quite right. Their test target will be high contrast, it is testing high contrast resolution. Not low contrast (where the high spatial frequency detail is that you actually want to see) such as texture of skin or tree bark, foliage in a distant landscape etc. Not how sharp your silhouette tree outline edges are against a bright white sky (still no high spatial frequencies there).

They didn't reveal the test target.

"in future
CLN issues, we will also publish more
detailed information on the test target
used"

If you want to go through future issues and find out what they used that'd be good.

Also take any dSLR image with a sharp lens, reduce it to 50% then blow it back up and compare to the original, you will see significant detail loss. Regardless, a line chart here is what can help determine maximum sensor resolution.


The other fact is, that getting this level of information off your film is going to be tricky unless you are a great printer with a great lens, most people will be scanning, so will be limited to 25-30 lp/mm (of real detail) from flatbeds, 80 lp/mm maximum at high contrast from coolscans, 125 lp/mm maximum at high contrast (Hasselblad X5) from 35mm only, MF is lower than a coolscan, about 60 lp/mm or less, plus they are disgustingly bad for colour neg from last time I had something done on one.

Printing colour at such great detail.. not many people traditionally print colour, even less compared to the film users doing b&w printing.

How are you going to get that detail off your Velvia? Ilfochrome is prohibitively expensive.

I mean for the super large print you need done for 35mm off a pristine slide (or neg) there is the drum scanning option at a few hundred for the one scan, though if you're printing that size, it's probably for a show or sale etc so has some justification.



But the point is, it's one thing to say this resolves X, but I think that's not much of a comparable figure to digital at all. (which is what the OP is after) as that detail is staying on the film strip.

Eg; lp/mm on the neg only compared to lp/mm from digital on the computer.. not comparable. Lp/mm from a print or lp/mm from a scan, comparable.

Shooting regularily 5dm2 and 1v with the same glass, it's rather obvious that resolution is absolutely not the reason to shoot film (of the same size). MF film is more comparable to FF digital, but still, there are many reasons to shoot film, resolution is just not one of them. (Even compared to large format, I can just stitch 50 20mpixel photos into a 500 mpixel one and print wall sized with no effort.)

When asked why I shoot film I say that I like the tones which are hard (for me) to reproduce in digital, the grain, smoothness of rendering, larger dynamic range, the whole process which makes every photograph more valuable in my eyes. There is no reason to shoot film if it doesn't motivate you and move you forward in my opinion. It is certainly not cheaper.

The detail is on the film, and can exceed your 5DII by some way, depending on the variables. But it's your workflow losing all the detail, and not the film "disappointing" you etc.

 

[barzune]

I know that I've seen it, either in another APUG discussion or in one of my books, that a "grain" in the emulsion coating is about 1 micron diameter. What affects the results is a process of clumping, brought about by the chemistry of development. While each "grain" is, in fact, analogous to the receptors in a sensor matrix, ti's the conversion of the digital signals, along with consideration of the Bayer mix, that I can't hope to understand.

 

[paul ron]

You read it here....

http://cool.conservation-us.org/coo...itale/2007-04-vitale-filmgrain_resolution.pdf

 

[stavrosk]

it's rather obvious that resolution is absolutely not the reason to shoot film (of the same size). [...]
there are many reasons to shoot film, resolution is just not one of them. [...]
I like the tones which are hard (for me) to reproduce in digital, the grain, smoothness of rendering, larger dynamic range, the whole process which makes every photograph more valuable in my eyes.

Totaly agree with this.

 

[andrew.roos]

andrew roos:

That isn't quite right. Their test target will be high contrast...

The other fact is, that getting this level of information off your film is going to be tricky unless you are a great printer with a great lens, most people will be scanning...

Hi Athiril

I did mention that other system limits, including contrast and diffraction, would prevent the theoetical maximum from being achieved. Lens resolution (both taking and printing) is another obvious system limitation.

I wasn't including output since I did not see this as part of the original question, which was phrased in terms of film grain versus sensor pixel pitch. If one is going to force a digital capture (scanner) on the analog process, then of course the limitations of the digital capture process may dominate system performance.

Personally, I shoot film when my primary purpose is to create a traditional print. If my primary purpose is to create a digital file then I shoot digital. But I agree with others who say that resolution is not the reason to shoot analog, even though I'm pretty sure my 645 film camera gives me better resolution than my (more expensive) DSLR. Both media are quite capable of giving me the resolution I need for the sizes I print. I shoot analog primarily because I enjoy it.

Andrew

 

[removed account4]

people suggest all sorts of stuff ...

better thani always subjective,and you will always find the prsonwho says ;that depends...

so ... exactly what ralph said --- that depends.

 

[summicron1]

i tell people that film and digital are too different to compare and that I use film because I enjoy the process and the fact that, being mature technology, it won't change or go obsolete on me in two years. Plus, it would cost numerous thousands of dollars to replace my amazingly wonderful film cameras with even moderately good digital cameras that would, again, be obsolete in 5 years while a Leica is forever.

You really need to learn the artist's shrug -- when asked a question that has no real good answer, just shrug, smile, and take a picture.

 

[Existing Light]

I had no idea that this thread would get so much attention. And I also had no idea how complicated film grain was. It seems like I underthought this whole thing before I posted it.

I think what I was getting at is "Where can I find technical data about how big developed grain for a specific film is?" and then compare that to the size of a pixel for a specific camera. I assumed Kodak and Ilford would have published that somewhere, but I might have been wrong about that. I didnt understand how complicated the emulsion was until skimming a few pdf's I found on google and what's been posted here. I never really thought about it too much.

I think from now on, I'll stick with "I like film better because I can do this with it [shows a few prints]"

 

[ic-racer]

Since "Digital Photography" is the end product of marketing and commoditization, digital sensor resolution is not usually specified. "Meapixel" is, of course not a unit of resolution. So, making a valid resolution comparison is likely to be meaningless to the avid digital consumer.

 

[L Gebhardt]

Seems the film grain thing was a bit more complex than I realized. I found this pdf that explains film particle and grain size. This might have to do until I can find technical publications of a particular film

http://cool.conservation-us.org/coo...e/2009-10-vitale-filmgrain_resolution_v24.pdf

That's the link I was going to send you to show it's more complicated than it appears on the surface. Also, keep in mind it's coming from a scanning comparison so mainly focused on digitizing film.

If you make prints in the darkroom, you really need to compare digital prints to analog prints. But it's hard to measure the results, and distill down to a number. So, the best way to demonstrate the difference is to show a well made print from an inkjet, and from your enlarger. Since it's subjective, there really isn't a right answer.

 

[Photo Engineer]

You read it here....

http://cool.conservation-us.org/coo...itale/2007-04-vitale-filmgrain_resolution.pdf

Blah, blah, blah is right.

Film grain varies from about 1 micron to about 10 microns in size depending on speed. You can compress this scale by proper Iodide addition and chemical addition so that a 5 micron grain can be made to act like a 10 micron grain. This is called a speed / grain enhancement. Two electron sensitization is an example of this technique. Anyhow, slow print films range in the region of 0.2 microns and smaller so ECP is in that range.

Sensors come in at about 5 - 10 microns per pixel, but 3 adjacent pixels are needed for full color so a true pixel is more like 15 x 15 or 30 x 30 microns when you view it in full color and with aliasing removed.

Resolution in film is a measure of how DEEP the layers of grains are stacked and how much absorbing dye can be introduced to prevent internal reflections of the light. In other words, scatter caused by turbidity is your problem in limiting resolution. Light hits a grain and causes an image to form. It must therefore arrive at the correct point to make a sharp image, otherwise you get a blurry unsharp image. This is not a problem with digital. The size of the pixel and the aliasing are.

As for the article quoted above, it is "fair" but does not explain things fully IMHO. Of course, this may just be the targeted audience.

This is the umpteenth thread on this subject on APUG in recent years. You may want to look them up.

PE

 

[michaelbsc]

....Resolution in film is a measure of how DEEP the layers of grains are stacked and how much absorbing dye can be introduced to prevent internal reflections of the light. In other words, scatter caused by turbidity is your problem in limiting resolution.

Ah-ha. Light dawns on Marble Head.
This is why the flat T grains are so much finer detail? Do I understand this now? Or at least begin to do so?

 

[Photo Engineer]

Well, yes. AgX is more or less transparent and is then sensitized by coloring it with a sensitizing dye that is the opposite color of the light it is to absorb. Thus red light goes through a green sensitive t-grain pretty much. But remember that a pan sensitized t-grain is gray or black and therefore absorbs light.

So the answer is different depending on how the emulsion was dyed and whether it is B&W or Color.

PE

 

[Alan Johnson]

One could get a rough estimate of grain size from Erwin Puts' results:
http://www.imx.nl/photo/Film/page169/page169.html
The microfilm grain is a lot smaller than the pixels.

 

[Maris]

I always though that the appearance of grain in positive photographs is a map of the spaces between the grains of the negative. Ok, smaller negative grains make for smaller spaces but the spacing is also influenced by the number of negative grains per unit area. A dense part of a negative has fewer widely spaced "gaps between the grains" so the ability of dense parts of the negative to carry detail and gradation information is reduced.

Trying to come up with a single number to compare negative grains or more realistically the spaces between them with pixels may be a lost cause.

 

[Diapositivo]

You are not the first to ask yourself a similar question. The "visually equivalent" resolution between film and digital is a very "hot" matter of study.

Some readings which I find interesting:

Vitale:
http://cool.conservation-us.org/coo...itale/2007-04-vitale-filmgrain_resolution.pdf
(very interesting)

Koren (which ends up with a very small number of equivalent pixels)
http://www.normankoren.com/Tutorials/MTF7.html

Rockwell
http://www.kenrockwell.com/tech/filmdig.htm
(an interesting read)

My personal experience is this: when comparing my film scans (Nikon Super Coolscan 5000 ED used at its maximum 4000 ppi resolution, 16 passes each slide, 2 exposures each slide, minimal ICE scratch removal) of E-6 slide films with my 10.8 mp digital camera (Sony DSC-R1) I can see a clear advantage of the film scans compared to the digital captures, as far as resolution is concerned. Digital captures appear "cleaner" but they do have less details.

With my setup the equivalence would probably be around 20-30 mp. If comparing with a better scanning procedure ("wet" drum scanning) the equivalence would very probably be at a higher pixel count but costs would rise quite a lot.

To me digital has its advantage in being more versatile (not just faster workflow, but also the possibility to change ISO setting on the fly, which can be quite precious) and film has its advantage especially in the superior dynamic range. Slide scans behave much better than digital IMO under this respect. It goes without saying that negative colour film is even better and by a long shot.

I will not dwell into different pros and cons between digital and film to avoid being boring and off-topic.

Fabrizio

 

[Photo Engineer]

Grain, much as sharpness, is a measure of differences in density!

PE