Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: How does one compare grain with pixel size?  (Read 2811 times)
whawn
Jr. Member
**
Offline Offline

Posts: 71



WWW
« on: February 04, 2009, 02:52:04 PM »
ReplyReply

Was checking some specs on film and realized that film grain is not measured in microns.  Instead, a stat called 'rms' is used for slide film.  Kodak's Ektachrome 100G is said to have 8rms grain, while E100VS is listed at 11rms.  I gathered that 'rms' is determined by viewing a saturated color through a 48 micron aperature and counting the dots somehow.

To top that, Kodak has abandoned rms for its new Ektar negative film and instead is using something called 'print grain index' which is print size dependent -- not measured on the film, at all.  

How does this, or can this, compare with a micron measurement, and why can't we simply state the grain size in microns?

So you know, I was hoping to compare in-camera pixel size with film graininess.

Logged

Walter Hawn -- Casper, Wyoming
Guillermo Luijk
Sr. Member
****
Online Online

Posts: 1280



WWW
« Reply #1 on: February 04, 2009, 11:21:06 PM »
ReplyReply

grain in digital photography has not a particular structure nor measurable size, since it's a consequence of the process of demosaicing individual noisy pixels which locate in the RAW file according to a Bayer pattern distribution.

In its origin, noise has no spatial structure and can be modelled by just a gaussian distribution (possion+gaussian actually, but poisson can be aproximated by a gaussian under a reasonable number of converted photons) independent on each pixel.
The grain (luminance and colour noise) obtained after demosaicing will entirely depend on the demosaicing algorithm and the amount of noise present in the previous individual pixels.

As I see then, digital grain's size is constant since it is related to pixel size. And its shape and how it spreads to surrounding pixels in the form of luminance noise and colour blotches will depend on the development strategy.



BR
« Last Edit: February 04, 2009, 11:24:30 PM by GLuijk » Logged

whawn
Jr. Member
**
Offline Offline

Posts: 71



WWW
« Reply #2 on: February 04, 2009, 11:56:02 PM »
ReplyReply

Quote from: GLuijk
grain in digital photography has not a particular structure nor measurable size, since it's a consequence of the process of demosaicing individual noisy pixels which locate in the RAW file according to a Bayer pattern distribution.
Thanks for the very cool discussion of digital grain.  That's a step toward what I'm looking for.   In addition, I'm understanding that film grain follows a more stochastic, almost a dithered, distribution, that the individual grains are of a certain but somewhat variable size, and that some grains overlap others, unlike digital grain, producing an appearance of 'clumping,' so is there any way to compare the two numerically?  Or are we stuck with just saying, "It looks better or worse to me"?


Logged

Walter Hawn -- Casper, Wyoming
bill t.
Sr. Member
****
Offline Offline

Posts: 2693


WWW
« Reply #3 on: February 05, 2009, 12:23:17 AM »
ReplyReply

It's apples & oranges.  I have reams of old 8x10 prints made from negatives 35mm to 4x5 inch and larger.  Almost all prints at 8x10 and larger size from negatives 2.25" wide and smaller show some amount of visible grain.  Sometimes you don't notice it, sometimes it is easily seen, depends on subject and contrast.  I definitely notice it more now than I did then when grain was just a fact of life.

But I simply have not seen anything equivalent in digital.  Sure, you can try to compare noise and grain but they're just not the same animal.  About all I would care to say about it is that from an aesthetic point of view conspicuous film grain is somehow less objectionable than conspicuous digital noise, but maybe that's just me.  If you want to think of grain as somehow limiting the information in an image, then in digital photography compression artifacts, edge quantization and other issues limit most types of digital images before noise.

It might be more appropriate to compare the modulation transfer functions for film and digital.  But I misplaced my K&E sliderule sometime in the 70's and anyway I'd rather just go take some pictures.  
Logged
teddillard
Sr. Member
****
Offline Offline

Posts: 664


WWW
« Reply #4 on: February 07, 2009, 06:41:02 AM »
ReplyReply

Quote from: whawn
Was checking some specs on film and realized that film grain is not measured in microns.  Instead, a stat called 'rms' is used for slide film.  Kodak's Ektachrome 100G is said to have 8rms grain, while E100VS is listed at 11rms.  I gathered that 'rms' is determined by viewing a saturated color through a 48 micron aperature and counting the dots somehow.

To top that, Kodak has abandoned rms for its new Ektar negative film and instead is using something called 'print grain index' which is print size dependent -- not measured on the film, at all.  

How does this, or can this, compare with a micron measurement, and why can't we simply state the grain size in microns?

So you know, I was hoping to compare in-camera pixel size with film graininess.

wow, ok, this post here, explaining grain measurement is uber-propellerhead:
http://www.answers.com/topic/film-grain

I'm guessing here that it's not particularly useful to measure grain size in microns because it varies so much within a sample, and the problems with the dye layers and stuff you mention.  The rms method looks like it's reading density for an area by measuring the light passing through it, generalizing the sample, sort of like if you were describing soil particle size by measuring the rate that water passed through it.  

This image from that link actually does, however, show a micron range in the notes.  (It's from the early 1900s)

[attachment=11393:Film_Grain.jpg]

As far as your quest, I pretty much agree with the other posts, it's ultimately not going to tell you much, although it is kind of interesting to compare the two completely different processes.
« Last Edit: February 07, 2009, 06:59:08 AM by teddillard » Logged

Ted Dillard
whawn
Jr. Member
**
Offline Offline

Posts: 71



WWW
« Reply #5 on: February 07, 2009, 01:23:43 PM »
ReplyReply

Quote from: teddillard
The rms method looks like it's reading density for an area by measuring the light passing through it, generalizing the sample...
Thanks -- the link told me what I wanted to know.  I'm very familiar with the RMS concept from audio work, so now it becomes much clearer to me.  I shoulda made the leap on my own.  

Just realized a basic disjunction in terminology between audio and video.  In audio, the word 'tone' refers to frequency while in video it refers to amplitude.  No wonder we can't all get along...  

And, now I'm gonna have to buy your book(s).  Is that what you had in mind to begin with?  
Logged

Walter Hawn -- Casper, Wyoming
teddillard
Sr. Member
****
Offline Offline

Posts: 664


WWW
« Reply #6 on: February 08, 2009, 12:45:00 PM »
ReplyReply

Quote from: whawn
Thanks -- the link told me what I wanted to know.  I'm very familiar with the RMS concept from audio work, so now it becomes much clearer to me.  I shoulda made the leap on my own.  

Just realized a basic disjunction in terminology between audio and video.  In audio, the word 'tone' refers to frequency while in video it refers to amplitude.  No wonder we can't all get along...  

And, now I'm gonna have to buy your book(s).  Is that what you had in mind to begin with?  
 

no, actually...  just me geekin' out.  funny though, terminology seems like the core of all of it, whether you're talking color management, digital photography or even audio.  right now i'm finishing up the B/W book, and tone to me means shades of gray.  badda bing!

Logged

Ted Dillard
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad