Home of Professional Cinematography since 1996

class="style8">Film_Grain vs Video_Noise

>Published : 4th February 2005

>Regarding recent posts discussing film "grain" :

class="style9">>"I understand the desire to make the 2 sources look as much alike as possible, ie to remove obvious clues like grain."<

>Film "Grain" is essentially the same artifact as video "noise". They look the same, (or very similar), and are there for the same reasons.

>Lew Comenetz
Video Engineer
USA


class="style9">>Film "Grain" is essentially the same artifact as video "noise". They look >the same, (or very similar), and are there for the same reasons.

>Lew, only a video person would write that!

>We film people think grain is a thing of beauty, but video noise is
hideous.

>Jeff "grain = good, noise = bad" Kreines


>Lew Comenetz wrote :

class="style9">>Film "Grain" is essentially the same artifact as video "noise". They look >the same, (or very similar), and are there for the same reasons.

>No, that's not correct.

>In the land of film, the grain _is_ the image. Get a strip of neg under a micro densitometer and you'll see what I mean. See Peter Swinson's pages on the tig.

>Simon Burley
RPS Film Imaging Ltd


class="style9">>In the land of film, the grain _is_ the image.

>It is the correlated grain "that is the image". The random grain, which is uncorrelated with the image detail, is the part of the image which various of us praise or complain about and call "grain".

>Regards

>John lowry
Lowry Digital Images
Burbank CA


>To those of you who poo pooed my idea that film grain is the same as video noise :

>1) [Reply to the comment]"That comment could only come from a video engineer". Guess what? I was originally a still photographer! I preferred large format cameras: 4x5 inch sheet film--To minimize grain! Ansel Adams (and other large format advocates) shot large format in part to avoid the excessive grain of the smaller formats.

>2) When you "Push" (force develop) film, what do you get? More grain!! Same as pushing video, where you get more noise! And past a certain point? Its not pretty!

>3) I once was present at a test shoot of three different TV cameras. Most of the observers were film people. They liked the noisiest of the TV cameras "Because of the Beautiful Grain".

>I stand by my original premise: Film "grain" is the chemical equivalent of video "noise".

>Lew Comenetz
Video Engineer - Former large format photographer


>Lew Comenetz writes:

class="style9">>I stand by my original premise: Film "grain" is the chemical equivalent >of video "noise".

>I would say it's analogous, which it is, but not equivalent.

>Brian Heller
IA 600 DP


class="style10">>I stand by my original premise: Film "grain" is the chemical equivalent >of video "noise".

>With all due respect, I daresay that film grain "can look desirable" under the right circumstances, under the right conditions, and when the photographer is attempting to convey a certain "feel" to a scene.

>In my twenty plus years as a videographer I have yet to see an instance where video "noise" looked as "desirable" as film grain.

>It's like trying to compare a wood tone in a fine wine to "that subtle hint of aluminium" in a can of soda.

>Jeffery Haas
freelance editor, camera operator
Dallas, Texas

 

John Lowry writes:


class="style10">>It is the correlated grain "that is the image".

>I think that was Lew's point. By definition, signal is orderly and noise is random. It's the same with any recording or transmission medium, audio or visual.

>Even with drawing and painting you're always grappling with the native textures of your materials, and hopefully achieving a mix of qualities that combines the intentional and the unavoidable in a way that's pleasing, purposeful and unique.

>And to address Dale's point: In a simple analog system, removing detail does tend to remove grain and vice versa. But digital processing is capable of distinguishing various characteristics of an image and selectively adding or removing them.

>One might say analog systems follow the rules of the natural world and digital systems the rules of the mind. Whatever the mind can conceive of that can be expressed in mathematical terms can be realized in the digital domain. The only systemic limitations reside in the transducers (film/chemistry, chips, microphones, recording heads, monitors, projectors, etc. and also the related optics), which still have to grapple with the facts and physics of the natural world.

>Dan "there'll be a quiz in the morning" Drasin
Producer/DP
Marin County, CA


class="style10">>In my twenty plus years as a videographer I have yet to see an instance >where video "noise" looked as "desirable" as film grain.

>I'll agree with that. Every time I've seen noise added to a video signal in order to make it look more "film like" it just ends up looking like noisy video.

>Film grain just plain looks different.

Art Adams, DP [film|hidef|video]
San Francisco Bay Area - "Silicon Valley"
http://www.artadams.net/


>I wasn't trying to get into an argument - I find - with the F900 - 3db is pretty tolerable and 6db shows grain, but from the transfer I've seen (I'm using the BandPro transfer from a few years back as a reference) - I could see it, but I didn't think it looked better or worse than film grain.

>How does film grain and video grain differ? I can't see it, and I'm not sure I want to find out (sometimes ignorance is bliss), but - what the hell.

>John Lowry - what software programs add "smart" noise/grain?

class="style10">>I think that was Lew's point. By definition, signal is orderly and noise is >random.

class="style10">>And to address Dale's point: In a simple analog system, removing >detail does tend to remove grain and vice versa.

>Hey - enlighten me! That's what CML is all about!

>Dale Launer
Writer/Filmmaker
Santa Monica


>Art Adams wrote:

class="style10">>Every time I've seen noise added to a video signal in order to make it >look more "film like" it just ends up looking like noisy video. Film grain >just plain looks different.

>Think of each film frame as a sand painting, and each film as a sand painting flip book. The grains that form the picture are totally random and change position every frame.

>Video noise while random isn't as random because the pixels that produce the picture are in a fixed matrix.

>Maybe when they are making amorphous CCD’s it will get closer to film

>Mark Smith


class="style10">>Someone wrote]: "In my 20+ years [filming] I have yet to see an instance >where video "noise" looked as "desirable" as film grain."

>You guys miss the point: Who says film grain is "desirable"? Why do people shoot 4x5 and even 8x10 sheet film? Why do people shoot 70mm film instead of 35mm? Why do many avoid 16mm film?? (except when a "gritty" look is wanted?).

>TO AVOID THAT UGLY GRAIN!!!!

>I still own a 4x5 camera for that very reason.

>Lew Comenetz
Video Engineer(Ex photographer who still has his 4x5)


>I think of noise as "shimmering" more than grain, which "swirls" more.

>Grain seems more dimensional in that it varies in size.

>Not all grain is the same either -- for example, I find the graininess from using a silver retention process more attractive than the standard graininess of color negative, because of the difference between grains of silver versus clumps of dye clouds.

>David Mullen, ASC


>Art Adams wrote:

class="style10">> Film grain just plain looks different

>More and more, what we see on screen consists of effects which are composites of live action film and computer generated images. The basic procedure is to first de-grain the film layers, and then composite the film layers with the computer generated layers _and_ a layer of film grain. The grain usually consists of scanned images of actual film stock.

>This, of course, is usually done to try and make seamless cuts between the live action and effects footage. Whether it works well or not usually depends (as with most everything) upon the talent of those doing it.

>But the point is that "film grain" is an effect. It can be used with
raw computer generated images or with "video". Video noise is also an effect that can be added to video or film - if that's the effect the director wants.

>Noel Sterrett
Admit One Pictures
www.admitone.com


>David Mullen wrote:

class="style10">>I find the graininess from using a silver retention process more >attractive than the standard graininess of color negative

>And IMHO, B&W grain is prettier than color grain, and color reversal grain is prettier than color negative grain.

>Jeff Kreines


class="style10">>It's like trying to compare a wood tone in a fine wine to "that subtle hint >of aluminium" in a can of soda.

>This is the funniest analogy I have seen in a while. Made me laugh out loud.

>Thanks Jeffery.

>Steve Schklair
3ality Systems
Cobalt Entertainment


class="style10">>I stand by my original premise: Film "grain" is the chemical equivalent >of video "noise".

>Still can't accept this.

>Video noise is extraneous information superimposed upon the image. We think first of the noise due to amplification, where ground noise is amplified along with a weak signal: but from a mathematical point of view, the regular pixel structure can also be interpreted as noise. Both sorts of noise can only take away from the information present in the original image.

>Film grain (or dye clouds) is the basic "atomic" pattern of the image, so it's the equivalent of pixels, not of noise.

>True, it also introduces additional, extraneous data by the structure of the grains and it follows some of the same behaviour (amplify it or push it and you get more).

>But what we notice most about film noise is its random pattern changing from frame to frame. While this, too, introduces extraneous information, it's also true that the randomness of the grain structure actually allows MORE information to be recovered from the image than if it were regular.

>Different sampling patterns from frame to frame in a static subject mean that finer detail can be observed by averaging a number of frames, provided the sampling method is sufficiently fine-resolution itself. And that it is sufficiently finely-tuned not to destroy information at the same time. Of course this argument only applies to motion picture imaging.

>In short, video noise is destructive, while film noise is constructive. Hardly equivalents.

>Dominic Case
Atlab Australia


>Would it be more correct to say that the electronic equivalent, strictly speaking, of film grain, is the line structure, and or the pixel structure of the electronic image?

>Still, isn't the successive build up of grain through generations, similar in the way video noise builds up in generations? So at the first level, grain isn't "noise", it's the sampling structure, but at successive generations the cumulative effects of the random grain pattern is for practical purposes, "noise'?

>Steven Bradford
Collins College
Phoenix Arizona


>A lot of really good points have been made about grain. What stands out to me is :

>1/. Any discussion of film "grain" vs. video "noise" a love-hate thing - some love it, some hate it for reasons ranging from the most visceral creative reasons to the most objectively based science of structural (pixel/silver particle) observation. No matter how we attempt to define the distinction between grain and noise, in the end it seems to come down to whether we like one or the other.

>2/. The contribution of signal and noise varies with the nature of the capture medium, i.e. film vs. videotape or solid state, but at the end of the day there is, with current technology, an inevitable contribution of grain or noise to a captured image.

>3/. If, and I trust the sources, if it is a cinematographers goal to "normalize" all images, then it follows that it should be normal practice to initially de-grain film in the transfer/color correction DI process. With that as a logical given, in that circumstance, then it would seem natural, with that initial goal in mind, to "normalize" all electronic capture HD images by removing all noise from those images at the post-production DI ingest stage, since, from John Lowry's very trusted experience, that is possible in both mediums. In either case an appropriate grain structure can be layered back in to the final product to suit the aesthetic/visceral sensibilities of those who like that effect.

>4/. So, it seems unavoidably obvious that the nature and amplitude of the inherent initial grain/noise structure of film and electronic capture media is a moot point, IF DI is the chosen production/post production work flow process, given the current processing technology, available to all here, and by logical application, that Allen Daviau's choice of degraining was not only logical but extremely appropriate to the DI style film finish work flow, and, particularly, to the Genesis/Film Comparison process.

>IF grain/noise reduction achieves the same imperceptible levels in the initial DI of both the film and the electronically captured elements.

>George C. Palmer
HDPIX, INC.
www.hdpix.com