Can You Provide an Objective Defintion of "Noise" ? ...

[Detail Man]

However, it seems that the results of a study where subjects attempted to hold an imaging device as steadily as they could for 2 Seconds time reveals what is clearly stated to be "mostly random":



Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

Thanks for the reference. I think that as the exposure gets shorter, the amount of randomness in the image caused by camera motion decreases.

Can you provide any concrete bases for that thinking ? The power spectral densities displayed above significantly increase below around 5 Hz. Thus, we might expect that Exposure Times shorter than around 200 mSec might result in lower magnitudes of positional variations. However, such an assumption itself does not speak to the nature of the hand-tremor motions themselves.

From the text of "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik, IEEE Transactions on Consumer Electronics, Vol. 53, No. 4, November 2007 (where "Figure 2" below refers to the pair of time-domain plots appearing at the top of the graphic, and "Figure 3" below refers to the pair of time-domain plots appearing at the bottom of the graphic):

The results are summarized in Figs. 2 and 3. In Fig. 2, we present the mean standard deviation of the motion path as a function of time. The mean is over the 85 motions. The standard deviation is a good estimate for motion blur in most cases. Fitting the results to a power law:

σ[handmotion] (t) = A * t^(a)

yields for x-axis a=0.75 and for y-axis a=0.69. This result is close, yet higher, than that of a random walk. This is expected as the motions are not fully random, and contain some inertial part.

Do you have evidence supporting a hypothesis of "inertial non-randomness" (or something like that) ?

In the limit, since rotation seems to be the main culprit, ...

Quoting from the text cited above:

We first noticed that during one second of motion, the amount of rotation was negligible. The maximal rotation during a 1/96 Second period is 0.12 Degrees, for a full 1 Second it is 0.88 Degrees. In the rest of the analysis we ignore the rotation, as its effect is considerably smaller than that of the translation.

.

... with the subject plane distant, camera motion can be modeled by a direction and an amount of linear displacement during the exposure.

It could be - but with seemingly significant reductions in accuracy as camera-subject distance increases - as is indicated by the text quoted below. Rotational motion manifests on a level independent of camera-to-subject distance, however.



Source: "Image Stabilization Technology Overview", David Sachs, Steven Nasiri, Daniel Goehl

.

Angular to linear displacement calculations can be performed on 1-dimensional (angular) movements - but that is a mathematical fact which exists indepedent from the evident loss of accuracy involved.

.

This appears to be the Ps deblur model, although I haven't played with it much.

I assume that you mean this. item (have yet to see one positive assessment by Adobe PS users).

Also, let us not forget about subject-motions. Ever tried to photograph a flower on a breezy day ?

Got me there, DM. I would characterize the flower's motion as chaotic, not random, but I think I previously agreed to conflate these for the purpose of this discussion.

My experiences attempting such surely feel "chaotic" ! I know nothing about "chaos theories" - but my thinking is that the propogation of such breezes may well be effectively random in nature - though (in a linear model) "filtered" by "system-responses" of relevant nearby physical objects).

.

It seems that one might reconsider assuming camera/subject-motion as being non-random in nature.

DM

 

[Allan Olesen]

In the limit, since rotation seems to be the main culprit, ...

Quoting from the text cited above:

We first noticed that during one second of motion, the amount of rotation was negligible. The maximal rotation during a 1/96 Second period is 0.12 Degrees, for a full 1 Second it is 0.88 Degrees. In the rest of the analysis we ignore the rotation, as its effect is considerably smaller than that of the translation.

.

Perhaps I am misunderstanding the context. Anyway:

If you take a photo of a subject 100 meters away, and you rotate the camera 0.12 degrees around an axis in the sensor plane, you get the same displacement as if you moved the camera 0.2 meters parallel to the sensor plane without rotating it.

0.2 meters in 1/96 second is 18.2 m/s. I am pretty certain that I don't normally move the camera at that speed while taking a photo.

Of course, the closer you get to the subject, the more influence linear camera movement will have, compared to the influence from camera rotation. But you will have to be really close before linear movement becomes the main source of image blur.

 

[JimKasson]

Thanks for the reference. I think that as the exposure gets shorter, the amount of randomness in the image caused by camera motion decreases.

Can you provide any concrete bases for that thinking ?

Just hand-waving arguments. Nothing like your well thought out references. Just take the swoops of the images of LEDs that I showed you earlier. They were exposed at 1/3 second or so. Exposures of 1/300 second would give you arcs of the same shape, but 1/100 as long. Now they look pretty straight. Go to arcs that are 1/1000 as long and they look really straight.

In the limit, since rotation seems to be the main culprit, ...

Quoting from the text cited above:

We first noticed that during one second of motion, the amount of rotation was negligible. The maximal rotation during a 1/96 Second period is 0.12 Degrees, for a full 1 Second it is 0.88 Degrees. In the rest of the analysis we ignore the rotation, as its effect is considerably smaller than that of the translation.

Boy, that's not my experience. Maybe the subjects were close. With subjects at infinity, translation wouldn't do anything to the image.

My experiences attempting such surely feel "chaotic" ! I know nothing about "chaos theories" - but my thinking is that the propogation of such breezes may well be effectively random in nature - though (in a linear model) "filtered" by "system-responses" of relevant nearby physical objects).

It seems that one might reconsider assuming camera/subject-motion as being non-random in nature.

I am willing to do that.

Jim

 

[Detail Man]

EXPERIMENTAL SETUP

We measured hand motion as a function of time, using small
board with a CMOS sensor. The board (shown in Fig. 1) form
factor was chosen to resemble that of a camera-phone. It is
approximately 5cm by 9cm and can be held like a camera-
phone. During each measurement, the sensor recorded VGA
video at 96 fps. The motion was later extracted from the video
sequence.

Each subject was asked to hold the board as steady as
possible for 2 seconds, after pointing the sensor at a standard
resolution chart (using a nearby computer screen). The
illumination conditions were regular indoor conditions
(around 300 Lux).

25 unpaid subjects (20 males and 5 females, ages from
25 to 50 years old) participated, each making 3 different
shots. The total number of different measurements
(motions) was 85.

Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

 

[Detail Man]

In the limit, since rotation seems to be the main culprit, ...

Quoting from the text cited above:

We first noticed that during one second of motion, the amount of rotation was negligible. The maximal rotation during a 1/96 Second period is 0.12 Degrees, for a full 1 Second it is 0.88 Degrees. In the rest of the analysis we ignore the rotation, as its effect is considerably smaller than that of the translation.

.

Perhaps I am misunderstanding the context. Anyway:

If you take a photo of a subject 100 meters away, and you rotate the camera 0.12 degrees around an axis in the sensor plane, you get the same displacement as if you moved the camera 0.2 meters parallel to the sensor plane without rotating it.

I agree.

0.2 meters in 1/96 second is 18.2 m/s. I am pretty certain that I don't normally move the camera at that speed while taking a photo.

It seems that the camera-subject distance used in the study was considerably shorter in distance:

Each subject was asked to hold the board as steady as
possible for 2 seconds, after pointing the sensor at a standard
resolution chart (using a nearby computer screen).

Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

Of course, the closer you get to the subject, the more influence linear camera movement will have, compared to the influence from camera rotation. But you will have to be really close before linear movement becomes the main source of image blur.

As nicely shown in the latter graphic posted.

 

[Detail Man]

Thanks for the reference. I think that as the exposure gets shorter, the amount of randomness in the image caused by camera motion decreases.

Can you provide any concrete bases for that thinking ?

Just hand-waving arguments.

Excellent pun there.

Nothing like your well thought out references. Just take the swoops of the images of LEDs that I showed you earlier. They were exposed at 1/3 second or so. Exposures of 1/300 second would give you arcs of the same shape, but 1/100 as long. Now they look pretty straight. Go to arcs that are 1/1000 as long and they look really straight.

I see.

In the limit, since rotation seems to be the main culprit, ...

Quoting from the text cited above:

We first noticed that during one second of motion, the amount of rotation was negligible. The maximal rotation during a 1/96 Second period is 0.12 Degrees, for a full 1 Second it is 0.88 Degrees. In the rest of the analysis we ignore the rotation, as its effect is considerably smaller than that of the translation.

Boy, that's not my experience. Maybe the subjects were close.

Each subject was asked to hold the board as steady as
possible for 2 seconds, after pointing the sensor at a standard
resolution chart (using a nearby computer screen).

Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

With subjects at infinity, translation wouldn't do anything to the image.

What sort of translation are you talking about (1-dimensional angular-linear conversions, I suppose?). We know, however, that angular rotation is a distinctly different matter. Am not sure how your comment relates to the above quoted text, though [or was it intended to (?)].

My experiences attempting such surely feel "chaotic" ! I know nothing about "chaos theories" - but my thinking is that the propogation of such breezes may well be effectively random in nature - though (in a linear model) "filtered" by "system-responses" of relevant nearby physical objects).

It seems that one might reconsider assuming camera/subject-motion as being non-random in nature.

I am willing to do that.

 

[Detail Man]

Use of the term "stochastic" completely ignores all periodic components. Following that line of thinking, it would appear that any/all components of the output-data (that are unrelated to the input-data) which are periodic in nature do not constitute "noise".

Right, if purely periodic.

One little "beef" (that also came up when conversing with GB on his recent thread). If one follows your above statement, then it would seem that any periodic ("aka "pattern") "noise" that appears within a recorded image is (also, accordingly) not considered by you to be "noise" ? If and when such things appear in recorded images, it seems (for me) hard to refer to them as constituting "desired signal(s)". Is it the case that such periodic phenomena are not "noise" ?

Does it seem to make reasonable sense in your thinking to fashion three separate categories:

"random noises"; and

"signals"; and

"periodic signals" generated by the imaging hardware that are not "signal" and are not "noise" ?

Your statement appears to clearly exclude "almost periodic functions".

If there are stochastic elements, those are excluded.

Are they "signal", then ?

Again, I'm not going there.

Are they "signal", then?

I'm not taking a position on that. I don't think there's a universal answer to that question. (Perhaps I haven't been making myself clear here. If that's the case, I apologize.)

No need to apoligize when sincere individuals attempt to exchange ideas and opinions, my friend !

(If) it is your position that "signal" cannot be clearly, objectively defined, (then) how could/would any thinker(s) endeavor to attempt to clearly, objectively define what the term "noise" means ?

I just did so, without defining "signal". Maybe you don't like the definition. Maybe I don't either, in some contexts.

.

But let me give you an example in everyday, non-engineering, usage. I once moved my engineering department into temporary facilities while a new building was being constructed. The temp space was open plan, and everyone had had private offices up to then. I got complaints about people having difficulty concentrating. I dealt with that by having sound generators installed to mask low-level distant conversations. Everybody, including me, called the sound producers "white noise generators", even though, from another perspective, their output was the desired "signal".

Your point [which, to me, implies that clearly described elements of all related context(s) must also be included within any such "knowledge claims"] seems to indicate that your answer to the question posed in the title of the original post in this thread (appears to me) to be "no" ? Please advise ...

 

[JimKasson]

One little "beef" (that also came up when conversing with GB on his recent thread). If one follows your above statement, then it would seem that any periodic ("aka "pattern") "noise" that appears within a recorded image is (also, accordingly) not considered by you to be "noise" ? If and when such things appear in recorded images, it seems (for me) hard to refer to them as constituting "desired signal(s)". Is it the case that such periodic phenomena are not "noise" ?

Does it seem to make reasonable sense in your thinking to fashion three separate categories:

"random noises"; and

"signals"; and

"periodic signals" generated by the imaging hardware that are not "signal" and are not "noise" ?

First, in my way of thinking, more than one noise source generates "noise" not "noises".

If you must construct your imaging system model this way, please substitute "deterministic"* for periodic.

Thanks,

Jim

*always producing the same output if the underlying machine passes through the same sequence of states.

 

[JimKasson]

EXPERIMENTAL SETUP

We measured hand motion as a function of time, using small
board with a CMOS sensor. The board (shown in Fig. 1) form
factor was chosen to resemble that of a camera-phone. It is
approximately 5cm by 9cm and can be held like a camera-
phone. During each measurement, the sensor recorded VGA
video at 96 fps. The motion was later extracted from the video
sequence.

Each subject was asked to hold the board as steady as
possible for 2 seconds, after pointing the sensor at a standard
resolution chart (using a nearby computer screen). The
illumination conditions were regular indoor conditions
(around 300 Lux).

25 unpaid subjects (20 males and 5 females, ages from
25 to 50 years old) participated, each making 3 different
shots. The total number of different measurements
(motions) was 85.

Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

96 fps is too slow to measure all photographically interesting blur; that's less than 1/100 sec.

It sounds like the subjects were holding the instrument at arms length, like you'd do with a camera with an LCD back screen, not pressed against your body like a "real" camera.

I wonder what "nearby" means...

Jim

 

[Detail Man]

One little "beef" (that also came up when conversing with GB on his recent thread). If one follows your above statement, then it would seem that any periodic ("aka "pattern") "noise" that appears within a recorded image is (also, accordingly) not considered by you to be "noise" ? If and when such things appear in recorded images, it seems (for me) hard to refer to them as constituting "desired signal(s)". Is it the case that such periodic phenomena are not "noise" ?

Does it seem to make reasonable sense in your thinking to fashion three separate categories:

"random noises"; and

"signals"; and

"periodic signals" generated by the imaging hardware that are not "signal" and are not "noise" ?

First, in my way of thinking, more than one noise source generates "noise" not "noises".

Only once noise-sources are combined. Could you possibly answer my questions posed (above) ?

If you must construct your imaging system model this way, please substitute "deterministic"* for periodic.

Thanks,

Jim

I'll try to remember to use your certainly more impressive-sounding term (in place of the term "periodic", when applicable) when communicating with you in the future. It is my hope that an absence of reply to my questions is a "random" (as opposed to a "deterministic") phenomenon ...

*always producing the same output if the underlying machine passes through the same sequence of states.

 

[Detail Man]

EXPERIMENTAL SETUP

We measured hand motion as a function of time, using small
board with a CMOS sensor. The board (shown in Fig. 1) form
factor was chosen to resemble that of a camera-phone. It is
approximately 5cm by 9cm and can be held like a camera-
phone. During each measurement, the sensor recorded VGA
video at 96 fps. The motion was later extracted from the video
sequence.

Each subject was asked to hold the board as steady as
possible for 2 seconds, after pointing the sensor at a standard
resolution chart (using a nearby computer screen). The
illumination conditions were regular indoor conditions
(around 300 Lux).

25 unpaid subjects (20 males and 5 females, ages from
25 to 50 years old) participated, each making 3 different
shots. The total number of different measurements
(motions) was 85.

Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

96 fps is too slow to measure all photographically interesting blur; that's less than 1/100 sec.

Good point there - but not a big-time "deal-buster" where it comes to the subject, it seems to me.

.

It sounds like the subjects were holding the instrument at arms length, like you'd do with a camera with an LCD back screen, not pressed against your body like a "real" camera.

"Pressed against" (any part) of the research-subjects body would have been problematic in terms of determining the "homogeneity" (a truly impressive word indeed) of the test-conditions used.

Some people (self included) do not use viewfinders - as the perspectives that I desire in relation to the floral subject-matter that I photograph almost never exist at "eye-level" (and I am getting too old and feeble to contort my body in ways that might sometimes make such to be true).

Here are the additional "qualifiers" stated in the "Results" section of the paper:

At this point, we cannot explain the difference between the two axes. It is most likely the result of the USB cable (standard cable with 0.5 cm diameter) that connected the board to the PC.

Another factor could be the way the camera is held. Most subjects gripped the camera board by encircling the base of the board using one hand.

Fig. 3 presents the mean spectrum of the hand motions. This plot demonstrates that most of the energy of the motion lies in low frequencies (99% lies below 10Hz). This result is in agreement with previous studies of involuntary hand in agreement with previous studies of involuntary hand motion measuring hand motions in eye surgeons ([1]).

It should be noted that due to the sensor’s electronic rolling shutter (rather than the traditional mechanical shutter), our measurements are not completely accurate. The effect of the rolling shutter is that of a low pass filter, that of a rectangular function of width T[shutter] T in the time domain. In frequency domain this transforms to sinc (ω * T[shutter]). At low frequencies ( 50 < Hz), the effect is negligible.

Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

.

I wonder what "nearby" means...

It certainly would have been nice of them to specify the distances clearly - but, they did not ...

 

[JimKasson]

In the limit, since rotation seems to be the main culprit, ...

Quoting from the text cited above:

We first noticed that during one second of motion, the amount of rotation was negligible. The maximal rotation during a 1/96 Second period is 0.12 Degrees, for a full 1 Second it is 0.88 Degrees. In the rest of the analysis we ignore the rotation, as its effect is considerably smaller than that of the translation.

Boy, that's not my experience. Maybe the subjects were close.

Each subject was asked to hold the board as steady as
possible for 2 seconds, after pointing the sensor at a standard
resolution chart (using a nearby computer screen).

Source: "Hand Motion and Image Stabilization in Hand-held Devices", Etay Mar Or, Dmitry Pundik.

With subjects at infinity, translation wouldn't do anything to the image.

What sort of translation are you talking about (1-dimensional angular-linear conversions, I suppose?). We know, however, that angular rotation is a distinctly different matter. Am not sure how your comment relates to the above quoted text, though [or was it intended to (?)].

By translation, I mean moving the camera up and down, side to side, or forward and back with no rotation -- no yaw, no pitch, and no roll. This started when I made the statement in purple above that rotation is the long pole in the tent, and you quoted the study saying, in their experiment, that rotation was more important. I was assuming camera/subject distances of several meters. It may be that their distances were much less. That would explain the discrepancy.

Jim

 

[JimKasson]

One little "beef" (that also came up when conversing with GB on his recent thread). If one follows your above statement, then it would seem that any periodic ("aka "pattern") "noise" that appears within a recorded image is (also, accordingly) not considered by you to be "noise" ? If and when such things appear in recorded images, it seems (for me) hard to refer to them as constituting "desired signal(s)". Is it the case that such periodic phenomena are not "noise" ?

Does it seem to make reasonable sense in your thinking to fashion three separate categories:

"random noises"; and

"signals"; and

"periodic signals" generated by the imaging hardware that are not "signal" and are not "noise" ?

First, in my way of thinking, more than one noise source generates "noise" not "noises".

Only once noise-sources are combined. Could you possibly answer my questions posed (above) ?

If that's what you meant, the convention is to call them "noise sources". "Noises" sounds like something craching in the kitchen.

If you must construct your imaging system model this way, please substitute "deterministic"* for periodic.

I'll try to remember to use your certainly more impressive-sounding term (in place of the term "periodic", when applicable) when communicating with you in the future. It is my hope that an absence of reply to my questions is a "random" (as opposed to a "deterministic") phenomenon ...

DM, I'm not trying to come up with erudite language here, nor am I playing word games. I'm trying to reframe the concept. The opposite of noisy is not periodic; it's deterministic. For example,veiling flare's first-order effect is not periodic, unless you consider zero frequency a period. However, given a scene and a lens model, we can predict the veiling flare in the image. In a confusing interaction, that flare will give rise to photon noise, which is not deterministic, but I think you probably get the idea.

And yes, with those changes, I'd consider your three-part taxonomy useful.

Jim

 

[Luke Kaven]

One little "beef" (that also came up when conversing with GB on his recent thread). If one follows your above statement, then it would seem that any periodic ("aka "pattern") "noise" that appears within a recorded image is (also, accordingly) not considered by you to be "noise" ? If and when such things appear in recorded images, it seems (for me) hard to refer to them as constituting "desired signal(s)". Is it the case that such periodic phenomena are not "noise" ?

Does it seem to make reasonable sense in your thinking to fashion three separate categories:

"random noises"; and

"signals"; and

"periodic signals" generated by the imaging hardware that are not "signal" and are not "noise" ?

First, in my way of thinking, more than one noise source generates "noise" not "noises".

Only once noise-sources are combined. Could you possibly answer my questions posed (above) ?

If that's what you meant, the convention is to call them "noise sources". "Noises" sounds like something craching in the kitchen.

If you must construct your imaging system model this way, please substitute "deterministic"* for periodic.

I'll try to remember to use your certainly more impressive-sounding term (in place of the term "periodic", when applicable) when communicating with you in the future. It is my hope that an absence of reply to my questions is a "random" (as opposed to a "deterministic") phenomenon ...

DM, I'm not trying to come up with erudite language here, nor am I playing word games. I'm trying to reframe the concept. The opposite of noisy is not periodic; it's deterministic. For example,veiling flare's first-order effect is not periodic, unless you consider zero frequency a period. However, given a scene and a lens model, we can predict the veiling flare in the image. In a confusing interaction, that flare will give rise to photon noise, which is not deterministic, but I think you probably get the idea.

And yes, with those changes, I'd consider your three-part taxonomy useful.

Hi Jim,

What in principle would rule out deterministic phenomena from being counted as noise? Surely the scratches on my (former) LPs are deterministic both in their origin and in their effects. Crosstalk also comes to mind.

 

[JimKasson]

Hi Jim,

What in principle would rule out deterministic phenomena from being counted as noise? Surely the scratches on my (former) LPs are deterministic both in their origin and in their effects. Crosstalk also comes to mind.

Luke, you can define noise anyway you like. My personal preference is to use the term only for stochastic data, whether desired or not. Defining noise as anything one doesn't like in an image is not a firm rock on which to build testing and modeling protocols. Defining noise as any departure from the original scene is such a high bar that just about any image would consist mostly of noise, since things like capture metameric error and lens distortion would then be noise.

But that's just the way I look at things, and there's plenty of room in my world for people who look at things differently.

Jim

 

[Luke Kaven]

Hi Jim,

What in principle would rule out deterministic phenomena from being counted as noise? Surely the scratches on my (former) LPs are deterministic both in their origin and in their effects. Crosstalk also comes to mind.

Luke, you can define noise anyway you like. My personal preference is to use the term only for stochastic data, whether desired or not. Defining noise as anything one doesn't like in an image is not a firm rock on which to build testing and modeling protocols. Defining noise as any departure from the original scene is such a high bar that just about any image would consist mostly of noise, since things like capture metameric error and lens distortion would then be noise.

But that's just the way I look at things, and there's plenty of room in my world for people who look at things differently.

I would not think that one could define noise any way one wants, any more than one could say that one has "arthritis of the thigh." One supposes there is a fact of the matter.

I did offer a principle downthread. I realize contemporary philosophy of science and the semantics of theory terms is a bit obscure to engineers, but that's where the ground floor is.

Surely you'd agree that a scratch on a record produces noise, and that such noise has a deterministic cause, no?

 

[JimKasson]

Surely you'd agree that a scratch on a record produces noise, and that such noise has a deterministic cause, no?

No, I wouldn't. Not in this context. But, like you say, I'm an engineer, and probably incapable of understanding you.

[Added later: Or, more to the point, I'm comfortable in using words that are part of the common parlance that have different meaning in engineering and color science contexts than they do in normal speech. "Negative feedback", is an example, which I noticed recently that none other than Robert Rubin used to mean what I'd call "positive feedback with adverse results". The word "color" (or colour, depending on your location), is a word that, when I'm using it as a color scientist, has specific meaning which varies with context, and in no case, should I use it in those ways when I'm picking out paint with my wife, or I'll be in big trouble. "Hue" is another word like that. So is "brightness". So I'm not uncomfortable using the word noise in a particular way in a particular context.]

Jim

--
http://blog.kasson.com

 

[Luke Kaven]

Surely you'd agree that a scratch on a record produces noise, and that such noise has a deterministic cause, no?

No, I wouldn't. Not in this context. But, like you say, I'm an engineer, and probably incapable of understanding you.

[Added later: Or, more to the point, I'm comfortable in using words that are part of the common parlance that have different meaning in engineering and color science contexts than they do in normal speech. "Negative feedback", is an example, which I noticed recently that none other than Robert Rubin used to mean what I'd call "positive feedback with adverse results". The word "color" (or colour, depending on your location), is a word that, when I'm using it as a color scientist, has specific meaning which varies with context, and in no case, should I use it in those ways when I'm picking out paint with my wife, or I'll be in big trouble. "Hue" is another word like that. So is "brightness". So I'm not uncomfortable using the word noise in a particular way in a particular context.]

Just trying to disambiguate. Are you saying that what the scratch on my record produces on playback is not noise? Or are you saying that it doesn't have a deterministic cause? Imagine a thousand scratches on one record, enough to produce a random distribution of sonic disturbances on playback. Does that move us closer together in your view?

We've already got some general acceptance on a cluster definition for "noise", so I don't see a reason /in principle/ why the scratched-record case wouldn't be admitted.

BTW, I would not have used the word "incapable" in connection with you. There is an interfield exchange involved here, and it would not be my intention to diminish anyone.

 

[JimKasson]

Just trying to disambiguate. Are you saying that what the scratch on my record produces on playback is not noise?

Yes. Let's imaging that it's a perfect, mathematically well defined, scratch.

Or are you saying that it doesn't have a deterministic cause?

No, I'm accepting what I think the thrust of your example, that it is deterministic. Say it's completely radial and you hear it once per revolution.

Imagine a thousand scratches on one record, enough to produce a random distribution of sonic disturbances on playback. Does that move us closer together in your view?

A thousand such scratches would still be deterministic, modulo some strange stuff happening when they cross.

We've already got some general acceptance on a cluster definition for "noise", so I don't see a reason /in principle/ why the scratched-record case wouldn't be admitted.

Maybe I'm just an outlier.

BTW, I would not have used the word "incapable" in connection with you. There is an interfield exchange involved here, and it would not be my intention to diminish anyone.

Thank you for that.

Jim

 

[JimKasson]

We've already got some general acceptance on a cluster definition for "noise", so I don't see a reason /in principle/ why the scratched-record case wouldn't be admitted.

Maybe I'm just an outlier.

On 2nd thought, let's not give up so soon on that. What's the current accepted definition of noise?

Jim