[Dartmoor Dave]

Ian will correct me if I'm wrong, but that seems to be exactly what he did. The left image is a 100% crop from the 18mm shot and the right one is about a 3:1 (=53:18) downsample of the shot taken with the 53mm lens. About 9X the number of photons were collected and averaged to make each screen pixel of the image on the right relative to the image on the left.

Hopefully Ian will clarify for us whether the one on the left is upsampled or the one on the right is downsampled. My past experience is that proponents of equivalence tend to post upsampled examples and then say: "Look! More noise!" So I've assumed that that's what he did.

And of course I fully agree that the downsampled 4800x7200 image in your own example will have less noise than the downsampled 2400x3600. I've never suggested otherwise.

I think perhaps the thing is that you're using the term "total light" in a correct sense, while every time I've encountered the term in equivalence threads before it has been so garbled that it has seemed to contradict the basic facts of photographic exposure.

---

Last edited by Dartmoor Dave; 11-17-2019 at 09:19 AM.

[photoptimist]

Hopefully Ian will clarify for us whether the one on the left is upsampled or the one on the right is downsampled. My past experience is that proponents of equivalence tend to post upsampled examples and then say: "Look! More noise!" So I've assumed that that's what he did.

And of course I fully agree that the downsampled 4800x7200 image in your own example will have less noise than the downsampled 2400x3600. I've never suggested otherwise.

I think perhaps the thing is that you're using the term "total light" in a correct sense, while every time I've encountered the term in equivalence threads before it has been so garbled that it has seemed to contradict the basic facts of photographic exposure.

Alas, there are tons -- both metric & Imperial -- of garbled nonsense about "total light," many other photographic concepts, and just about everything in the realms of human experience.

It's like all the ill-informed people who talk about "equivalence" but then speak WRONGLY of having to change the camera-to-subject distance to handle a change in sensor size. ARGH!

[Ian Stuart Forsyth]

I'm assuming that the one on the right is a 100% crop, and the one on the left is a crop of far fewer pixels that you've upsampled to appear the same size on screen. So the appearance of more noise in the one on the left is a by-product of the upsampling,

No the one of the left was viewed at 100% and the one on the right was down sampled to the same size.

and nothing at all to do with "more light" being used to make the one on the right.

It has everything to do with more light. As a test we could shot the same scene the other way the only difference is that I would make so that both images would receive the same amount of light, would you hazard a guess as to the outcome of how the noise would appear if I was to again down sample the image with more pixels ?

Because they are made up of about the same amount of light how noise will appear would be very close

And of course, in the case of digital photography we are almost never looking at an upsampled image. Even if you use a 4K monitor, a photograph taken with any camera with more than 8 megapixels has to be downsampled so that you can see it all on screen. Up sampling only becomes an issue in digital photography if you decide to print bigger than the size that your actual pixel resolution would give you at the standard 300dpi.

Again none of this had to do with up sampling

Here is a simple test done using down sampling


https://photos.smugmug.com/Temp/Temp/i-w7QRvC7/0/39653a22/O/100mm.jpg

The above image was taken at 100mm iso 100 ƒ8 1/8sec and viewed at 100%


https://photos.smugmug.com/Temp/Temp/i-LmJ2GJg/0/18d8a043/O/200mm.jpg

And this image taken at equivalent settings 200mm iso400 ƒ16 1/8sec and downscaled to the same output size.

As you can see when they are made up of the same amount of light the appearance of noise is very similar in nature.

---------- Post added 11-17-2019 at 05:58 PM ----------

Perhaps in the interests of fair play, you could now post a 100% crop of the 18mm image, then downsample a crop from the 53mm to appear the same size on screen, so that we can all consider the difference in noise when you do that.

This was what was done so that there was no pixelization.

We can really see that its indeed the amount of light that has the greatest influence in how much noise we see in our final image unless there is a radical shift if sensor tech

And as shown above when you capture the same amount of light even when one image contains more pixels, and when they are viewed at the same output they will have very similar noise. Would this not tell you that its the total light that has the greatest influence on the noise that we capture.

[swanlefitte]

Been following this thread watching you fight an uphill battle, I know how difficult it can be over something that is very basic
lens light-gathering abilities - Page 15 - PentaxForums.com

It went on in this thread.
Focal length = focal length, I know . . . . . . . . . . BUT - PentaxForums.com
It changed my mind. So many variable are involved it's confusing as to what is being compared at any given time.
Thanks to all of you for elucidating the topic in it's various forms. Dynamic range at the moment.

[Dartmoor Dave]

Thank you for clarifying how you did that comparison:

No the one of the left was viewed at 100% and the one on the right was down sampled to the same size.

In this instance you have used the idea of "total light" in a valid way, as your aperture, shutter speed and ISO were consistent.



But then you've immediately gone and changed a significant factor in noise -- the ISO -- and once again used the concept of "total light" in an invalid way:

The above image was taken at 100mm iso 100 ƒ8 1/8sec and viewed at 100%

And this image taken at equivalent settings 200mm iso400 ƒ16 1/8sec and downscaled to the same output size.

This is like trying to nail jelly to the wall, and I've had enough of it.

[Class A]

But then you've immediately gone and changed a significant factor in noise -- the ISO -- and once again used the concept of "total light" in an invalid way:

No, he hasn't.

You appear to be overestimating the influence of the ISO setting on noise.
The reason why higher ISO settings appear to correlate with higher noise levels is that one typically raises the ISO setting when light levels are low. The lower light levels available to high-ISO captures are what primarily leads to higher noise levels.

Please see photoptimist's post about what the main source of noise in digital images nowadays is. The stochastic nature of light itself typically causes more noise than the image capturing technology.

If you had the choice between taking an image at ISO 100 and or at ISO 200 with the latter option having double the amount of photons available to form the image, say by increasing the illumination level (without any overexposure occurring in either case), which situation would you prefer?

Hint: The noise-reducing effect of more photons (total light) trumps any potential minor increase of the conversion noise floor implied by the ISO setting increase. A number of modern sensors are referred to as being essentially "ISO-less" because their performance does not depend on the ISO setting. Pretty much the same overall image noise is observable whether the shot is taken at, say ISO 800 or at ISO 100 and then pushed by three stops in post production. In other words, the ISO 800 image doesn't show any noise that has been introduced by the high ISO setting as such, but just the noise that was already present in the ISO 100 image.

P. S.: Ian used the correct equivalent ISO 400 setting.

[BigMackCam]

You know you've been following a thread discussion too long when... it's only 10:45am and already you feel like you need a stiff drink...

[Dartmoor Dave]

You know you've been following a thread discussion too long when... it's only 10:45am and already you feel like you need a stiff drink...

And apologies again to everyone for getting snappy. My real-world health problems have been making me too irritable and argumentative on the forum lately, which as I said a couple of pages back is not my usual style. I occasionally take a break for a couple of weeks to cool myself off, and I think it's probably time to do that again.

[BigMackCam]

And apologies again to everyone for getting snappy. My real-world health problems have been making me too irritable and argumentative on the forum lately, which as I said a couple of pages back is not my usual style. I occasionally take a break for a couple of weeks to cool myself off, and I think it's probably time to do that again.

Sorry to hear that, David - hope you see some improvement or get some relief soon...

[Rondec]

And apologies again to everyone for getting snappy. My real-world health problems have been making me too irritable and argumentative on the forum lately, which as I said a couple of pages back is not my usual style. I occasionally take a break for a couple of weeks to cool myself off, and I think it's probably time to do that again.

I hope you can get to feeling better. Certainly the internet isn't exactly a relaxing environment -- at least not all of the time.

[Ian Stuart Forsyth]

But then you've immediately gone and changed a significant factor in noise -- the ISO -- and once again used the concept of "total light" in an invalid way:

In most of the some cameras we have today as you increase the iso the seldom produce more noise in the image and some cameras as you increase the iso you can decrease the amount of noise the camera produces ( this is they very reason why you would want to increase the iso). the reason why we see more noise is that because what your are doing when selecting a higher iso is telling the cameras metering system to decrease the size of the exposure and its this decrease in exposure causing that noise.

There is a very good link on PF The Role of ISO Sensitivity - PentaxForums.com

Here Dan goes thru it

---------- Post added 11-19-2019 at 10:40 PM ----------

Hint: The noise-reducing effect of more photons (total light) trumps any potential minor increase of the conversion noise floor implied by the ISO setting increase. A number of modern sensors are referred to as being essentially "ISO-less" because their performance does not depend on the ISO setting. Pretty much the same overall image noise is observable whether the shot is taken at, say ISO 800 or at ISO 100 and then pushed by three stops in post production. In other words, the ISO 800 image doesn't show any noise that has been introduced by the high ISO setting as such, but just the noise that was already present in the ISO 100 image.
P. S.: Ian used the correct equivalent ISO 400 setting.

What is so silly is that if this was not true then it would be very simple to prove that this wrong but to this day no one has, if they had run the test they would see that they are wrong

[automorphism]

One thing you can do to see how ISO affects noise is take a photo at ISO 3200 on a bright sunny day or with flash. It will still look pretty good compared to ISO100, and pretty close to the same if your camera is ISO invariant.

[kaseki]

Against my better judgment, and feeling like Cyrano de Bergerac inspecting a Guillotine blade track, I have a few comments related to some of the recent posts.

Depth of field is an angular effect at the focal plane, wherein there is a zone of acceptable resolution due to the convolution of the optics circle of confusion and the pixel size (or film resolution) on each side of the exact focal point (not really a point, is the point). The larger the extreme ray angles, the shorter this zone is in the z direction. Short distances at the focal plane correspond to short boundaries on the focused ranges in the object/target/subject zone. The ray angle extremes are those determined by the aperture diameter (per Class A) and the focal length, a combination called f-number when the scene is far away. Hence I would always argue for calling depth of field an f/no phenomenon, with the caveat that object distance, lens resolution, and detector resolution are also factors.

Etendue is a simplification of the Lagrange Invariant, which derives from the laws of thermodynamics. The rule is: the étendue cannot be improved (made smaller). For large f/no, or tight laser beams, the conservation is simply given by beam diameter times aperture diameter. For larger beams or fields of view, the conservation is in terms of area x solid angle (steradians). In general, the integrals over the differential areas dotted with the differential directions of weighted rays impinging on (or emitted by) each area cannot be improved upon (focused tighter or made brighter).

Poisson noise (for small counts) or its Gaussian equivalent for large counts is the root mean square of deviations from the mean, and in both cases is given by the square root of the count number. It is not the noise that improves with higher counts, but the signal to noise ratio.

[StiffLegged]

Eh? Y’know that circle of confusion? Well, that’s me near the middle. Yeah, just there.

[Rondec]

Our sensors are pretty iso invariant meaning that if you underexpose iso 100 and push it 3 stops you should have similar end results as iso 800, at least with respect to noise. I mainly shoot iso 100 for the dynamic range, which is definitely better than at higher isos, but I must confess that the recent discussion has pretty much lost me.