If you’re talking film the answer is film has literal grain because it’s physical

Grain is only one part of the noise function. The other is the random nature of light itself.

If you’re talking digital the answer is sensors use gain to increase and decrease sensitivity

Nope. We can of course define “sensitivity” is multiple ways, but typically in this context the image sensors have one sensitivity (though different for each wavelength of light).

When you increase the gain you start having cross-talk as you overload individual pixels and the detection kinda bleed across them.

No, this is plan wrong.

Here’s what happens typically:

1. Lens draws a noisy image (because of photon shot noise - light is noisy, and this noise dominates outside of very light starved places)
2. Pixels capture this image - at this stage pixels add such a tiny amount of noise to the signal that is usually irrelevant.
3. Signal moves to PGA (programmable gain amplifier) - here the analogue signal is amplified according to selected settings (typically the ISO setting controls this)
4. Signal moves to ADC (analogue to digital converter) where it is digitized - this process is slightly noisy, thus some noise is added to the signal
5. A raw file is written or the camera processes a JPG and writes it

The reason why there is step 3 is in step 4 noise. If you increase the signal before ADC, the noise that the ADC adds becomes less relevant in comparison. The drawback for amplification is that the signal may “burn”, or the ADC operational range is exceeded.

A fun experiment is that you take two shots with identical exposure settings: one at ISO 100, the other at ISO 6400, both in raw format, and then process them in Lightroom or other such raw processor to the same lightness. For practically all today’s cameras the ISO 6400 will appear cleaner.

Sensors have a native ISO

No they do not. ISO is an output format (e.g. JPG) metric defined by ISO 12232 standard. It has nothing to do with image sensors.

Most cameras do change the operational parameters of the image sensor when the ISO setting is changed, typically the PGA (programmable gain amplifier) setting is changed and the signal is amplified the more the higher the camera’s ISO setting is.

and any extended (higher or lower) are simply the sensor amplifying or decreasing the signal.

That is not right.

Typically the ISO 100 (sometimes something else) setting on a camera is such that the image sensor is run at the lowest PGA amplification setting. Anything above that and the amplification is increased (and/or digital multiplication is used in software). The “extended lo” settings typically operate the sensor at the very same setting the ISO 100 does, just change the metering and processing of the data.

And no… high ISO doesn’t cause noise, it just amplifies it.

No it doesn’t.

ISO setting is a lightness parameter. Adjusting it ajusts JPG lightnesss. This is very different from amplifying noise.

Regarding raw files and the underlaying information, the sensor tries to capture - using a high ISO does not amplify noise. Instead it on typical camera slightly reduces the little noise that the analogue to digital convenrsion adds, and some cases also reduces the noise that the pixel adds.

What is noise?

Typically noise is deviation from the expected value. Expected value can be measured by taking multiple samples and the more samples you get, the more precise the expected value will be. In practise expected value is the average of the samples. Thus noise is deviation from that. Signal to noise ratio (SNR) tells you how much noise there is.

But where does noise REALLY come from?

Light itself is noisy - light particles, photons, hit the image plane randomly, following poisson distribution. Thus the image that the lens draws is in itself noisy. The more light particles you collect, the larger the SNR will be.

Almost all the noise in the raw-file data comes from randomness of light.

Image sensor adds a tiny little amount to this and it’s irrelevant unless very little light is present.

Slightly wider aliasing free shooting envelope.

I used to use unmodified Sony NEX-7 for some IR work, but the exposure times were very long as the camera’s IR filter cuts by far most of IR, thus a tripod was an absolute must. I imagine it’s pretty much the same for A6000 as it’s not likely that Sony has made the camera’s IR filter any weaker as that would influence colour reproduction siginificantly (like with the fiasco called Leica M8).

So you need an aproperiate filter - that R72 is likely fine, a tripod and a lens which is decently corrected (+decent coatings) for the IR wavelenghts to avoid hotspots. I think you can google for a list of suitable lenses.

If I’m going for the grainy retro look for some lowlight photos would it make sense to keep ISO low and add grain in post, or would the grain from turning up the ISO give a different kind of grainy look

“Grain” is a property of film, not digital. If you want to have a “film” like grain, the best option is to add it in processing or post-processing stage in computer - you have infinite options then to define what kind of uglyness you want.

Generally one should capture as much information as possible - it’s always easy to make low quality (for example the retro look you’re after) from high quality information, than try to get a low quality shot right away on the camera and hope it’s the right kind of low quality ;)

To collect lots of information, you should collect as much light as possible. Thus go for as long exposure and as large aperture as possible (and then you can set the ISO to as large as possible without burning anything, though often it’s much more practical just to let the ISO float automatically.)

Well, I simply wanted to clarify what you presented in a way which seemed to me to be causing more confustion to OP in this context than necessary.

Basically this was the problem for me:

don’t forget that focal length also affects light transmission. A longer focal length gathers light from a smaller area of the scene and therefore less light than a shorter focal length.

In the context of format comparison, it can get confusing to the OP as in this context the focal lengths collect light from the same area of the scene, not different.

Thus, it answers OP’s real question about why exposure isn’t changing even when the entrance pupil diameter changes

We interpreted OP’s question differently. Not saying either interpretion is better or worse.

> If you put a 50 mm lens at f3 on a full frame sensor and take a picture from 10feet away from your subject with the background 10 feet behind your subject. And then take the same photo with the same lens, but now set at F2, on a crump sensor, and you want the same framing you will have to

I only commented on your " There’s no such thing as aperture equivalent. " as being false as it is.

Thus the context is 50/3 on FF and 33/2 on APS-C. In this context the apertures are equivalent, thus your statement was false.

Focal length affects transmission in any context

It doesn’t, at least in principle.

The f-number is calculated the same regardless of format size, incorporating the focal length. Exposure settings values, including the aperture as expressed as an f-number, work the same for every format size.

Of course. Have I said anything else?

This discussion is about exposure purposes

Really? You went way beyond that for example when you wrote:

But don’t forget that focal length also affects light transmission. A longer focal length gathers light from a smaller area of the scene and therefore less light than a shorter focal length.

This has nothing to do with exposure purposes at all, but about how much light is collected. And as the context was FF vs APS-C and the size of aperture, what you wrote was hardly helpful.

> Most modern cameras are ISO invariant anyway.

Very few are truly ISO invariant as upping the ISO reduces the read noise slightly even with conventional sensors, plus many are dual gain which are never as there is one spot where the read noise collapses. Including your A9.

Interestingly, it was a long time ago when many cameras behaved in read-noise invariant manner (the CCD era). Some even only marked the ISO as metadata in the raw files.

> My Sony A9 gains nothing beyond ISO 800 compared to increasing exposure in post-processing, so I generally have my ISO set to Auto 100-800.

Sony A9 has read noise of about 2.5 electrons at ISO 800, it goes down to something like 1.2 or or 1.1 at the much higher ISOs. normally it indeed is quite irrelevant, thus I’d probably use the same ISO range myself that you’re using.

> I am a low light event pro, and I regularly increase my exposure 3-4 EV in post

Being a pro, I wish you’d teach the others better than this as you know that exposure and lightness are not the same and that one can’t increase the exposure in post. It would help the beginning learners to understand the fundamentals better and learn faster to make fewer mistakes.

Nice photo, btw.

A 50mm lens at F2.8 on a crop sensor camera will give you exactly the same depth of field and light as a 50mm lens at F2.8 on a full frame.

Different light actually. Cropping throws away light.

And diffent DOF as you enlarge the images from image plane by different amounts to the viewing size, though as the field of views are also diffent, comparing DOFs isn’t IMHO too meaningful.

So this meant that they would say that a 25mm f2.8 is the equivalent of a full frame 50mm f5.8 – but this only equated to the depth of field and not the light gathering capabilities of the lens.

The light collecting is a function of several parameters: scene luminance, field of view, exposure time and aperture size.

We’re only interested in the aperture size here. 50mm/5.6 and 25mm/2.8 both have about 9mm aperture diameter, or area of 63mm^2 or so. Thus the same amount of light goes through.

The practical implication of this is that the pictures from these systems would look identical to each other.

If we were to increase the light collecting of either system via making the aperture larger, we’d improve the image quality (noisewise) but also decrease the depth of field.

Larger aperture diameter (with the same field of view, as in this case) will collect more light over the same amount of time (and also the depth of field, DOF, will be more shallow). Thus with equal exposure time the larger format will have better image quality (due to more light being collected).

The density of light will be the same though as the light is spread over different areas (FF is 2.3 times larger than APS-C). This is relevant for the exposure calcualtion of the cameras exposure meter. Thus if you use an automatic exposure program, then same f-number on different formats gives you the same exposure time. That’s kind of the point of the whole f-number system.

There’s no such thing as aperture equivalent.

Of course there is. If one compares different formats, then f/2 on APS-C and f/3 on FF do identical job - in principle it would be impossible to know which system was used. Thus the apertures are equivalent.

But don’t forget that focal length also affects light transmission. A longer focal length gathers light from a smaller area of the scene and therefore less light than a shorter focal length.

This is not the case in this context as it was specifically about APS-C vs FF. 23mm APS-C lens and 35mm FF lens have the same field of view, thus the same light reflects from the scene to the lens.

A 16mm entrance pupil diameter lets in less light while a 23mm focal length lets in more light, and that equals out to the same as a 25mm entrance pupil diameter letting in more light and paired with a 35mm focal length letting in less light.

A 16mm entrance pupil of 23mm f/1.4 on APS-C collects about 2.3 times less light than 35mm f/1.4 on FF with it’s 25mm entrance pupil - the same amount of light reflects from the scene to the lens, but the system with the larger entrance pupil collects more light.

For exposure purposes this is of course not relevant, but for image quality and DOF it is.

An f/2.8 lens will shoot as fast as f/2.8 does, no matter what you put it on.

But what is this “speed” or “fastness”? The f-number is not just some abstract unit, but it’s tied to physical reality.

To answer to that one needs to understand that this “fastness” refers to an exposure time (or shutter speed) one can use to achieve desired image quality. Thus f/2.8 on one format does not equal f/2.8 on another.

In principle f/4.2 on FF, f/2.8 on APS-C do an identical job - they give the same image quality (noisewise) with the same exposure time, the depth of field is the same and the blur from diffraction will be the same.

it is said that full frame aperture equivalent of 2.8 to aps-c is 4.2

Assuming same field of view, in principle f/4.2 on FF and f/2.8 on APS-C will create create identical results: DOF, noise, diffraction blur all will be the same.

In practise there may be some differences:

• FF can usually capture more light before burning, thus one may be able to use a longer exposure time to get better image quality.
• FF image is enlarged less for the final output, thus the lens can be somewhat worse to get the same resolution
• The smaller f-number requirement for APS-C may mean more complex and/or inferior optics

For most shooters it’s quite irrelevant if one uses FF or APS-C - not that much difference usually.

does it mean that shutter speed of aps-c is one stop slower that full frame on the same aperture?

Shutter speed can be what ever the user wants. If one let’s the camera to decide it instead, then an autoexposure program of the camera would likely use a different exposure time in each case, assuming the same ISO and light levels.

Native ISO is the natural ISO range that your sensor can handle without extra digital manipulation

The image sensor has no ISO range, or ISO setting at all.

Typically different ISO settings on the camera cause the sensor to be operated with different parameters (leading to different noise performance curve). May sound like nitpicking, but I think it’s important to understand the difference.

So for example that would be like ISO100-6400 as a native range and the low setting (ISO50) and high setting (iso12800) would be extended ISO as they are digitally manipulated.

This is wrong.

Typically the image sensor operates at the same setting at “extended low ISOs” and with the “lowest normal ISO” (though some extended settings may use the same parameters of ISO 125 or ISO 160 for example). What is different is metering - with the extended settings the camera exposure program simply causes a larger exposure to be used by default, thus there will be less headroom (in the JPGs, thus no longer ISO standard compliant). That’s it.

The “extended high” settings on the other hand typically add (digital) multiplication to the data. All it does for raw is that it cuts the headroom. For JPG shooters it’s not relevant if it’s “extended high” or not.

Pretty much right, though I’ll clarify a bit if I may:

Seeing more noise at higher ISOs is because light noise is related to the square root of the number of photons

Noise is related to the number of photons. What the square root gives is the standard deviation (due to the Poisson distribution that light particles follow) and “noise” is the signal to noise ratio, SNR, the ratio of the number of photons to the standard deviation. And this is of course funnily also the square root of the number of photons.

A lower number has a relatively higher square root. Less light has relatively higher noise

I do understand what you mean, but this is IMHO rather confusingly put and especially for beginners impossible to comprehend what you mean.

Simply, more light means larger SNR. The standard deviation of the signal goes up, but the signal itself goes up much more. What an observer sees as “noise” is simply a product of low SNR.

The base ISO is the one where the sensor system is designed to have the highest signal to noise ratio at the “correct” exposure.

The image sensor has no ISO at all. It can be run at different parameters and typically changing the camera’s ISO setting changes there operational parameters.

The concept of “base ISO” isn’t officially defined (by ISO, the organization) and can interpreted in many ways. If we go by your definition above, then it would be typically the lowest extended ISO setting as the exposures are typically larger than the ones with the lowest “regular ISO” settings. Many people on the other hand seem to consider the lowest “regular ISO” to be the “base ISO”. FWIW, usually the image sensor is driven with the same parameters in both cases.

ISO value itself is really a property of output formats, like JPG.

IMHO the concept of base ISO should be abolished for above reasons. It’s likely more harmful than useful for beginners.

1. Normally, as you raise the ISO the maximum amount of light the image sensor can collect goes down.
2. The more light you collect the better the image quality because light itself is noisy - the more you capture, the less noisy things appear to you.
3. Raising ISO makes the JPG brighter, reducing it darker.
4. Raising ISO normally reduces the camera added noise slightly - this is mainly relevant if you shoot raw.