"Net Noise" continuing the iso invariance discussion.

So we were having a good discussion about iso invariance:

ISO Invariance, why not?

And what came to light is there are quite a few versions of noise, which then lead to discussions of the cause of that noise, and then to discussions of is it really noise, etc...

In the end, all I really care about is "Net Noise"

Net Noise - The amount of noise left in the image when viewing or printing.

I want to know how different things affect it during capture like:

Exposure, Dynamic range, contrast, sensor saturation, light, etc...

I want to know what I can do to mitigate it:

Different sensors, Post processing techniques, including Denoisers, Flash, selecting better iso starting points (this is where a lot of the back and forth starts).

The need for various noise definitions depend on where you are in the photography sphere, designer of photography products, capturer of the image or just a viewer of the image. I find SNR, beginning noise, noise floor, etc... things that I don't need when capturing an image. As a capturer of images, what I found amazing was that my m43 camera sensors were producing no more noise than my FF sensor, yet the "net noise" in low light photos was unacceptable to me. Crazy right?

But in the end the only noise I care about is net noise, what am I looking at in the final product.

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Are you asking how to reduce visible noise in your final images as viewed?

Andrew

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.

These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.

I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

ahaslett said:

Are you asking how to reduce visible noise in your final images as viewed?

Andrew

Click to expand...

Yes that's the place from where I start, and once you bring up noise the conversation invariably goes down the various noise rabbit holes. As an example:

You can say Higher ISO causes more noise. And in a practical image sense it does.

BUT once you start discussing the image creation process you go down the road of

"ISO doesn't cause noise, only exposure"
"Inability to capture the full dynamic range causes noise, not ISO"
etc...

But I know when I move the iso on my camera to a higher value, the noise in the final image increases.

jhunna said:

ahaslett said:

Are you asking how to reduce visible noise in your final images as viewed?

Andrew

Click to expand...

Yes that's the place from where I start, and once you bring up noise the conversation invariably goes down the various noise rabbit holes. As an example:

You can say Higher ISO causes more noise. And in a practical image sense it does.

BUT once you start discussing the image creation process you go down the road of

"ISO doesn't cause noise, only exposure"
"Inability to capture the full dynamic range causes noise, not ISO"
etc...

But I know when I move the iso on my camera to a higher value, the noise in the final image increases.

Click to expand...

That’s because ISO affects metering because digital cameras are designed to mimic film cameras when producing jpegs.

If you shoot RAW, for any given camera, the appearance of noise is determined by exposure (T-stop times shutter speed) and the local scene luminance in the areas where you are concerned about noise. That is shot noise, then there are other sources of noise.

You can dramatically change the appearance of noise by processing.

As was pointed out in the previous thread, in many cameras you get less input referred read noise in the dark parts of your picture by using a higher ISO, at the expense of restricting how much light you can capture in the highlights.

My exposure strategy is to use the largest aperture that fits my composition, the longest exposure also. I tend to shoot either at base ISO or in Auto ISO with a setting that depends on whether the lens has OSS or not.

I use EC to make best use of the available DR at base ISO. In cameras with less DR than the R4, I sometimes use UniWB to avoid the worst excesses of jpeg metering and EC to push the headroom on the histogram that I know the OEM left.

For long exposures, I use LENR which adds shot noise but removes annoying noise from thermal effects.

For most of my images, shot noise dominates.

Sometimes, stacking multiple images is necessary. This increases noise and signal. Since signal is increased a lot more, when you adjust the brightness of the final image to be constant, the appearance of noise is reduced compared to a single image.

Andrew

jhunna said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.
These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.
I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

Click to expand...

First: Why do you compensate -2 if less noise is your goal?

jhunna said:

ahaslett said:

Are you asking how to reduce visible noise in your final images as viewed?

Andrew

Click to expand...

Yes that's the place from where I start, and once you bring up noise the conversation invariably goes down the various noise rabbit holes. As an example:

You can say Higher ISO causes more noise. And in a practical image sense it does.

BUT once you start discussing the image creation process you go down the road of

"ISO doesn't cause noise, only exposure"
"Inability to capture the full dynamic range causes noise, not ISO"
etc...

But I know when I move the iso on my camera to a higher value, the noise in the final image increases.

Click to expand...

Do you reduce exposure when raising the ISO? If so, sure you will see more noise!

If exposure is kept the same, the results will look pretty much the same.

Magnar W said:

jhunna said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.
These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.
I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

Click to expand...

First: Why do you compensate -2 if less noise is your goal?

Click to expand...

I think the 7c over exposes but at least 1 stop of light. I am able to recover more shadow detail so I overcompensate buy another stop. -2 EV is a personal preference, that gives me the results I like. I may be able to lower the net noise of the image, but don't feel the need to do so.

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.
These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.
I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

Click to expand...

First: Why do you compensate -2 if less noise is your goal?

Click to expand...

I think the 7c over exposes but at least 1 stop of light. I am able to recover more shadow detail so I overcompensate buy another stop. -2 EV is a personal preference, that gives me the results I like. I may be able to lower the net noise of the image, but don't feel the need to do so.

Click to expand...

If you really are concerned about as little visible noise as possible from your RAW files, you expose slightly "to the right" at base ISO, and then adjust brightness when post processing the files.

If you are pleased with the results, just do what works for you.

Magnar W said:

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.
These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.
I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

Click to expand...

First: Why do you compensate -2 if less noise is your goal?

Click to expand...

I think the 7c over exposes but at least 1 stop of light. I am able to recover more shadow detail so I overcompensate buy another stop. -2 EV is a personal preference, that gives me the results I like. I may be able to lower the net noise of the image, but don't feel the need to do so.

Click to expand...

If you really are concerned about as little visible noise as possible from your RAW files, you expose slightly "to the right" at base ISO, and then adjust brightness when post processing the files.

If you are pleased with the results, just do what works for you.

Click to expand...

That's what I do, i just gave my specifics to my process.

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.
These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.
I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

Click to expand...

First: Why do you compensate -2 if less noise is your goal?

Click to expand...

I think the 7c over exposes but at least 1 stop of light. I am able to recover more shadow detail so I overcompensate buy another stop. -2 EV is a personal preference, that gives me the results I like. I may be able to lower the net noise of the image, but don't feel the need to do so.

Click to expand...

If you really are concerned about as little visible noise as possible from your RAW files, you expose slightly "to the right" at base ISO, and then adjust brightness when post processing the files.

If you are pleased with the results, just do what works for you.

Click to expand...

That's what I do, i just gave my specifics to my process.

Click to expand...

So you are happy to give up all the noise advantage of the larger sensor and a bit more?

Andrew

ahaslett said:

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.
These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.
I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

Click to expand...

First: Why do you compensate -2 if less noise is your goal?

Click to expand...

I think the 7c over exposes but at least 1 stop of light. I am able to recover more shadow detail so I overcompensate buy another stop. -2 EV is a personal preference, that gives me the results I like. I may be able to lower the net noise of the image, but don't feel the need to do so.

Click to expand...

If you really are concerned about as little visible noise as possible from your RAW files, you expose slightly "to the right" at base ISO, and then adjust brightness when post processing the files.

If you are pleased with the results, just do what works for you.

Click to expand...

That's what I do, i just gave my specifics to my process.

Click to expand...

So you are happy to give up all the noise advantage of the larger sensor and a bit more?

Click to expand...

That's not what happens in practice. It is still better by far than my smaller sensor cameras.

ahaslett said:

Andrew

Click to expand...

jhunna said:

So we were having a good discussion about iso invariance:
ISO Invariance, why not?

And what came to light is there are quite a few versions of noise, which then lead to discussions of the cause of that noise, and then to discussions of is it really noise, etc...
In the end, all I really care about is "Net Noise"
Net Noise - The amount of noise left in the image when viewing or printing.

I want to know how different things affect it during capture like:

Exposure, Dynamic range, contrast, sensor saturation, light, etc...

I want to know what I can do to mitigate it:

Different sensors, Post processing techniques, including Denoisers, Flash, selecting better iso starting points (this is where a lot of the back and forth starts).

The need for various noise definitions depend on where you are in the photography sphere, designer of photography products, capturer of the image or just a viewer of the image. I find SNR, beginning noise, noise floor, etc... things that I don't need when capturing an image. As a capturer of images, what I found amazing was that my m43 camera sensors were producing no more noise than my FF sensor, yet the "net noise" in low light photos was unacceptable to me. Crazy right?

But in the end the only noise I care about is net noise, what am I looking at in the final product.

Click to expand...

This is not scientific.

You can take an image of a face being lit with a candle at high ISO with little noise, whereas you can take an image of the same face in the shadows (no lights source) with lots of noise.

Further, certain types of noise are easily remedied with software cleanup tools. This works for all formats, but the success depends again very much on image content.

You also have to compare RAW+post-processing-software, as opposed to JPG. JPG processing can reduce noise already, at the cost of detail. In some cases, the noise is not disturbing, and processing too heavily can lead to washed out results.

In my experience, if there is direct (or bounced) light on your subject, any ISO value will yield satisfactory results. If your subject is unlit, no ISO value will give you the results that you seek.

I don't want to name this shadow-noise, as there is more too it, but in the first case, the same face being lit or unlit to me is the difference between a picture worth taking versus skipping.

The best denoising tools cannot overcome poor exposure and poor illumination.

I have seen side by side results where the difference between sensor formats may appear to be little, whereas in other shots the difference between said formats is very noticeable.

Bill Claff maintains a site which measures the various sensor formats, and translates the discussions into scientific data points: https://photonstophotos.net/

Check e.g. EM1 versus A7R4:
https://photonstophotos.net/Charts/PDR.htm#Olympus OM-D E-M1 Mark III,Sony ILCE-7RM4


OM-D EM1 M3 vs ILCE-7RM4

• Camera Model ---------------------- Maximum PDR --- Low Light ISO --- Low Light EV
• Olympus OM-D E-M1 Mark III -------------- 9.74 ---------------- 2,472 -------------- 9.63
• Sony ILCE-7RM4 ------------------------------- 11.62 ---------------- 4,744 ------------ 10.57

The 'Low Light ISO' score is showing a 2x factor, which is expected by sensor format comparison.

This is scientific. An observation that e.g. my 'cell phone camera' has the same noise as a different sensor camera for a particular scene is anecdotal.

--
Cheers,
Henry

Net noise:

We have two tresholds.

If the image sensor recieves too few photons no image is recorded. Noise only, no image.

If the image sensor recieves too much light the image is saturated. No noise, no image.

Dynamic range is the difference in stops between those tresholds.

The image sensor does NOT become more sensitive to light when ISO is raised. What basically happens is that the exposure is reduced and the image is brightened by increasing the gain (kind of a fine tuned version of that).

Read noise is a function of the camera electronics (the image sensor is part of that). Read noise actually decrease when ISO is raised, but we loose dynamic range.

Shoot noise increase when the image is less exposed.

Exposure is about shutter speed and aperture (f/number). Like 1/500sec at f/8.

Raising the ISO:

Most let the camera choose shutter speed and aperture when raising the ISO. That leads to a less exposed image. Shoot noise increases and the image becomes more noisy. Raising the ISO is kind of equal to underexposing and brightening the image in post processing.

To take full advantage of the dynamic range of the image sensor and to get as noise free images as possible, use base ISO and expose properly (manual settings and no auto ISO).

In low light using the second gain stop have merit as read noise is reduced. However read noise is usually swamped by shoot noise. Under expose, and shoot noise increases.

Just test for yourself (skip JPGs if seriously interested in dynamic range):

1) Increase ISO and expose by the camera meter (reduce one EV per raised ISO stop).

2) Set the exposure, keep the settings, and take images at different ISOs.

What do you see?

jhunna said:

ahaslett said:

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

jhunna said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

That's what this thread is attempting to amplify and clarify. We all have our own perspectives and tests. I am constantly testing, and then I run additional tests once I learn about new tests. BUT what matters is the end image.
These technical discussions can go so deep that we lose site that all that matters is you have an image you enjoy.

Most people are absolutely fine putting the camera in auto, and firing away. But as you learn more, you want to control your look. Noise is one of the things I have been trying to control, and I do mean control as sometimes it adds to the look.
I take lots of pictures, primarily of people indoors, so I have a solution that works for me:

A7C, aperture mode, Lens wide open, ISO 640, -2 EV

This gives me the most flexibility with the raw, and I don't really care what the image looks like in the camera, as I have post processing tools that can turn them into what I want in the final image. The one "trick" that seems to have allowed me to do this is iso invariance. A concept that is debatable, but seems to work very well for me in practice with this particular set up.

So back to the original point which is when discussing these esoteric concepts when I refer to noise I am almost always describing/discussing the noise in the end product and what to do to better manipulate that "net noise".

Click to expand...

First: Why do you compensate -2 if less noise is your goal?

Click to expand...

I think the 7c over exposes but at least 1 stop of light. I am able to recover more shadow detail so I overcompensate buy another stop. -2 EV is a personal preference, that gives me the results I like. I may be able to lower the net noise of the image, but don't feel the need to do so.

Click to expand...

If you really are concerned about as little visible noise as possible from your RAW files, you expose slightly "to the right" at base ISO, and then adjust brightness when post processing the files.

If you are pleased with the results, just do what works for you.

Click to expand...

That's what I do, i just gave my specifics to my process.

Click to expand...

So you are happy to give up all the noise advantage of the larger sensor and a bit more?

Click to expand...

That's not what happens in practice. It is still better by far than my smaller sensor cameras.

jhunna said:

Andrew

Click to expand...

Click to expand...

I too have FF and MFT bodies, but both have Sony sensors.

I don’t understand your exposure strategy.

Andrew

Tristimulus said:

Most let the camera choose shutter speed and aperture when raising the ISO. That leads to a less exposed image. Shoot noise increases and the image becomes more noisy. Raising the ISO is kind of equal to underexposing and brightening the image in post processing.

Click to expand...

Which is a huge mistake, right? One should raise ISO only if one has reached the limit of desired aperture and shutter speeds.

The other mistake is needlessly underexposing by lowering ISO without changing exposure (i.e. aperture and shutter speed) cause it brings the signal only closer to the noise floor.

Hope I got this right ---

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

I like the quote: "One test is worth a thousand expert opinions."

The Ryantist said:

Magnar W said:

Why not just do a few real world tests to find out how the gear you use perform under different conditions? Doesn't take long and the result tells more than graphs and theoretical values posted by others (which will be confirmed by your tests anyway). This is fun, and you will probably learn a lot too.

Theories are great, and a pragmatic approach to photography is real world use of the theories.

Click to expand...

I like the quote: "One test is worth a thousand expert opinions."

Click to expand...

Agreed, but the discussion opens up different ideas. I take what I like, and then discard the rest. I don't know who these methical people we are trying to convence. I ask questions for clarity, and for other's experience.

You could shoot your sensor out into space and shield it from the sun like the sensors on JWST. Or something like this:

https://astronomy-imaging-camera.com/product/asi6200mc-pro-color