Does a bigger sensor make a better picture?

[Ags]

This is something I have wondered about... does a bigger sensor result in a better picture, if we keep the scope the same and select similar targets appropriate to the framing of each sensor? Also assume the pictures are displayed at normal picture size, so no pixel peeping! The noise grain would be finer on the larger sensor, the stars would be smaller and the larger sensor would obviously capture more light/data... In conventional photography a larger sensor gives better bokeh etc, but does a similar rule apply in astrophotography?

[Elp]

Define "better"...

[Ags]

More nicerer.

More seriously, I am just looking for different perspectives, so i am sure people will have their own standards of 'better' which will lead to divergent answers.

Edited May 24 by Ags

[Elp]

I think the term you're tending toward is resolution. We can get better resolution by using larger aperture which normally means longer focal length so a targets details will be spread over more pixels so more resolution. If you image at a better resolution and mosaic to compare with the same FOV with a wide field lens, then yes, that way the former will look better.

You also see a small marginal improvement in quality when imaging mono over OSC as you utilise all pixels per bandpass.

If you talk about subject framing, then you're getting into composition, which can also be subjective to the individual.

Edited May 24 by Elp

[Oscar23]

My knowledge on this one is somewhat surface-level but I will explain as best I can. Increasing sensor size, all else being equal, will also increase the field of view. So, you'll be imaging a larger patch of sky, and any subjects within that patch will look smaller, i.e. covering fewer pixels. But, you can get that back by increasing focal length, using a barlow (or a stronger barlow), etc. So you're back at the same field of view, but with a larger sensor. If that sensor is the same resolution, every pixel will be physically bigger and thus more sensitive to light, so you can get a brighter image or reduce ISO for the same brightness and thus reduce noise. Conversely, if the pixel size is unchanged, there will of course be more pixels in the larger sensor, so while it won't be any more sensitive to light, now any subject in the field will cover more pixels, which should in theory yield greater detail. And of course if it's somewhere in the middle you can get a combination of both benefits, to proportionately lesser degrees.

I understand "better" is rather subjective in this case, but, as I see it, assuming you can achieve your desired FOV with both, without sacrificing optical quality, a larger sensor should make for a better picture by some metric, yes. That's no problem for non-astro applications which is why the best and most expensive "flagship" DSLRs/MILCs (Canon 5D, Sony Alpha-7, Nikon Z 5, etc) are usually full-frame. The main reasons to not go full-frame in photography are size, weight, and cost. You don't get better pictures for it. But, it can be challenging to get a narrow enough FOV in AP, especially if the scope doesn't have a very long focal length. Conversely, it can also be hard to get a wide enough FOV for large DSOs if your scope does have a long focal length... sensor size has some obvious parallels with eyepiece field stop size here.

Hopefully not too wordy, I tried to explain as straight-forward-ly as I could

Edited May 25 by Oscar23

[Mr Spock]

A larger sensor will give a wider field of view for the same focal length. So when you display them side by side the image from the larger sensor will look smaller.

The same amount of light will fall on each pixel, assuming pixel size is constant, so you won't get in more light per se. 

Terrestrial photography is completely different. Picture quality depends a lot on the dynamic range of the sensor and the ISO used. The field of view changes with sensor size. So can pixel size, so, a full frame 24mp and a 24mp crop frame will have the same number of pixels. However, the FF camera will have larger pixels and so better low light performance.

I'll assume like most photographers you misunderstand what bokeh is. Bokeh is how out of focus areas are rendered, not how shallow the depth of field is. It is a function of lens design not aperture.

[vlaiv]

Yes, and in different ways - depending on how you utilize it.

First of - bigger sensor is faster on a given target if you pair it with appropriate scope - and thus produces nicer image in sense of less noise. However in order to achieve this effect - you need to

- use larger scope (that will provide you with equal FOV) - which has larger aperture

- use some sort of binning to achieve target resolution (you should really compare two equal sampling rates in this case)

Second - bigger sensor will produce nicer looking image with same scope and same sampling rate provided that:

- there is some residual noise in both images and

- you view larger image at "fit to screen" sort of scenario - where screen is smaller in area than total pixel area of larger image.

Above conditions lead to display software using scaling for larger image and not for smaller image. Scaling has very similar effect to binning - it reduces noise or improves SNR (difference being that it also introduces pixel to pixel correlation and has unpredictable SNR improvement, usually less than binning - so for all intents and purposes, binning is the right way to do this).

However in scenarios where you don't have this scaling - two images will look equally "nice" - this is when cropping the larger sensor to get same true field of view or viewing both images at 100% (or same scale for that matter).

So we could say that first is true improvement in image quality while second is apparent improvement in image quality (that will go away under different circumstances), although - you can also make second be permanent if you scale image yourself in processing (but then again - why not use binning instead?)

[Ags]

3 hours ago, Mr Spock said:

I'll assume like most photographers you misunderstand what bokeh is. Bokeh is how out of focus areas are rendered, not how shallow the depth of field is. It is a function of lens design not aperture.

I should have been clearer. With a larger sensor (e.g. full frame camera vs crop) you would use a longer lens for the same framing. The longer lens has shallower depth of focus so it has more pronounced bokeh. All quite irrelevant for AP of course.

Edited May 25 by Ags

[Oscar23]

Yes, bokeh and other depth-of-field effects are of no concern here - to a camera, the cosmos are flat

[Fegato]

I think pixel size is an important parameter here - are you assuming pixel size is identical?  So if e.g. you "upgrade" from an ASI2600 to ASI6200 on the same scope - in my book, the overlapping (APS C size) image area will surely be identical?  As for the extra bits around the edge of the frame on the full frame sensor - well it depends on how well your scope / setup can cope with the extra size. You might suffer from vignetting, or greater issues with stars if you have any tilt or other problems with the precision of your image train.

[Ratlet]

One nice thing about a smaller sensor is that you are basically pre-cropping the image compared to a larger sensor so you can lose some aberrations and imperfections.

With a 533 sensor on my 130pds I don't really need a coma corrector because the bulk of the coma lands of the sensor.  I could achieve the same effect by using a larger sensor and cropping the offending section out of course.

[Second Time Around]

I've learnt a lot from reading the articles by Roger Clark, a former NASA photographer, in which he explodes some popular myths by showing the results of actual experiments.

Go to https://clarkvision.com/articles/characteristics-of-best-cameras-and-lenses-for-nightscape-astro-photography/

Scroll down to the section "Crop versus full frame sensors".

I think many will be surprised!

[Oscar23]

Steve, that's a very dubious and possibly disingenuous article (or, at least, article section) - Mr. Clark did not adjust ISO or focal length to compensate for the increased light sensitivity; you can actually see that the 6D image is over-exposed in his own example! The stars are blown out, and the nebulosity is brighter. Signal-to-noise is "extremely close to equal" because ISO is equal. Cropping is not a good way to take advantage of a larger sensor... you are literally throwing away the extra working area!

[Ags]

I think the answer to my question is a qualified yes.

 - for small targets that frame well on both sensors, there is no advantage from a larger sensor size

- for large targets that requires a mosaic in one or both sensors, the larger sensor has an enormous advantage in light gathering

- a larger sensor will provide more artistic options for target framing

- a larger sensor can be used more easily with larger apertures

- when comparing different size sensors with the same scope but different targets appropriate to each sensor, the large sensor image can be significantly better because it will have gathered far more light from its target (I.e. the large sensor target covers more mm2 on the sensor, so if surface brightness of the two DSOs is the same, the larger sensor will have collected more light)

- the results from both sensors may appear more similar if one insist on 100% zoom pixel peeping, but doing so is like imaging with both sensors reduced to the same ROI, which would defeat the point of comparing the sensor sizes

Edited May 25 by Ags

[Vroobel]

I have a Canon 6D (modded) which is full-frame and also have APS-C cameras (2600MM and 2600MC). I matured to a conclusion that the full-frame DSLR will work with lenses somewhere under a darker sky soon while the APS-C astro-cameras work in my garden under Bortle 9(+). Presently it's much cheaper to buy an affordable high quality APS-C astro-camera and another, smaller scope, even if it's APO, than the same quality full-frame camera which additionally requires much higher quality scope. The another smaller scope gives you a bigger FoV and more ways for the artistic composition. This way I have 2" 230mm Askar, 4" 572mm Altair and 10" 1285mm Newtonian which give me around 2.5x of difference between the FoV of each. And I always can do the crop, even if it's a waste of the sensor surface or photons. 

[Elp]

One thing that hasn't been mentioned, the camera sensor size has to fit within an optics illuminated circle to reap the benefit of using the larger sensor.

Also not all bodied cameras perform the same so image qualities will vary, the case is different if we're talking about same sensor tech astro cameras which levels the playing field somewhat between models.

[Bivanus]

Does a bigger sensor make a better picture...so just the sensor size we're talking about ? Then the answer is simply NO. 

Photography - just like philosophy - can't be bought by the kilogram. Something about both subjects requiring a systemic aproach  with many parts of said system having to work toghether to get a 'better' result.

Let me simply enumerate a couple thinghies: is your telescope big enough for that big sensor ? is it colimated & aclimatised ? is it an Achro/APO/Mirror/Catadioptric ? What sky are you working with? Can you see the sky - hello Cumullus my old friend...  How are your post processing skills ( mine are so bad the wavelets refuse to wave or something😅)

If BIGGER would be BETTER just like that , I have a feeling there would be WAY much more discussions about how to adapt those pesky Hasselblads or other medium format camera's to a telescope and we would all do it like Teruyasu Kitayama...   Teruyasu Kitayama | Capturing Nebulas and Pinwheel Galaxies (hasselblad.com)

Edited May 25 by Bivanus
link added

[Oscar23]

Being able to fully illuminate the sensor is assumed; using a sensor too physically large for the telescope to illuminate it would obviously be a configuration error. I didn't think about it because it makes no sense to consider in this context. The question is not whether using the wrong gear together is better.

Bivanus, I suspect the reason there's little talk about medium format cameras is because they are huge, extremely expensive, not astro-specialised, and the sensor size exceeds what's practical to fit using available hardware; full-frame sensors are already slightly too big for the standard T-2 thread and vignette slightly when used as such. Conversely, I would say that, if bigger were NOT better, modern medium-format cameras would not exist! And, evidently, the results of actually getting one hooked up to a telescope speak for themselves - the website you linked has many absolutely fantastic photos on it that were taken with a medium-format camera, so I wonder what point you were trying to make with it. It seems to me that bigger really IS better.

[pipnina]

People have already explained the technical details quite well here so I'll just add a bit of info that "answers" the question in its own way.

The HST WFC3 has a 45x45mm sensor

JWST's NIRCAM has two groups of four 42x42mm sensors, giving each group a square size of 85x85mm, which is getting close to traditional 4x5 sheet film size. Albeit both of thse telescopes work at optical speeds slower than f20...

[Ags]

The ESA Euclid space telescope has an array of 6x6 CCDs, each with about 4000x4000 12um pixels. That gives it a sensor of about 300x300 mm.

[Steve Ward]

And there's always the LSST ... 😉

https://petapixel.com/2024/04/20/the-biggest-digital-camera-ever-made-6600-pounds-and-3200-megapixels/#:~:text=Each sensor captures a 4%2C000,of capturing 3%2C200-megapixel photos.

[ollypenrice]

I'm afraid I think it's dead simple:

The size of the sensor affects absolutely nothing other than the field of view, but your optics' image circle needs to be able to cover it.

The size of the pixels decides the resolution up to the limit of the seeing. Pixel size and sensor size are , in principle, unrelated.

The noise is determined by the camera's electronics, cooling and software.

The final framing depends on the imager's crop. There will be more to crop on a larger sensor but this is of no consequence. Ironically, I have found over the last 15 years, that I nearly always crop long focal length/small sensor images and nearly always use short focal length/big chip setups for making mosaics. This might strike you as the wrong way round but that's how my imaging works in practice.

Olly

 

[vlaiv]

3 minutes ago, ollypenrice said:

The size of the sensor affects absolutely nothing other than the field of view, but your optics' image circle needs to be able to cover it.

Sensor size affects one very important thing - total amount of light collected. How well is that light utilized - that is something else.

4 minutes ago, ollypenrice said:

The size of the pixels decides the resolution up to the limit of the seeing. Pixel size and sensor size are , in principle, unrelated.

Pixel size and sensor size are directly related by simple equation

sensor size (say width) = number of pixels x width of single pixel

6 minutes ago, ollypenrice said:

The noise is determined by the camera's electronics, cooling and software.

Very important, if not the most important source of noise is shot noise which is determined by "chunk of light" being collected at single sample (in simplest terms - how much light has been collected by any given pixel).

Software on the other hand can be used to manipulate noise or better put SNR - but it does not determine it at the time of capture at all

 

[ollypenrice]

21 minutes ago, vlaiv said:

Sensor size affects one very important thing - total amount of light collected. How well is that light utilized - that is something else.

 

The total amount of light collected is of no consequence at all if it is from parts of the sky which don't interest you. If they do, then yes, but that is the same as saying that sensor size affects FOV - which is what I did say!

21 minutes ago, vlaiv said:

Pixel size and sensor size are directly related by simple equation

 I was disappointed by the equation which followed: I was hoping you were going to go all relativistic on us and prove that pixels got larger or smaller depending on how many neighbours they have!

(I do find it irksome when daytime photography writers use 'resolution' to mean pixel count. Grrr. They should be made to place a wet mackerel in a hat and wear it for twenty minutes on a warm day. And that's being lenient!)

Olly

Edited May 28 by ollypenrice
typo

[vlaiv]

5 minutes ago, ollypenrice said:

The total amount of light collected is of no consequence at all if it is from parts of the sky which don't interest you. If they do, then yes, but that is the same as saying that sensor size affects FOV - which is what I did say!

But you can take larger sensor and make it if with the FOV you are interested in - by changing FL, and if you couple that change in FL with change in aperture size - then you get faster system which larger sensor allows for.

In that sense - it's not just the FOV.

5 minutes ago, ollypenrice said:

 I was disappointed by the equation which followed: I was hoping you were gong to go all relativistic on us and prove that pixels got larger or smaller depending on how many neighbours they have!

I know - but it is important equation to keep in mind - it goes along with above.

Say that you want to capture a galaxy at 1"/px. Also, for the sake of arguments - let that galaxy be 10 arc minutes across - so you really need to capture about 15 arc minutes in width to frame it nicely.

That is only 900 sampling points at 1"/px.

If we take a large sensor that has something like 20000 um in width (that is 20mm in metric ), you can see that you can have up to 20000 / 900 = 22.222um pixel size ... That in turn means ~4500mm of focal length or 20" RC

Very handy equation

Edited May 28 by vlaiv
pressed submit sooner than I've hoped :D