The crop myth.

[ollypenrice]

Although nobody experienced in imaging would fall for it, I wonder if nailing 'The crop myth' might be helpful for beginners? I have just seen it appear yet again this evening and it is perpetuated by the daytime photography term 'crop factor.' This term is not entirely invalid because in daytime photography there is the question of the relationship between distance and apparent size: the daytime photographer might, quite reasonably, want to use a system which gave the same distance/apparent size ratio as that perceived by the eye. But in astronomy we are shooting at infinity and this is not the same.

Here's the Horse and Flame.

And here's the Horse:

The second image has more Horse detail? No, they are precisely the same image but the second is cropped. The Horse is identical in both. Zoom in on the upper image and it has exactly the same resolution as the lower one.

Your PC, by default, will probably fit an image to your screen so, if it is cropped, the Horse may fill the screen. If it is not cropped the PC will shrink the entire image to fit the screen but, if you zoom in, you'll find that the level of Horse detail is identical.

When we see that we are looking at an image at '100%' or 'Full size,' what does that mean? It means that one camera pixel is given one screen pixel. And that is absolutely unaffected by cropping.

To those who've been imaging for a while all this is dead easy but, if you're new to it, I'm not sure that it is. In a nutshell cropping brings no new detail.

Olly

 

 

 

 

Edited June 27, 2019 by ollypenrice

[vlaiv]

Not sure what the myth is all about - but I think that shooting at infinity is not quite a problem when discussing crop factor - if I understand it correctly.

In my view it is more related to angle than size on sensor? Angle also holds true for infinity shots - smaller sensor will have smaller TFOV, and you can define ratio of full frame sensor and any particular sensor using certain scope - and you will get a "crop factor" that way.

Of course there are other ways to think about image size in AP, but size of sensor is sometimes very useful (I tend to forget that from time to time) - especially with latest CMOS sensors where read noise is small and one can afford all sorts of binning in software.

[JamesF]

Has your internet broken again Olly?  No image attached that I can see.

If you're talking about the often-repeated assertion that a 50mm lens is actually an 80mm lens when used on a 1.6x crop factor sensor then it frustrates me no end.  I have posted numerous times that a 50mm lens is a 50mm lens is a 50mm lens and nothing whatsoever is suddenly going to make it an 80mm lens.  What changes is how much of the field of view is captured on the sensor, not anything to do with the lens, and to suggest otherwise is hugely misleading.

James

[-Joe_]

I would say it relates to the 35mm film. A 50mm lens provides a certain fov on a 35mm film frame (A full frame). Those numbers have ended where they are due to the size of the human eye.

”crop factor” is just a way of relating back to that when trying to sell digital cameras with smaller than 35mm sensors.

What really matters is fov, what pixels you have in that fov, and how many photons are arriving.

i may be wrong (normally am)...

[-Joe_]

I guess perspective also comes into it with ‘normal’ photography. As well as fov, how far away you are is a major artistic decision. Not one we can change much with astrophotography.

[ollypenrice]

On 28/06/2019 at 11:03, -Joe_ said:

 

”crop factor” is just a way of relating back to that when trying to sell digital cameras with smaller than 35mm sensors.

 

Yes, but there is a lurking potential confusion: the term 'crop factor' is innocent enough if you want know what lens for an APSc chip will give the same field as 'lens x' on a 35mm chip. The danger is that it can lead folks to believe that reducing the chip size is somehow equivalent to increasing the focal length, when it isn't. An experienced imager wouldn't make this mistake but in posts in the beginners' section it is quite often made.

Olly

[Datalord]

Let me add some fuel to the fire. The "crop" factor is "real" in terms of FoV, but the reason it is so massively misleading is that it comes at the cost of lowered "/pp, which artificially puts more pressure on the mount and guiding which stresses the entire data gathering.

This is the caveat I didn't realise until much later and that for a beginner is very difficult to fathom. Both the resolution, the importance of it and the mount requirements it ultimately poses.

[vlaiv]

1 minute ago, Datalord said:

Let me add some fuel to the fire. The "crop" factor is "real" in terms of FoV, but the reason it is so massively misleading is that it comes at the cost of lowered "/pp, which artificially puts more pressure on the mount and guiding which stresses the entire data gathering.

This is the caveat I didn't realise until much later and that for a beginner is very difficult to fathom. Both the resolution, the importance of it and the mount requirements it ultimately poses.

I think that this was exactly what @ollypenrice tried to explain with this post - "crop factor" has nothing to do with "/pixel.

Resolution in terms of "/pixel is defined only by focal length and pixel size. Sensor size is not directly related to this - there are small sensors with large pixels and large sensors with small pixels (and anything in between).

I initially misunderstood the post because it lacked attached images for some reason, so I did not pay attention and I thought that we are talking of crop factor in terms of FOV / lens rather than cropping image and thinking it will somehow change "/pixel - give one more "zoom".

This topic actually touches upon a subject that I wanted to start discussion on for some time now - difference between "recording" resolution and "presentation" resolution. In principle you can record on higher resolution (less "/pixel) but still "present" image at proper sampling/resolution given conditions image was taken in (seeing and even guiding). In this case "sub optimal" guiding is not such a big deal - one only has to accept that they will be able to do images of certain detail scale (what we call resolution), even if one has scope/camera that is capable of recording finer detail scale in principle (higher resolution).

I think I've develop very good "fractional" binning algorithm that would facilitate actually choosing presentation, or work resolution based on measurements of the image.

[pete_l]

7 hours ago, vlaiv said:

This topic actually touches upon a subject that I wanted to start discussion on for some time now - difference between "recording" resolution and "presentation" resolution. In principle you can record on higher resolution (less "/pixel) but still "present" image at proper sampling/resolution given conditions image was taken in (seeing and even guiding). In this case "sub optimal" guiding is not such a big deal - one only has to accept that they will be able to do images of certain detail scale

Yes. I think this is a more relevant issue. It amuses me when we see (for example) 6MP images displayed on a standard HD monitor of only 2MPix. We are fortunate that 4K monitors allow us to display 10 MPix images natively. But when you sit at a normal distance for computer work (much closer than one would watch a TV programme) it is not easy to take in the entire view.
My "work" screen is 107cm wide. At 3840 pixels that is 36px/cm or 90 DPI. I get perfectly good images displayed on that, despite only sitting 60-80 cm away. Yet mobile phone screens make a big deal of "retina" displays with much higher resolutions. I assume the users hold their phones very close to their eyes.

I recall having a conversation with a "professional" photographer many years ago. They insisted that all photos had to be printed at 300 DPI. No matter their size. No matter how far away they would be viewed from. I feel that if they revisited the subject now, with a more accepting mind, they would see (!) the error of their ways.

[Demonperformer]

On 27/06/2019 at 23:01, ollypenrice said:

Your PC, by default, will probably fit an image to your screen so, if it is cropped, the Horse may fill the screen. If it is not cropped the PC will shrink the entire image to fit the screen but, if you zoom in, you'll find that the level of Horse detail is identical.

When we see that we are looking at an image at '100%' or 'Full size,' what does that mean? It means that one camera pixel is given one screen pixel. And that is absolutely unaffected by cropping.

I agree with what you are saying, but my "astro-snapshot" approach is slightly different. My screen is only 1280x7-something. Having an image that is any bigger than that is pointless because, as you say, the computer will automatically shrink it. So I will tend to either crop the image for framing (my images don't support the use of the word "artistic") reasons, or to resize it, so that I end up with a file that will display "at 100%" on my screen. This file tends to be considerably smaller than the one I get out of the camera, so saves me storage space on the disk and (if anyone should be foolhardy enough to want a copy!) bandwidth (mine and theirs). It also saves space on the SGL servers if I choose to upload it here. Anything bigger is just a waste of resources as far as I am concerned.

[StuartJPP]

The thing is that if you have a 24MP full frame sensor and a 24MP "crop" sensor there will be more pixels on the Horse and hence more detail (all else being equal).

Making the sensor smaller without making the pixels smaller as well just gives you a smaller FOV with the same detail.

[ollypenrice]

52 minutes ago, Demonperformer said:

I agree with what you are saying, but my "astro-snapshot" approach is slightly different. My screen is only 1280x7-something. Having an image that is any bigger than that is pointless because, as you say, the computer will automatically shrink it. So I will tend to either crop the image for framing (my images don't support the use of the word "artistic") reasons, or to resize it, so that I end up with a file that will display "at 100%" on my screen. This file tends to be considerably smaller than the one I get out of the camera, so saves me storage space on the disk and (if anyone should be foolhardy enough to want a copy!) bandwidth (mine and theirs). It also saves space on the SGL servers if I choose to upload it here. Anything bigger is just a waste of resources as far as I am concerned.

I think there's a place for an image which is way too big to fit the screen but which can be 'cruised around' while zoomed in on. Well, I hope there is since Tom and I spent 400 hours capturing one!

 

12 minutes ago, StuartJPP said:

The thing is that if you have a 24MP full frame sensor and a 24MP "crop" sensor there will be more pixels on the Horse and hence more detail (all else being equal).

Making the sensor smaller without making the pixels smaller as well just gives you a smaller FOV with the same detail.

Yes, the point of my post was simply to say that you can't 'zoom in' on a target by using a smaller sensor, a misconception which comes up fairly regularly. You can, of course, do so by using smaller pixels until other factors limit the real resolution possible.

Olly

[alacant]

Phew. A Spanish mate just called asking me to explain this thread. Ahgghh!

We have 24mp and a 12mp dslrs. The 12mp is better. Is that it?

Cheers and clear skies from a suffocating Alicante.

[Craney]

With this crop, the more you zoom in, the more detail you get..... 

[andrew s]

1 hour ago, ollypenrice said:

I think there's a place for an image which is way too big to fit the screen but which can be 'cruised around' while zoomed in on. Well, I hope there is since Tom and I spent 400 hours capturing one!

 

Yes, the point of my post was simply to say that you can't 'zoom in' on a target by using a smaller sensor, a misconception which comes up fairly regularly. You can, of course, do so by using smaller pixels until other factors limit the real resolution possible.

Olly

Quite right. Another related issue is digital zoom v optical zoom. Discuss.

😉

Regards Andrew 

[ollypenrice]

4 hours ago, alacant said:

Phew. A Spanish mate just called asking me to explain this thread. Ahgghh!

We have 24mp and a 12mp dslrs. The 12mp is better. Is that it?

Cheers and clear skies from a suffocating Alicante.

If both cameras have the same sized chip then this isn't within the remit of the thread - but that's not a problem. I just want things to be clear. Assuming the chips are the same size then, clearly, one has much smaller pixels than the other. Is this of itself good or bad? It could be either, depending on the camera's performance and its suitability for the optics in which it's used. There is no point in using pixels which are so small that they are trying to resolve details which will be be blurred out by the seeing. You might get the same real resolution from the larger pixel camera, and more light per pixel, which would be an advantage. But if you were using a very short FL, as in a camera lens, you might be glad of the smaller pixels' ability to render finer details.

I think folks are trying to find in my original post complexities which are simply not there. It really is a simple point: if we take a small chip camera and a large chip camera, both with the same pixel size, the Horsehead might fill the frame on the small chip and only quarter fill it on the large - but the size of the image of the Horsehead as captured, and the details it contains, are identical in both. A smaller chip does not 'get you closer' because as soon as you look at the Horse at 100% in both you'll see that they are identical. On the day I posted, a beginner wanting to image the planets was advised that to get a larger image of a planet you'd need a smaller chip. But no, the only way to get a larger image of the planet is to get a longer focal length or smaller pixels or both. (Of course in fast frame imaging larger chips need longer download times and can't reach the fastest frame rates, all being equal, but that's a separate matter.)

Hot here, too, and the telly's stopped working on day one of the Tour de France. This is a poor do...

Olly

Edited July 6, 2019 by ollypenrice

[Datalord]

16 minutes ago, ollypenrice said:

I think folks are trying to find in my original post complexities which are simply not there.

But there are. The reason we all start running into the weeds is that pixel size and focal length has so much impact on the precision you need from your mount, which is exactly the dark magic no beginner understands. They (and I include my own 3 years past self in this) focus on the focal length, the camera chip size and aperture because that's what translates from the "real" world.

If we want beginners to have a better chance at pairing, we have to make it easy to comprehend what choices in scope and camera means to the mount. 

[ollypenrice]

27 minutes ago, Datalord said:

But there are. The reason we all start running into the weeds is that pixel size and focal length has so much impact on the precision you need from your mount, which is exactly the dark magic no beginner understands. They (and I include my own 3 years past self in this) focus on the focal length, the camera chip size and aperture because that's what translates from the "real" world.

If we want beginners to have a better chance at pairing, we have to make it easy to comprehend what choices in scope and camera means to the mount. 

My intention was to deal with just one variable in isolation, the size of the chip, and to point out that, of itself, it has no bearing on how 'zoomed in' we are. That's all. I think it important to understand this before going on to image scale.

Image scale has been discussed at length elsewhere, when the variables are pixel size and focal length. Indeed I did an AN article on that subject a while ago and concluded that a shorter focal length/smaller aperture/smaller pixel setup holds up remarkably well (though not completely) against a longer FL/larger aperture/larger pixel rig.  Yes, this is indeed the key thing to understand in deep sky AP but it wasn't my topic in this thread.

Olly

[Demonperformer]

21 hours ago, ollypenrice said:

I think there's a place for an image which is way too big to fit the screen but which can be 'cruised around' while zoomed in on. Well, I hope there is since Tom and I spent 400 hours capturing one!

But your images are in a slightly different league to mine.

[MilwaukeeLion]

I like cropping elongated stars out of my AVX captures.  Yes this confused me early on especially using Stellarium.  Switching between cameras the selected object in Stellarium would be different sizes in the frame leading me believe one camera would give me greater FL . 

[ollypenrice]

11 minutes ago, MilwaukeeLion said:

I like cropping elongated stars out of my AVX captures.  Yes this confused me early on especially using Stellarium.  Switching between cameras the selected object in Stellarium would be different sizes in the frame leading me believe one camera would give me greater FL . 

Yes, that's all I was getting at.

Another simple way to put it: On how many pixels does the object's projected image land? 

Olly