Jump to content

Narrowband

Reducer Myth: On Point


Rodd

Recommended Posts

Well, here we ago again.  No, not really, as I do not really have much of a point other than to say here are data sets perfectly suited to such a comparison.  Both were shot with the TOA 130 and STT-8300 with 3um Ha filter, and both contain 11 30 min subs.  The first image is from the reduced scope (.7x reducer).  The second image is at native focal length with no reducer.  Both images are fully calibrated and aligned, but otherwise unprocessed.  The same STF stretch was applied to both.  It is quite apparent that the reduced stack is much brighter, perhaps as expected.  But I am not seeing much of a resolution boost in the unreduced image.  While the details are larger--they do not seem more defined.  In fact, I think I see more details in the reduced image.  I realize that if a detail is not bright enough to be seen, it makes no difference how much resolution one has.  But I find it hard to believe that is the case here.  The f ratios were F5.4 and F7.7, not F3 and F15.  The one difference that I could not correct was orientation, as they were captured at different camera angles.  This should not matter for empirical analysis--but it may make visual interpretation more difficult.  I originally posted a rotated unreduced image, but rotation changes the scale somehow.

So the question is--if you enlarge the reduced image so that the scale is the same (or bigger) than the unreduced image, are the details more or less apparent?  Or equal.  This is a big question, because if they are at least the same, there is no reason not to use a reducer (other than you don't have one--which is true for the TEC 140 folks).

All thoughts, comments welcome.  

 

Reduced:

5afcbd7cf2ae7_Ha-11reduced.thumb.jpg.253e422e83930a99516182ca52c8f049.jpg

 

Unreduced:

5afcbf2d91ac2_Haunreduced2.thumb.jpg.11cf2c62f98003cc3973f0db7fcce3c6.jpg

Link to comment
Share on other sites

  • Replies 87
  • Created
  • Last Reply
1 hour ago, Rodd said:

I guess we are all talked out on this matter.  

No. Never. :wink:

With .7 reduction, you put twice as much sky on each pixel. That means that with an extended target, each pixel collects double the number of photons during those 30 minutes. But the same amount of camera noise (and dark current). Wouldn't that make the reduced image a better quality?

Link to comment
Share on other sites

54 minutes ago, wimvb said:

No. Never. :wink:

With .7 reduction, you put twice as much sky on each pixel. That means that with an extended target, each pixel collects double the number of photons during those 30 minutes. But the same amount of camera noise (and dark current). Wouldn't that make the reduced image a better quality?

well--according to the theory, the reduced image should be brighter than the reduced image, but at the expense of resolution.  But I am not sure that is the case here.  Zoom in on the reduced image and fine details are more apparent, not less.

Link to comment
Share on other sites

Had a quick play with the images, Rodd (hope you don't mind).

Took me a while to figure how those pics matched until I realised one was rotated at 90 degrees to the other.

So downloaded each picture, rotated one through 90 degrees to get them aligned, reduced the "reducer" image to 70% to compensate for the size difference and extracted a 500*350 pixel chunk centred on roughly the same point, the doubled the size of each.

Clearly the "reducer" image is still considerably brighter, but as to resolution and which one you prefer ... I will leave others to decide ...

no reducer.jpg

reducer.jpg

Link to comment
Share on other sites

I prefer the reduced image as it is brighter and looks more detailed. I don't think there's much difference in the actual resolution of the two images, but I think the improved SNR of the reduced image gives better contrast so making details easier to perceive.

Cheers, Ian.

Link to comment
Share on other sites

1 hour ago, iansmith said:

I prefer the reduced image as it is brighter and looks more detailed. I don't think there's much difference in the actual resolution of the two images, but I think the improved SNR of the reduced image gives better contrast so making details easier to perceive.

Cheers, Ian.

If you are right--and it seems that way, then it flies in the face of accepted theory.  

Rodd

Link to comment
Share on other sites

12 hours ago, Rodd said:

well--according to the theory, the reduced image should be brighter than the reduced image, but at the expense of resolution.  But I am not sure that is the case here.  Zoom in on the reduced image and fine details are more apparent, not less.

Matter of resolution is complicated. True, mathematical / physical resolution should not be measured by eye. Human eye (and brain) perceive image differently based on many parameters.

Increase contrast of the image - it will be perceived as sharper. Add noise to the image and it will be perceived as sharper. We don't even perceive color the same way always, nor the brightness.

Here is example for the brightness:

Whites_illusion.jpg

Both a column and b column gray rectangles have the same level of gray - we see them as being different in intensity / brightness.

Non linear stretch of data alters frequency response for that data (not sure exactly how, and I don't thing there is simple / obvious rule, but I do know that it often causes high frequency harmonics to be added) - again making things sharper than they are.

Way we scale the image has an impact on perceived detail and sharpness - just take regular image and enlarge it 2x2 using different algorithms / filters (nearest neighbor, bilinear, bicubic, mitchell, lanczos) and in each image you will see different amount of "sharpness", and you will be able to judge which one looks the sharpest. You have not added any resolution to this image - yet results look differently sharp. Take another image (different one in brightness, composition and color) and do the same. It is likely that different resampling algorithm will yield sharper result this time. Again you have not added any resolution to the image.

Having said all of this, I have an idea (exploiting nature of human eye) how I can make obvious tiny differences in resolution between these two images.

Could you please post original stacks (unaltered) so I can have a go at it?

Link to comment
Share on other sites

8 hours ago, Demonperformer said:

Clearly the "reducer" image is still considerably brighter, but as to resolution and which one you prefer ... I will leave others to decide ...

Don't mind at all--It seems to me that the reduced image shows more detail, though that may be an illusion created by greater contrast.  Still--its close enough for doubt.

Rodd

Link to comment
Share on other sites

6 minutes ago, vlaiv said:

Matter of resolution is complicated. True, mathematical / physical resolution should not be measured by eye. Human eye (and brain) perceive image differently based on many parameters.

Increase contrast of the image - it will be perceived as sharper. Add noise to the image and it will be perceived as sharper. We don't even perceive color the same way always, nor the brightness.

Here is example for the brightness:

Whites_illusion.jpg

Both a column and b column gray rectangles have the same level of gray - we see them as being different in intensity / brightness.

Non linear stretch of data alters frequency response for that data (not sure exactly how, and I don't thing there is simple / obvious rule, but I do know that it often causes high frequency harmonics to be added) - again making things sharper than they are.

Way we scale the image has an impact on perceived detail and sharpness - just take regular image and enlarge it 2x2 using different algorithms / filters (nearest neighbor, bilinear, bicubic, mitchell, lanczos) and in each image you will see different amount of "sharpness", and you will be able to judge which one looks the sharpest. You have not added any resolution to this image - yet results look differently sharp. Take another image (different one in brightness, composition and color) and do the same. It is likely that different resampling algorithm will yield sharper result this time. Again you have not added any resolution to the image.

Having said all of this, I have an idea (exploiting nature of human eye) how I can make obvious tiny differences in resolution between these two images.

Could you please post original stacks (unaltered) so I can have a go at it?

Here they are.  Keep in mind that the above stacks are unaltered except for identical stretches--that is one aspect of the STF tool in PI that can be useful, it can ensure that 2 images are stretched exactly the same.  But all that you say, while true, brings this issue into the realm of a loss that is really a win (ie a college team scrimmages a pro team and almost wins--losing in the last seconds of the game).  That is a moral victory (Not sure why morality has been attached to this saying, but that is the saying).  Anyway--in the end, it is the human eye that perceives and judges/admires such images.  If the difference in resolution between a reduced image and an unreduced image requires lengthy explanations and optical illusions to see, then that is close enough for me to say that there is no appreciable (even noticeable) difference.  It means that one can use a reducer to collect data faster (or more data in the same amount of time), crop the extra unwanted FOV and have an image of similar detail that was collected faster than using an unreduced system.  And that is contrary to the myth.  And as I said--even if it turns out that it is not quite as detailed--if its close enough to require hard inspection, its close enough for general purposes (The moral victory).

Ha-11 reduced.fit

Ha-11-unreduced.fit

Rodd

 

Link to comment
Share on other sites

I did a brief comparison of a small part of of the image (the thumb shaped pillar) and I felt that the unreduced image was sharper and showed the detail more clearly. Did it show additional detail? I don't think so, but perhaps that's because there isn't the additional detail in that part of the object to show? I agree that the reduced image is brighter, appears to show more detail,  and is aesthetically more pleasing, but close up, I reckon there is sharper detail in the unreduced. 

PS: I looked at the first images in the post, not the TIFFs.

Link to comment
Share on other sites

18 minutes ago, RobertI said:

I did a brief comparison of a small part of of the image (the thumb shaped pillar) and I felt that the unreduced image was sharper and showed the detail more clearly. Did it show additional detail? I don't think so, but perhaps that's because there isn't the additional detail in that part of the object to show? I agree that the reduced image is brighter, appears to show more detail,  and is aesthetically more pleasing, but close up, I reckon there is sharper detail in the unreduced. 

PS: I looked at the first images in the post, not the TIFFs.

Not sure that is what I see in the example above (the rotated and enlarged section).  Again, at some point, even if you are right, if the difference is small enough.....MV!

Rodd

Link to comment
Share on other sites

6 minutes ago, Rodd said:

Here they are.  Keep in mind that the above stacks are unaltered except for identical stretches--that is one aspect of the STF tool in PI that can be useful, it can ensure that 2 images are stretched exactly the same.  But all that you say, while true, brings this issue into the realm of a loss that is really a win (ie a college team scrimmages a pro team and almost wins--losing in the last seconds of the game).  That is a moral victory (Not sure why morality has been attached to this saying, but that is the saying).  Anyway--in the end, it is the human eye that perceives and judges/admires such images.  If the difference in resolution between a reduced image and an unreduced image requires lengthy explanations and optical illusions to see, then that is close enough for me to say that there is no appreciable (even noticeable) difference.  It means that one can use a reducer to collect data faster (or more data in the same amount of time), crop the extra unwanted FOV and have an image of similar detail that was collected faster than using an unreduced system.  And that is contrary to the myth.  And as I said--even if it turns out that it is not quite as detailed--if its close enough to require hard inspection, its close enough for general purposes (The moral victory).

Ha-11 reduced.fit

Ha-11-unreduced.fit

Rodd

 

I don't dispute much of what you said here. As I already emphasized in another thread dealing with resolution - depend on what you want out of resolution. There are different approaches to the resolution. One is scientific - lets you measure things more accurately, the other is "esthetic" - let's you produce nicer images.

If you are not concerned with measuring things, then by all means - use focal reducer, it will help you get better SNR in shorter amount of time (just how much, depends on LP - so narrow band imaging clearly has an edge in this over broadband in all but pristine skies) at a minimal / imperceptible (to the human eye) loss of resolution.

What I do object is calling theory flawed or myth based on nonscientific conclusions.

Too many times I've read in these forums when discussing theoretical side of things something along the lines "Yes, it might be theoretically so but in practice ...". Theory can differ from the practice only on two grounds: First - theory is wrong, then it must be abandoned and replaced with another one that fits observational data, or second: There is systematic error in measurement / theory is being applied out of realm of what it's explaining (not all aspects have been taken into account when matching practice and theory).

Anyway I believe we are ought to strive to have theory match the practice, either by correcting our theory or making sure we are applying theory to practice in correct way - that will just broaden and deepen our understanding and ultimately bring better results to the community as a whole (even to people that just enjoy other aspects of this hobby and are not concerned with much of theoretical background).

"If the difference in resolution between a reduced image and an unreduced image requires lengthy explanations and optical illusions to see ..." - I probably did not express myself properly. All I wanted to do is create animated gif (out of couple of images processed linearly as to not loose resolution) because it helps eye/brain see difference more easily than comparing images side by side. Eye is more sensitive to motion - people observing DSOs often use this trick - tapping slightly OTA to make it shake a bit - it lets you spot faint DSO more easily.

Link to comment
Share on other sites

3 minutes ago, ollypenrice said:

Nobody disputes the usefulness of reducers on a field-filling target such as this. The myth concerns itself with small targets entirely covered by the chip with or without reducer.

Olly

But olly--what is the difference between a cropped section of this image, and say a smaller target that does not cover the chip--once you crop out a section, its the same....no?  Once you crop out, say a certain section--that certain section did not cover the entire chip and is synonymous with--a galaxy for instance.

Rodd

Link to comment
Share on other sites

4 minutes ago, Rodd said:

Not sure that is what I see in the example above (the rotated and enlarged section).  Again, at some point, even if you are right, if the difference is small enough.....MV!

Rodd

I agree! But when I examined the originals you posted, I saw something different. It could be that the viewer I am using is doing its own 'smoothing' of the image? I'll have a look at the TIFFs later and see what I get.

Link to comment
Share on other sites

6 minutes ago, vlaiv said:

"If the difference in resolution between a reduced image and an unreduced image requires lengthy explanations and optical illusions to see ..." - I probably did not express myself properly. All I wanted to do is create animated gif (out of couple of images processed linearly as to not loose resolution) because it helps eye/brain see difference more easily than comparing images side by side. Eye is more sensitive to motion - people observing DSOs often use this trick - tapping slightly OTA to make it shake a bit - it lets you spot faint DSO more easily.

Let me rephrase my statement as well.  If it requires ANY explanation at all....its too much.  You go to a museum and look at a replica of a masterpiece....if it requires a docent with a magnifying glass to point out the differences....are there really any differences at normal viewing conditions?

Link to comment
Share on other sites

48 minutes ago, Rodd said:

Let me rephrase my statement as well.  If it requires ANY explanation at all....its too much.  You go to a museum and look at a replica of a masterpiece....if it requires a docent with a magnifying glass to point out the differences....are there really any differences at normal viewing conditions?

Well, again it depends whether one requires certain explanation over a certain domain. You have two glasses, they are filled to the same level (indistinguishable difference), liquid looks the same, has the same transparency - for purpose of representing glass of water on the image, both will do fine.

But then you examine both liquids - you find that one is indeed water, other is alcohol, they have different density - because of density they refract light in different way, you notice that background which you view thru the glass with liquid is stretched  differently in one image compared to the other (very slight difference). Comparing two images next to each other, hardly anyone can spot the difference, and all who spot the difference agree that it is not important difference at such a small level.

But it turns out that such a small difference lets you determine if glass actually contains water or alcohol.

Link to comment
Share on other sites

1 minute ago, vlaiv said:

Well, again it depends whether one requires certain explanation over a certain domain. You have two glasses, they are filled to the same level (indistinguishable difference), liquid looks the same, has the same transparency - for purpose of representing glass of water on the image, both will do fine.

But then you examine both liquids - you find that one is indeed water, other is alcohol, they have different density - because of density they refract light in different way, you notice that background which you view thru the glass with liquid is stretched  differently in one image compared to the other (very slight difference). Comparing two images next to each other, hardly anyone can spot the difference, and all who spot the difference agree that it is not important difference at such a small level.

But it turns out that such a small difference lets you determine if glass actually contains water or alcohol.

True--but for the purposes of looking at the glass--not what is behind the glass--just the glass--or IMAGE itself, they are the same.  They, of course, are NOT the same, but we are talking about how they appear, which is what AP is all about when viewing images without the aide of analytical technology.

Rodd

Link to comment
Share on other sites

1 hour ago, Rodd said:

But olly--what is the difference between a cropped section of this image, and say a smaller target that does not cover the chip--once you crop out a section, its the same....no?  Once you crop out, say a certain section--that certain section did not cover the entire chip and is synonymous with--a galaxy for instance.

Rodd

That's what I was wondering, and why I cut out a piece covering the same portion (and importantly size) of the object.

Link to comment
Share on other sites

2 minutes ago, Demonperformer said:

That's what I was wondering, and why I cut out a piece covering the same portion (and importantly size) of the object.

Precisely...that is the crux of the myth and the point behind the post.  Thank you

Rodd

Link to comment
Share on other sites

2 minutes ago, Rodd said:

Precisely...that is the crux of the myth and the point behind the post.  Thank you

Rodd

Wait, can we actually describe the "myth" for my benefit?

If I'm understanding correctly, "myth" in this case was: Using focal reducer reduces resolution of the acquired image compared to obtaining the same image without focal reducer?

And by giving this example you are in fact confirming that above statement is a myth?

 

Link to comment
Share on other sites

4 minutes ago, vlaiv said:

Wait, can we actually describe the "myth" for my benefit?

If I'm understanding correctly, "myth" in this case was: Using focal reducer reduces resolution of the acquired image compared to obtaining the same image without focal reducer?

And by giving this example you are in fact confirming that above statement is a myth?

 

More specifically the myth states that if you are imaging an object that does not fill the sensor--like a galaxy--or a portion of an image, using a reducer will not allow you to crop the image later and have a image with more signal or detail than if you did not use the reducer and compared it to an uncropped image.  I know that sound convoluted, I am typing fast.  Someon else can state it better I am sure.

Link to comment
Share on other sites

Just now, Rodd said:

More specifically the myth states that if you are imaging an object that does not fill the sensor--like a galaxy--or a portion of an image, using a reducer will not allow you to crop the image later and have a image with more signal or detail than if you did not use the reducer and compared it to an uncropped image.  I know that sound convoluted, I am typing fast.

Ok, I'll state my view on that:

Using focal reducer will let you shoot the same object, which when cropped after will provide the same FOV as image taken without focal reducer. It will also provide you with higher SNR for the same acquisition parameters (sub time, number of subs) and same equipment (camera) compared to imaging without focal reducer (just how much SNR gain there will be, depends on number of other factors). Only part that I disagree with is that you will end up with same or more detail in the case of focal reducer vs native.

I am assuming that focal reducer does not act as corrective element (field flattening, coma correction, etc are out of the scope of this statement) and that it purely reduces focal length. I'm also assuming (as stated above) that all other parameters are the same - same camera, same sky conditions, ...

There is border line case where both using focal reducer and native will provide the same level of detail and that is if with both focal lengths, reduced and native, sampling resolution is higher than critical sampling for a given scope. In all other cases using focal reducer will indeed result in lower resolution over native focal length. How much lower? It also depends on number of factors, like sampling rate on both focal lengths (reduction factor of focal reducer), aperture size and sky conditions.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.