Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

How Big does a Target need to be on a Sensor before you feel you "Got It"?

  • Please log in to reply
21 replies to this topic

#1 GraySkies

GraySkies

    Ranger 4

  • -----
  • topic starter
  • Posts: 381
  • Joined: 25 Feb 2011
  • Loc: Ontario, Canada

Posted 24 April 2019 - 11:11 AM

So an interesting philosophical question for all you imagers...

 

The sky is full of faint fuzzies and given different cameras focal lengths and sensor sizes etc... if you decide to image say a galaxy, how many pixels (diameter or longest edge) do you feel you should get at a minimum before you consider it "captured", or is there another factor you go by?

 

I'm asking as I'm looking to do a Messier Object Challenge and image all 110 (Similar to This with my initial results here) but I've been debating the minimum dimension requirements as I restart the challenge.

There is also the question of how deep to go as well, but that's a discussion for another day...


Edited by GraySkies, 24 April 2019 - 11:12 AM.

  • elmiko likes this

#2 jdupton

jdupton

    Surveyor 1

  • *****
  • Posts: 1776
  • Joined: 21 Nov 2010
  • Loc: Central Texas, USA

Posted 24 April 2019 - 12:13 PM

GraySkies,

 

   That is an interesting question. I'm sure you will get a full range of answers. For me, I use two different criteria.

 

   If the object in question is going to be the main subject of the image I am capturing, I want to get at least 100+ pixels across it to see some structure and details. More often than not, I shoot for objects that can fill at least 25% of my field of view. Since I like galaxies and planetary nebulae most, I went for relatively long focal length for my imaging system (a Celestron 11" EdgeHD; often used with the EdgeHD 0.7X focal reducer) in order to get the plate scale and field I felt I needed.

 

   The second criteria comes into play when I am simply trying to identify objects in an image. I like searching out just how faint and far away the objects I see in my images really are -- how deep did I get, so to speak. I consider an object "detected" if it subtends at least three or four pixels in my image. That corresponds to a size of about 1.5 to 2 arc-seconds at my usual plate scale. 

 

 

John

 

[Edit] PS: After seeing Alex's post below, I see I totally misunderstood your question! I have never done an imaging marathon so have nothing to draw on, sorry.


Edited by jdupton, 24 April 2019 - 08:25 PM.


#3 Alex McConahay

Alex McConahay

    Cosmos

  • *****
  • Posts: 7976
  • Joined: 11 Aug 2008
  • Loc: Moreno Valley, CA

Posted 24 April 2019 - 08:17 PM

>>>>>>> minimum before you consider it "captured", or is there another factor you go by

 

I don't understand your question. Your example(s)  are about how you are displaying you images, not about how you are capturing them.  

 

Check out my example of what you are talking about.

 

 http://alexastro.com...ssierindex.html

 

At the top is a big poster I made from getting all of the Messiers I got in one night. But below that are other versions of the Messiers taken with much more care over the years. The original pictures of either version were the same size (within reason, they were taken with different scopes and different cameras, but all were the central subject of a full set of exposures). All of them, within the limits of composition could have been full pictures. It was only that I wanted the whole "all-110-in-one-night" thing that I shrunk them and pasted them together. 

 

In short, you take them at full size, and then shrink them as needed. And what you will need depends on how you are presenting them. 

 

So, as long as they are somewhat recognizable as the object (with the exception of that imposter----M40) it counts. But your artistic vision has to decide how big that is. 

 

Alex


Edited by Alex McConahay, 24 April 2019 - 08:20 PM.

  • DSO_Viewer likes this

#4 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23350
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 24 April 2019 - 08:38 PM

So an interesting philosophical question for all you imagers...

 

The sky is full of faint fuzzies and given different cameras focal lengths and sensor sizes etc... if you decide to image say a galaxy, how many pixels (diameter or longest edge) do you feel you should get at a minimum before you consider it "captured", or is there another factor you go by?

 

I'm asking as I'm looking to do a Messier Object Challenge and image all 110 (Similar to This with my initial results here) but I've been debating the minimum dimension requirements as I restart the challenge.

There is also the question of how deep to go as well, but that's a discussion for another day...

My opinion:

 

When it comes to galaxies, I want them as large as I can get them, without over-filling the frame such that they don't feel like they are out there hanging in pace. That means you need some space around the galaxy to portray that feeling, but not so much that the galaxy starts to feel like another little blob in the field. 

 

The same would apply to any small object. Globs. Planetary nebula. Even small groups of galaxies. I would want them all to be large enough to be as detailed as possible, but still have that sense of hanging in space. 


  • psandelle, Astrola72, DSO_Viewer and 1 other like this

#5 Alex McConahay

Alex McConahay

    Cosmos

  • *****
  • Posts: 7976
  • Joined: 11 Aug 2008
  • Loc: Moreno Valley, CA

Posted 24 April 2019 - 10:33 PM

>>>>>>> I want them as large as I can get them, without over-filling the frame

 

I'd say that is close to true of any picturesque galaxy. (and even then, a few field stars, and maybe something else to tell a story might be nice in the frame. )

 

But there are some pretty plain galaxies out there. 

 

But, really, that is an artistic decision.

 

For instance, M86 is a pretty plain, boring galaxy. However, put it together with M84 and a few other companions, it gets pretty interesting. M86 filling the frame---not so much. M86 as an eye on a face-----cool. 

 

Alex


  • pinzmann and psandelle like this

#6 t_image

t_image

    Gemini

  • -----
  • Posts: 3086
  • Joined: 22 Jul 2015

Posted 25 April 2019 - 05:58 AM

So an interesting philosophical question for all you imagers...

 

The sky is full of faint fuzzies and given different cameras focal lengths and sensor sizes etc... if you decide to image say a galaxy, how many pixels (diameter or longest edge) do you feel you should get at a minimum before you consider it "captured", or is there another factor you go by?

 

I'm asking as I'm looking to do a Messier Object Challenge and image all 110 (Similar to This with my initial results here) but I've been debating the minimum dimension requirements as I restart the challenge.

There is also the question of how deep to go as well, but that's a discussion for another day...

I presume that with "given different cameras focal lengths and sensor sizes etc." you understand the concept of spatial resolution.

Since your question asks about pixel resolution of object (as far as capture), you are in other words asking about sampling.

But IMO the question focuses too narrowly on capture which has biased the responses.

 

It is also just as important to consider the entire workflow from capture to delivery.

 

  • delivery method should determine sampling requirement

Because, your delivery method (the sampling resolution level of display) is the age-old thing to evaluate in determining your sampling requirement for best efficiency.....

Posting as a low-res image on some social media site would mean something entirely different than preparing for 8K IMAX presentation.....

So you might want to work backwards to determine capture sampling...

 

Regardless of whether or not people have thought this through,

I guarantee tested it will play out to be true. I also would predict those that have a particular notion of criteria also consistently have the same delivery method for the images they evaluate...

 

So although the concept of pixel resolution is a good place to start, (ex:as in an AI image recognition database would have to have a minimum for accurate differentiation),

the context of your real-world application would best also take into account how you desire to display these images.....

 

  • A final thing to consider (as the concept of cropping to frame the images together neatly is established),is consistency of look.

This may be somewhat solved by keeping the pixel count resolution (diameter) of the objects the same,

but it may be good strategy to evaluate the smallest target and start with the minimum requirement(pixels) of what it would demand.....

Then any over-sampling with the larger objects can be matched with image processing as you prepare your layout....

Otherwise the crisp fine resolution of a larger object matched in grid with an object that was small but is cropped so that it is all jagged/pixelated would be a poorer presentation of the effort....


  • GraySkies likes this

#7 GraySkies

GraySkies

    Ranger 4

  • -----
  • topic starter
  • Posts: 381
  • Joined: 25 Feb 2011
  • Loc: Ontario, Canada

Posted 25 April 2019 - 11:09 AM

I presume that with "given different cameras focal lengths and sensor sizes etc." you understand the concept of spatial resolution.

Since your question asks about pixel resolution of object (as far as capture), you are in other words asking about sampling.

But IMO the question focuses too narrowly on capture which has biased the responses.

 

It is also just as important to consider the entire workflow from capture to delivery.

 

  • delivery method should determine sampling requirement

Because, your delivery method (the sampling resolution level of display) is the age-old thing to evaluate in determining your sampling requirement for best efficiency.....

Posting as a low-res image on some social media site would mean something entirely different than preparing for 8K IMAX presentation.....

So you might want to work backwards to determine capture sampling...

 

Regardless of whether or not people have thought this through,

I guarantee tested it will play out to be true. I also would predict those that have a particular notion of criteria also consistently have the same delivery method for the images they evaluate...

 

So although the concept of pixel resolution is a good place to start, (ex:as in an AI image recognition database would have to have a minimum for accurate differentiation),

the context of your real-world application would best also take into account how you desire to display these images.....

 

  • A final thing to consider (as the concept of cropping to frame the images together neatly is established),is consistency of look.

This may be somewhat solved by keeping the pixel count resolution (diameter) of the objects the same,

but it may be good strategy to evaluate the smallest target and start with the minimum requirement(pixels) of what it would demand.....

Then any over-sampling with the larger objects can be matched with image processing as you prepare your layout....

Otherwise the crisp fine resolution of a larger object matched in grid with an object that was small but is cropped so that it is all jagged/pixelated would be a poorer presentation of the effort....

It's a bit of both pixel sampling and spatial resolution...

Right now I have 24MP (FF) and 16MP (MFT) cameras for astroimaging, from there I have 3 Astrograph Telescopes at 250, 436mm, 2023mm focal range with respective optical trains. So I have various combinations in order to size my targets with obvious realities that the longer focal length telescopes are slower and the MFT (full spectrum) is significantly more noisy than the FF (unmodded) camera.

So for say a given galaxy, I can image it where it's 100 pixels wide, some detail, or I can image at ~200 pixels wide more detail & noise. or I can Image 500-1000 pixel wide but it will take 10x a long to do so.

Plus as you pointed out, with a wider field I can image 2-3 or in some cases more M/NGC objects at once with a really wide field. This is what I did with my RedCat51, I imaged M51 AND M63 in the same frame, both are tiny (M63 is ~150 pixels wide on the long end) but there is still quite a bit of detail in each and I can easily cover all the Messier objects like this and display them as 250x250 icon squares in a mosaic of all 110 for a total of 2750x2500 pixel final image (if I did it in as close to square) which is ~7MP image so that's suitable for web/small print.

* I also have a 58mm noctural lens, but I haven't fully tested at what f-stop is it truely astrograph quality


Edited by GraySkies, 25 April 2019 - 11:10 AM.


#8 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 23350
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 25 April 2019 - 11:29 AM

>>>>>>> I want them as large as I can get them, without over-filling the frame

 

I'd say that is close to true of any picturesque galaxy. (and even then, a few field stars, and maybe something else to tell a story might be nice in the frame. )

 

But there are some pretty plain galaxies out there. 

 

But, really, that is an artistic decision.

 

For instance, M86 is a pretty plain, boring galaxy. However, put it together with M84 and a few other companions, it gets pretty interesting. M86 filling the frame---not so much. M86 as an eye on a face-----cool. 

 

Alex

I think it depends on how you capture it. wink.gif

 

https://apod.nasa.go...Hanson_1143.jpg

 

And what I said earlier goes for groups of galaxies as well as singular galaxies. You don't really want a group of galaxies, say Leo Triplet, to be framed such that all three galaxies end up right near the edge of the frame without any space around the outside of them. That is what I've been stuck with with my current scope, and it is not a particularly pleasing framing. Galaxies, or groups of galaxies, or groups of other objects, need to have some empty space around them in the frame to get a good sense of the object, and to have an aesthetically pleasing feel. 


Edited by Jon Rista, 25 April 2019 - 12:01 PM.

  • psandelle and t_image like this

#9 weh

weh

    Explorer 1

  • -----
  • Posts: 64
  • Joined: 24 Aug 2017
  • Loc: Northeastern North Carolina USA

Posted 26 April 2019 - 01:01 AM

* I also have a 58mm noctural lens, but I haven't fully tested at what f-stop is it truely astrograph quality

If you are referring to the 58mm f/1.2 AiS Noct Nikkor, I have one also (purchased new from B&H about 4 months before it was discontinued -- now, it's often $3500+ used). Fantastic lens. I use it most often at either f/2.8 or f/4.


  • GraySkies likes this

#10 lakeorion

lakeorion

    Apollo

  • -----
  • Posts: 1293
  • Joined: 03 Aug 2010
  • Loc: Lake Orion MI

Posted 26 April 2019 - 02:09 PM

My first capture of the Horsehead was about an 8 pixel smudge.  I was using a 35mm prime and taking a picture of the whole constellation.   But I was happy there was a horsie there...


  • pinzmann, psandelle and GraySkies like this

#11 GraySkies

GraySkies

    Ranger 4

  • -----
  • topic starter
  • Posts: 381
  • Joined: 25 Feb 2011
  • Loc: Ontario, Canada

Posted 26 April 2019 - 05:23 PM

If you are referring to the 58mm f/1.2 AiS Noct Nikkor, I have one also (purchased new from B&H about 4 months before it was discontinued -- now, it's often $3500+ used). Fantastic lens. I use it most often at either f/2.8 or f/4.

yep :)

 

do you find a big improvement in f2.8 vs 4?



#12 weh

weh

    Explorer 1

  • -----
  • Posts: 64
  • Joined: 24 Aug 2017
  • Loc: Northeastern North Carolina USA

Posted 26 April 2019 - 05:55 PM

do you find a big improvement in f2.8 vs 4?

I've occasionally used it as a portrait lens on DX bodies where I tend to use it more at f/2-f/2.8 (like the isolation  but f/1.2-f/1.4 is often too much). I've tried a few star pictures with it on my D850 and SkyGuider Pro. There it seems to do a better job at f/4. By f/5.6, I think there are probably better choices. I have not done any exhaustive testing -- I'm more a grab-what-I-feel-like-using and go for it person. So, take my evaluations as very subjective rather than objective.



#13 GraySkies

GraySkies

    Ranger 4

  • -----
  • topic starter
  • Posts: 381
  • Joined: 25 Feb 2011
  • Loc: Ontario, Canada

Posted 27 April 2019 - 05:08 PM

I've occasionally used it as a portrait lens on DX bodies where I tend to use it more at f/2-f/2.8 (like the isolation  but f/1.2-f/1.4 is often too much). I've tried a few star pictures with it on my D850 and SkyGuider Pro. There it seems to do a better job at f/4. By f/5.6, I think there are probably better choices. I have not done any exhaustive testing -- I'm more a grab-what-I-feel-like-using and go for it person. So, take my evaluations as very subjective rather than objective.

Fair enough, I've been using it to film movies etc because the fast glass allows me to really shoot "in the dark" and with my new camera.... shivers... its a great combo!



#14 weh

weh

    Explorer 1

  • -----
  • Posts: 64
  • Joined: 24 Aug 2017
  • Loc: Northeastern North Carolina USA

Posted 27 April 2019 - 09:27 PM

Fair enough, I've been using it to film movies etc because the fast glass allows me to really shoot "in the dark" and with my new camera.... shivers... its a great combo!

It's a great hunk of glass -- its aspheric element was hand-ground, and, according to an engineer at Nikon whom I knew ages ago, they rejected about two-thirds of the resulting aspheric lens elements in final testing. Its rendition for portraiture using DX-format bodies is amazing, in my opinion, better than the current 58mm f/1.4G AF-S. And, its coma correction is legendary. Many of Nikon's classic lenses -- as good as they were -- don't hold up that well in the age of 45 megapixel density 35mm full-frame sensors. The 58 f/1.2 AiS Noct blasts it out of the park.


Edited by weh, 27 April 2019 - 09:30 PM.


#15 rhcrooks

rhcrooks

    Vostok 1

  • *****
  • Posts: 102
  • Joined: 28 Aug 2013

Posted 27 April 2019 - 11:53 PM

10 pixels or so - enough to distinguish it from a star. For my DSLR and AT72, that's like 25" or so.



#16 weh

weh

    Explorer 1

  • -----
  • Posts: 64
  • Joined: 24 Aug 2017
  • Loc: Northeastern North Carolina USA

Posted 28 April 2019 - 09:34 AM

Back to the original question (as my posts have been particularly off-topic) ... the answer is: it depends. The object of the image should be the predominant focus of the image, by placement or by impact. Where you are constructing a mosaic of individual images, there needs to be some overall plan to give the resultant mosaic impact but each of its elements must also hold their own.

 

Decide on the dimensions of each element, then decide how each object can create the greatest impact within its own element. Not all need to be centered. Their shape and other characteristics should dictate their placement within each of their spaces. As to a minimum size: if it isn't easily recognizable, you didn't nail it, fuzzy or not. A blur that isn't a star is still a blur.


  • GraySkies likes this

#17 weh

weh

    Explorer 1

  • -----
  • Posts: 64
  • Joined: 24 Aug 2017
  • Loc: Northeastern North Carolina USA

Posted 28 April 2019 - 03:11 PM

A further comment: essentially, you are creating a portrait of each of the Messier objects. You might take as examples some of the classic portraits from the Renaissance masters: depending on the subject and augmenting props and/or elemental framing (in this particular case, surrounding stars and clusters), they sometimes centered subjects, sometimes used an offset center, sometimes applied a loose rule of thirds (whereby the object, itself, the direction of motion (implied or actual), and the center of focus are at or near three of the four intersections of a tic-tac-toe alignment), or aligned image elements with the golden ratio -- the Fibonacci sequence -- a series of diminishing squares inscribed within an central, expanding spiral -- of which, BTW, the rule of thirds is a simplification.

 

And while you want to include the entire set, the posters I've seen with them all in order are rather boring, regardless of the impressive feat of accomplishment having captured them all. Reordering them so as to follow an implied alignment, especially with implied motion focusing inward along the Fibonacci sequence squares, could give you a mosaic with tremendous visual impact. Besides, it then becomes a game to locate them in order within the scrambled mosaic matrix -- especially if you do not number them.


Edited by weh, 28 April 2019 - 03:13 PM.

  • psandelle and GraySkies like this

#18 schmeah

schmeah

    Fly Me to the Moon

  • *****
  • Posts: 5100
  • Joined: 26 Jul 2005
  • Loc: Morristown, NJ

Posted 28 April 2019 - 04:51 PM

Another idea would be to present them to scale, to give an accurate representation of relative angular size, and in a collage. I’ve been doing this with galaxies that I capture.

 

https://pbase.com/ds...image/165313649

 

Derek


  • psandelle, GraySkies and weh like this

#19 weh

weh

    Explorer 1

  • -----
  • Posts: 64
  • Joined: 24 Aug 2017
  • Loc: Northeastern North Carolina USA

Posted 28 April 2019 - 10:06 PM

Another idea would be to present them to scale, to give an accurate representation of relative angular size, and in a collage. I’ve been doing this with galaxies that I capture.

 

https://pbase.com/ds...image/165313649

 

Derek

Some other choices: arranged by increasing or decreasing physical size, by estimated age, by distance from earth. Moreover, the individual mosaic elements need not necessarily be the same size: e.g., the smallest might be 128x128 pixels with others being multiples of that in either or both dimensions -- 128x256, 128x384, 256x256, 256x384 and 384x384 -- that is, 1x2, 1x3, 2x2, 2x3 and 3x3. There is no limit to the ways 110 objects can be displayed together other than the limits placed by ones own imagination.


  • schmeah likes this

#20 johnsoda

johnsoda

    Viking 1

  • *****
  • Moderators
  • Posts: 947
  • Joined: 30 May 2014
  • Loc: SW Ohio, USA

Posted 07 May 2019 - 09:11 PM

Okay, here is M101 and friends.  Many, many, many friends. 

 

I wasn't actually trying to get this image.  I was just playing around with my Star Adventurer and my Sigma 50 mm lens, and I chose to put M101 in the center, for no real reason - mostly just practice at aiming the Star Adventurer, which is very far from GoTo.  This was practice for doing narrowband widefields at a more appropriate time of year and to see how well the Star Adventurer does with RA guiding.  I was going to throw the hundreds of images away, and I remembered this thread.

 

Bottom line is that I've captured M101 (not for the first time, but certainly it's the most pixelated version).

 

Click on the small image for a higher resolution image:

 

get.jpg?insecure

 

M101 and many friends small.png

 

 



#21 johnsoda

johnsoda

    Viking 1

  • *****
  • Moderators
  • Posts: 947
  • Joined: 30 May 2014
  • Loc: SW Ohio, USA

Posted 07 May 2019 - 09:12 PM

And here's our old friend M101 in ultra-high resolution:

 

M101 pixelated.png



#22 bmhjr

bmhjr

    Apollo

  • ****-
  • Posts: 1205
  • Joined: 02 Oct 2015
  • Loc: Colorado

Posted 07 May 2019 - 10:26 PM

And here's our old friend M101 in ultra-high resolution:

 

attachicon.gif M101 pixelated.png

 

I would say you got it.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics