•

# image scale, under/over sample, type of camera's for focal lengths

31 replies to this topic

### #1 belbakri

belbakri

Sputnik

• topic starter
• Posts: 26
• Joined: 07 May 2019

Posted 19 August 2019 - 06:47 PM

Hi All,

I am looking at getting into AP work and I want to know where there is some information, in one place if possible to help me understand the following topics

• Image Scale
• Under/over sampled
• arc sec/ pixels

Along with that how does focal length/aperture effects the above and how to pick a guide camera and main camera for various scopes (size) both for OAG and with using a separate guide scope.   I know this is a lot of information but it would be nice if I could find in one place.  Perhaps there is a book that does of this  or a good write up somewhere, IDK  but any help would be greatly appreciate.

thanks

Brent

### #2 spacemunkee

spacemunkee

Gemini

• Posts: 3,274
• Joined: 22 Oct 2017
• Loc: Cincinnati, Ohio

Posted 19 August 2019 - 07:07 PM

This book is highly recommended. Helped me understand a lot. Very in depth.

https://www.amazon.c...la-401450758693

• ks__observer likes this

### #3 belbakri

belbakri

Sputnik

• topic starter
• Posts: 26
• Joined: 07 May 2019

Posted 19 August 2019 - 07:19 PM

Thanks.... as  a matter of fact I just found it and ordered before I saw your post ... thanks!!

### #4 Stelios

Stelios

Cosmos

• Posts: 8,276
• Joined: 04 Oct 2003
• Loc: West Hills, CA

Posted 20 August 2019 - 04:31 AM

I’ll try to give a rather simple and rough answer that will hold you while that excellent book arrives:

Image scale is given by the simple formula: 206.3*pixel_sixe_in_microns/focal_length_in_mm

That also helps show what’s most important. Note that aperture doesn’t appear.

As an example, for a 600mm F/L refractor and a Canon T3i DSLR, the formula gives 206.3*4.3/600=1.48(arcsec/px)

OK, what does the value mean? The value is relevant compared to the seeing. For most of the US seeing is around 2”, ranging from 1 to 3. In the Canary Islands, parts of the Caribbean and Chile it can be half that or less.

Roughly speaking, you are “under sampled” when your image scale is greater than your seeing, “oversampled” when it’s less than 1/2 your seeing, and normal(ish) otherwise. None of these conditions is fatal, but in general you want to be in the normal range, as otherwise you are either not maximizing your equipment or wasting it. So a general rule-of-thumb is to aim for 1-2” image scale, and not worry as long as you’re not wildly over or under. 0.7 or 2.8 is OK—0.2 or 4.1 is usually not unless your conditions are very atypical.

For guiding, a common rule-of-thumb is to have the guider image scale be no more (less is OK) than 3-5x the imaging camera scale. In practice, you can guide almost anything under 500mm with a 30mm guide scope, anything under 1000mm with a 50mm guidescope, and for more (and especially for movable mirror scopes) you want an OAG. (This assumes commonly sold guidescopes and cameras).

An OAG is OK even with shorter focal lengths, but the relative setup complexity and extra cost makes it inadvisable for most beginners.

Please note that the above rules are general approximations, that will serve you fairly well till you become knowledgeable enough to navigate the many other pitfalls and requirements of our time-and-money pit of a beloved hobby .

• kingjamez, Gene3, GoFish and 3 others like this

### #5 belbakri

belbakri

Sputnik

• topic starter
• Posts: 26
• Joined: 07 May 2019

Posted 20 August 2019 - 05:16 AM

Stelios,

thanks. I've always enjoyed your comments and feedback in the forums. Keep it up

### #6 dhaval

dhaval

Vendor

• Posts: 1,813
• Joined: 21 Jul 2008
• Loc: Round Rock, TX

Posted 20 August 2019 - 07:34 AM

Continuing on the general theme that Stelios has provided - while over/under sampling are not "fatal" - there are some downsides associated with either.

The primary downside of oversampling is that you will need an exceptional mount with very good knowledge of guiding. This is where terms like "chasing the seeing" become more important.

While there are very few downsides to undersampling, the challenge with undersampling is that you are not utilizing your equipment to its fullest. However, such a set up does allow a person to learn and experiment because you can "get away" with certain flaws in guiding, image processing, etc. That is the reason why most people will suggest an 80mm scope to most newcomers.

With regards to OAG, I like to deviate from the norm - OAG's may be difficult to set up, but they are one and done type activity. You don't have to set it up every night you go out to image. The upside is that you can set it up during the day as well. Most people who are looking for a "cheaper" guiding solution will most likely face an issue with differential flexure - they won't even realize they have that issue because quite honestly, for a beginner, guiding is the last thing that they will learn. This results in many bad subs, not realizing why you don't have more subs that keepers, etc. I like to suggest OAGs to new comers, regardless of their skill or equipment. Besides, when you think about it, an OAG is not that expensive than a guidescope set up. Where you may have a challenge would be in figuring out the right back-focus, just because the equipment that you select may not have enough - but you can always ask around before buying.

CS!

Edited by dhaval, 20 August 2019 - 07:35 AM.

• Gene3 likes this

### #7 ks__observer

ks__observer

Apollo

• Posts: 1,118
• Joined: 28 Sep 2016
• Loc: Long Island, New York

Posted 20 August 2019 - 09:18 AM

Agree with above re OAG.

Re oversampling:

For mono -- test downsample/binning to recover SNR -- as you waste SNR otherwise.

Re Undersampling:

There are great widefield shots on Astrobin at 10 asp, 20 asp, etc.  As long as your stars don't look blocky i think you are fine.

### #8 bobzeq25

bobzeq25

Hubble

• Posts: 18,325
• Joined: 27 Oct 2014

Posted 20 August 2019 - 09:46 AM

The book is best, here's an overview.

As Stelios says 1-2 arc sec per pixel is a typical range.

Closer to 1 can give better resolution, _if_ a number of other things (seeing, tracking, ...) cooperate.

Closer to 2 always gives a better signal to noise ratio, with equal total imaging.

Both are important for good images.

Some (very) experienced imagers will go numerically below 1.  Things like deconvolution can work better.

Imaging at short focal lengths (particularly imaging with camera lenses) frequently goes numerically over 2.  Nothing wrong with it.

I would suggest that a beginner go in the vicinity of 2, it's more "forgiving" of typical beginner problems.

THE most important thing to consider.  The two most common beginner mistakes are skimping on the all important mount, and/or getting too big a scope.  Visual astronomy experience often misleads people, imaging is completely different.  That's more important than details like image scale.

Your 9.25 would be far too big, the typical SCT fork mount far too little mount.

Here's another good (but more basic) book.  Scroll down the page to the picture of the very expert author.  That's a \$500 70mm refractor on a \$1200 Sirius mount.  It's just about the ideal setup for getting into DSO AP.  The right tool for the job.

http://www.astropix....bgda/index.html

Here's where you do not want to go.

"I started out with a CPC 800 on a heavy duty wedge and a Canon 450d.  In hindsight, I'd have started with an 80mm refractor <on a good mount>.  I would have saved a lot of money and gotten up the learning curve a lot quicker."

Edited by bobzeq25, 20 August 2019 - 10:00 AM.

Fly Me to the Moon

• Posts: 5,257
• Joined: 24 Feb 2007
• Loc: Snohomish, WA

Posted 20 August 2019 - 10:34 AM

The primary downside of oversampling is that you will need an exceptional mount with very good knowledge of guiding. This is where terms like "chasing the seeing" become more important.

...

While there are very few downsides to undersampling, the challenge with undersampling is that you are not utilizing your equipment to its fullest.

I kind of disagree with this.

The primary downside to over sampling is that you are giving up some per-pixel signal to noise ratio at capture time.  This is just because a smaller pixel collects less light than a larger one.  The only other down side to over sampling is that your file sizes are larger than they need to be.  There are some benefits to over sampling.  For example, over sampled stars tend to respond better to deconvolution.  Also, if you are a bit over sampled, you will be able to take advantage of nights of exceptional seeing.

The primary down side to under sampling is that you are not sampling the stars properly.  Small, dim stars will be pixelated and not well shaped.  If you are doing a wide field shot, where nobody will be zooming in on individual stars, then this is probably ok.  As with over sampling, there are some benefits to under sampling.  Chiefly, under sampled stars will hide tracking and guiding errors better than well sampled or over sampled stars (over sampled stars take no prisoners at revealing tracking and guiding problems, or optical correction problems).  Also, you can fix your star shapes by using drizzle integration.

All things considered, I would much rather be over sampled than under.  The exception is doing wide field work, where the subject is huge and individual stars are not going to be featured.

• dhaval and Jon Rista like this

### #10 GoldSpider

GoldSpider

Viking 1

• Posts: 888
• Joined: 20 Apr 2015
• Loc: Pennsylvania

Posted 20 August 2019 - 10:53 AM

AstronomyTools has some nice AP utilities, including this one for finding the right camera/scope combination.  This answered a lot of questions I had about focal length and sampling.

### #11 nimitz69

nimitz69

Surveyor 1

• Posts: 1,754
• Joined: 21 Apr 2017
• Loc: A barrier island 18 miles south of Cocoa Beach

Posted 20 August 2019 - 11:00 AM

Since a critical part of this is that your seeing conditions fall basically somewhere around 1”-2” is there a straight forward way to measure your local seeing conditions that is accurate enough for this purpose?

### #12 dhaval

dhaval

Vendor

• Posts: 1,813
• Joined: 21 Jul 2008
• Loc: Round Rock, TX

Posted 20 August 2019 - 11:02 AM

I kind of disagree with this.

The primary downside to over sampling is that you are giving up some per-pixel signal to noise ratio at capture time.  This is just because a smaller pixel collects less light than a larger one.  The only other down side to over sampling is that your file sizes are larger than they need to be.  There are some benefits to over sampling.  For example, over sampled stars tend to respond better to deconvolution.  Also, if you are a bit over sampled, you will be able to take advantage of nights of exceptional seeing.

The primary down side to under sampling is that you are not sampling the stars properly.  Small, dim stars will be pixelated and not well shaped.  If you are doing a wide field shot, where nobody will be zooming in on individual stars, then this is probably ok.  As with over sampling, there are some benefits to under sampling.  Chiefly, under sampled stars will hide tracking and guiding errors better than well sampled or over sampled stars (over sampled stars take no prisoners at revealing tracking and guiding problems, or optical correction problems).  Also, you can fix your star shapes by using drizzle integration.

All things considered, I would much rather be over sampled than under.  The exception is doing wide field work, where the subject is huge and individual stars are not going to be featured.

Completely agree with your assessments. I am not suggesting that there are not other downsides or even that they are more important - but my post was mostly in reference to the fact that the OP is a new comer and for them, it is hard to comprehend SNR and such (no judgment for or against the OP, just a general statement).

I also agree that I would rather be over sampled than under. In fact, at our remote site, we are doing 0.27"/px - and it is way more fun than properly sampled or under sampled. But you can only do that with really good equipment and awesome skies.

CS!

### #13 the Elf

the Elf

Vanguard

• Posts: 2,033
• Joined: 06 Sep 2017
• Loc: Germany

Posted 20 August 2019 - 12:04 PM

While it is a good idea to think about important parameters before any purchase I'd like to point out that there are more important considerations for a beginner: your tracking and guiding. Whatever pixel scale you pick, you need a mount that is able to track within that limit for the exposure times you want/have to use. If you go for a scale near your seeing you need either auto guiding or a premium mount. You probably know that you should put 50+% of your budget into the mount and then select the image scale accordingly. Some real life examples: the famous (poor, not recommended) AVX is troubled tracking within 3 arcsecs, with my EQ6-R I am a bit below 2 I guess. Everything becomes much easier if your scale is 3-5 arcsecs per pixel, you don't have much trouble with the mount in that case.

• Jim Waters likes this

### #14 bobzeq25

bobzeq25

Hubble

• Posts: 18,325
• Joined: 27 Oct 2014

Posted 20 August 2019 - 12:04 PM

Since a critical part of this is that your seeing conditions fall basically somewhere around 1”-2” is there a straight forward way to measure your local seeing conditions that is accurate enough for this purpose?

Something we'd all like, but not really available.

It's something less than your FWHM, and highly variable.  Both night to night, and within nights.  Mine can change dramatically in 10 minutes.

I can tell, at least relatively, two ways.

How much are the PoleMaster targets dancing around?  If it's less than their size, it's a decent night.

What's my guiding like?  0.7 is usual.  0.6 is good.  0.5 is great.  If it's above 1.0, seeing is so bad it's hardly worth imaging.  Maybe collect some color data binned 2X2.

About oversampling.  It's pretty much an experienced imager deal.  If you know specifically why you want to do it, great.  Beginners are better off undersampling.

Edited by bobzeq25, 20 August 2019 - 12:06 PM.

Fly Me to the Moon

• Posts: 5,257
• Joined: 24 Feb 2007
• Loc: Snohomish, WA

Posted 20 August 2019 - 12:25 PM

Since a critical part of this is that your seeing conditions fall basically somewhere around 1”-2” is there a straight forward way to measure your local seeing conditions that is accurate enough for this purpose?

I would agree with other responses on this so far, but would add that if you take a good look at the sky while you are imaging, you will get a feel for things.  I can get a general idea of "good seeing", "mediocre seeing" or "bad seeing" just by looking at the sky for a few minutes at the start of the night.  The thing to pay attention to is the "twinkle" in the stars.  The more twinkle, the worse the seeing.

• bobzeq25 likes this

### #16 kingjamez

kingjamez

Vanguard

• Posts: 2,160
• Joined: 03 Oct 2006
• Loc: Fairfax, VA

Posted 20 August 2019 - 12:43 PM

I look at my declination guide graph on PHD2 to determine how good the seeing is. If it’s smooth then seeing is good, if it’s jagged and jumping back and forth, seeing is bad.

### #17 Jon Rista

Jon Rista

ISS

• Posts: 24,079
• Joined: 10 Jan 2014

Posted 20 August 2019 - 01:07 PM

I kind of disagree with this.

The primary downside to over sampling is that you are giving up some per-pixel signal to noise ratio at capture time.  This is just because a smaller pixel collects less light than a larger one.  The only other down side to over sampling is that your file sizes are larger than they need to be.  There are some benefits to over sampling.  For example, over sampled stars tend to respond better to deconvolution.  Also, if you are a bit over sampled, you will be able to take advantage of nights of exceptional seeing.

The primary down side to under sampling is that you are not sampling the stars properly.  Small, dim stars will be pixelated and not well shaped.  If you are doing a wide field shot, where nobody will be zooming in on individual stars, then this is probably ok.  As with over sampling, there are some benefits to under sampling.  Chiefly, under sampled stars will hide tracking and guiding errors better than well sampled or over sampled stars (over sampled stars take no prisoners at revealing tracking and guiding problems, or optical correction problems).  Also, you can fix your star shapes by using drizzle integration.

All things considered, I would much rather be over sampled than under.  The exception is doing wide field work, where the subject is huge and individual stars are not going to be featured.

The other thing about oversampling is that you can always downsample your results to make a larger effective pixel. This is not perfect, you still have read noise in each pixel so total read noise in each effectively larger pixel will be higher...but there can still be a significant gain to pixel SNR by downsampling oversampled data. The benefit here is, you don't suffer from the consequences you do with undersampling (which is not just blocky stars...but stars that will saturate more quickly as well due to more signal concentrated into each pixel...which can lead to the stars dominating the field, making it hard to see what's behind them), while still gaining most of the benefits.

### #18 belbakri

belbakri

Sputnik

• topic starter
• Posts: 26
• Joined: 07 May 2019

Posted 20 August 2019 - 05:33 PM

I have lots to learn... hopefully I won't get too frustrated but will all helpful CNer's here I am sure I will find help when I need it .... thank you all!

### #19 Jon Rista

Jon Rista

ISS

• Posts: 24,079
• Joined: 10 Jan 2014

Posted 20 August 2019 - 06:46 PM

I have lots to learn... hopefully I won't get too frustrated but will all helpful CNer's here I am sure I will find help when I need it .... thank you all!

For a beginner, a happy place to be is around 1"/px. I wouldn't worry about sampling rate so much until you have the other basics down. As a beginner, an image scale around 1-1.5"/px will give you good SNR with more forgiving exposures, without resulting in stars that saturate so fast that you have problems with dynamic range, noise, etc.  Until you get a handle on how to track, guide, focus, etc. it is likely that you will have a sampling rate (at least for stars) of 3-4x, even with a 1.5" pixel. This is because a lot of things can and will blur the image until you get them under control...

Since you are just getting into AP, worrying about sampling in any capacity is just putting the cart before the horse. You will have a lot to learn before that ever becomes an issue. Not only will you need to get a handle on how to manage tracking, how to guide well, and how to get your guided RMS error under 1" (which is really going to be necessary before you have to start worrying about sampling rate), you will also need to get a handle on focus as well. Focus is an oft-overlooked but critical factor in image quality and star FWHM. Even if you are tracking optimally, if you are not keeping things focused, even with bigger pixels you can easily have stars that sample 3-4x just because of less than ideal focus.

Cover the basics first. There is a LOT to learn. Once you have the basics handled, then start pushing into the more advanced aspects of imaging, such as how to optimally match image scale to scope, how to get and maintain optimal focus, etc.

### #20 the Elf

the Elf

Vanguard

• Posts: 2,033
• Joined: 06 Sep 2017
• Loc: Germany

Posted 25 August 2019 - 07:00 AM

Here is an example of strong undersampling:

http://www.elf-of-lo...n_2019_1080.jpg

Taken with a mono modded T3i and a vintage 200mm Ashai Pentax Takumar and 7nm Baader Ha filter. The image is downsampled to full HD, so most stars are just one pixel. The FWHM was 1.0some pixels, so pinpoint sharp. The stars bloated by stretching but as most displays are full HD I did not bother correcting this. The downsampling also helps the lot of noise I have in the hot summer nights with my uncooled camera. Here is the full res:

http://www.elf-of-lo...n_2019_full.jpg

I'm not afraid of the undersampling monster. Are you?

Edited by the Elf, 25 August 2019 - 07:01 AM.

• bobzeq25 and ks__observer like this

### #21 steste1122

steste1122

Explorer 1

• Posts: 56
• Joined: 20 Sep 2013

Posted 25 August 2019 - 11:14 PM

Continuing on the general theme that Stelios has provided - while over/under sampling are not "fatal" - there are some downsides associated with either.

The primary downside of oversampling is that you will need an exceptional mount with very good knowledge of guiding. This is where terms like "chasing the seeing" become more important.

While there are very few downsides to undersampling, the challenge with undersampling is that you are not utilizing your equipment to its fullest. However, such a set up does allow a person to learn and experiment because you can "get away" with certain flaws in guiding, image processing, etc. That is the reason why most people will suggest an 80mm scope to most newcomers.

With regards to OAG, I like to deviate from the norm - OAG's may be difficult to set up, but they are one and done type activity. You don't have to set it up every night you go out to image. The upside is that you can set it up during the day as well. Most people who are looking for a "cheaper" guiding solution will most likely face an issue with differential flexure - they won't even realize they have that issue because quite honestly, for a beginner, guiding is the last thing that they will learn. This results in many bad subs, not realizing why you don't have more subs that keepers, etc. I like to suggest OAGs to new comers, regardless of their skill or equipment. Besides, when you think about it, an OAG is not that expensive than a guidescope set up. Where you may have a challenge would be in figuring out the right back-focus, just because the equipment that you select may not have enough - but you can always ask around before buying.

CS!

As a newcomer to the hobby and someone who has setup a Celestron off axis guider (although I have not yet used it for guiding) I can share my experience setting it up.  After getting frustrated with it and putting it aside for a month I came back to it determined to get it working.  What I found worked for me is first I found focus on the guider.  Next I moved the focuser out till I achieved focus on the main imager.  Since I knew I needed to add more length to get the main camera to focus I put a couple spacers behind the OAG and returned to the position where I had focus on the guider.  This brought my main camera close to focus because even though it moved it inward, the spacers still had it sitting just as far back  Now I just had to fine tune it which brought the guider out of focus but  that was easily correctable.  The only other real problem is all the different adapters you need to figure out.

Edited by steste1122, 25 August 2019 - 11:14 PM.

• dhaval likes this

### #22 bobzeq25

bobzeq25

Hubble

• Posts: 18,325
• Joined: 27 Oct 2014

Posted 26 August 2019 - 11:13 AM

Here is an example of strong undersampling:

http://www.elf-of-lo...n_2019_1080.jpg

Taken with a mono modded T3i and a vintage 200mm Ashai Pentax Takumar and 7nm Baader Ha filter. The image is downsampled to full HD, so most stars are just one pixel. The FWHM was 1.0some pixels, so pinpoint sharp. The stars bloated by stretching but as most displays are full HD I did not bother correcting this. The downsampling also helps the lot of noise I have in the hot summer nights with my uncooled camera. Here is the full res:

http://www.elf-of-lo...n_2019_full.jpg

I'm not afraid of the undersampling monster. Are you?

You call that undersampled?  <big grin>  In the words of Crocodile Dundee, _that's_ not undersampled...

Here's 31 (three one) arc sec/pixel.

http://www.astrobin....9102/B/?nc=user

Edited by bobzeq25, 26 August 2019 - 11:13 AM.

### #23 the Elf

the Elf

Vanguard

• Posts: 2,033
• Joined: 06 Sep 2017
• Loc: Germany

Posted 26 August 2019 - 01:33 PM

http://www.elf-of-lo...lkyWay2019.html

53 arcsec/pixel

Have you got a 180° fish eye? If so, you win.

• bobzeq25 and GoldSpider like this

### #24 bobzeq25

bobzeq25

Hubble

• Posts: 18,325
• Joined: 27 Oct 2014

Posted 26 August 2019 - 02:39 PM

The other thing about oversampling is that you can always downsample your results to make a larger effective pixel. This is not perfect, you still have read noise in each pixel so total read noise in each effectively larger pixel will be higher...but there can still be a significant gain to pixel SNR by downsampling oversampled data. The benefit here is, you don't suffer from the consequences you do with undersampling (which is not just blocky stars...but stars that will saturate more quickly as well due to more signal concentrated into each pixel...which can lead to the stars dominating the field, making it hard to see what's behind them), while still gaining most of the benefits.

When in the PI process do you typically downsample?

Messenger

• Posts: 418
• Joined: 03 Jun 2018
• Loc: Wisconsin

Posted 26 August 2019 - 04:26 PM

The primary downside to over sampling is that you are giving up some per-pixel signal to noise ratio at capture time.  This is just because a smaller pixel collects less light than a larger one.

Can you elaborate on this?  Why do smaller pixels collect less light for a given area than larger pixels?  Is it due to gaps between pixels causing photons to reflect? Therefore a 2x2 binned area would collect less light than one large pixel covering the same area.  If so, how much light is lost due to this effect?  I'm confused because isn't this is already factored into the QE of a sensor?

Maybe I'm way off and it has to do with something else such as how SNR is calculated.

Edited by chadrian84, 26 August 2019 - 04:59 PM.

## Recent Topics

 Cloudy Nights LLC Cloudy Nights Sponsor: Astronomics