Jump to content


Photo

Agile DSLR Astrophoto Imaging & Processing

  • Please log in to reply
72 replies to this topic

#26 Starhawk

Starhawk

    Space Ranger

  • *****
  • Posts: 5546
  • Joined: 16 Sep 2008
  • Loc: Tucson, Arizona

Posted 12 August 2012 - 12:10 PM

The point is, the K-5 has the most sensitive detector (ISO 51200) thusfar deployed in a DSLR, and while the noise is quite low, it is still the signal you see in an image.

Where the DSLRs do well is high signal. I've been using something I invented called the brightness factor to compare setups:

Br=1000/(f/#)^2

So, here are a few to give an idea how it works. It puts the brightness of the image on the focal plane on a common footing where a brightness factor twice as big means the image really is twice as bright.

AP130 EDFGT, f/6.3:

Br=1000/6.3^2= 25.2

AP130 EDFGT with 0.75 reducer, f/4.725:

Br=1000/4.725^2= 44.8

Note, a magnitude change in brightness reached on the camera in a given interval requires a 2.512 improvement in the brightness ratio. So, the focal reducer isn't even giving 1 magnitude improvement as there was only a 1.78x improvement in image brightness.

Here's another:

C11, f/10:

Br=1000/10^2= 10

C11 with f/6.3 reducer, f/6.3

Br=1000/6.3^2= 25.2

Note, all identical focal ratios give identical image brightness. Also note the 2.52x improvement in the brightness ratio means this is getting a full magnitude in performance improvement. Compared to the AP130 EDFGT, this has less than half the field of view, though.

C11 with hyperstar, f/2:

Br=1000/4^2= 250

This is the configuration DSLR cameras made possible. The 25x in brightness improvement means an exposure gets close to 3.6 magnitudes of improvement over an f/10 image.

But this comes at a cost- the very low back focus leaves no room for filter wheels, and the fact it's in the middle of a glass plate means the camera has to be light. But the DSLR has built in filters and a compact light form factor, so this is possible.

As for other magic, going wheels up without a PC following the telescope around is awesome.

-Rich

#27 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 13 August 2012 - 01:34 AM

Where the DSLRs do well is high signal.

But this comes at a cost- the very low back focus.../-/...DSLR has built in filters and a compact light form factor, so this is possible.

...going wheels up without a PC following the telescope around is awesome.


You bring up some important points. And thanks for your brightness factor details; I presume the 'factor' is in support of the premise of this discussion?

#28 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 13 August 2012 - 01:46 AM

I think you should revise one of the opening statements.


You may have taken some of the statements out of the context; there was a follow up statement "Since imaging agility will vary given the scope/camera/sky conditions, the idea is to produce decent images in less time. That less time, for example, may be 10 min (combined) for 24” scope, 30 min (combined) for 10” scope, and may be 45 min (combined) for 5” scope, etc. Note: These times are just examples, not actual recommendations."

This is a discussion/work in progress, not a conclusion per se. I'll try summarizing along the way as evidence accumulates.

#29 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 13 August 2012 - 02:45 AM

Following are few samples of ~30 min integration with a 254mm APO and Canon 60Da of larger brighter objects:

Note: All images are cropped.

NGC 7635 - Bubble Nebula in Cassiopeia
M51, NGC 5194 - Whirlpool Galaxy

#30 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 13 August 2012 - 02:47 AM

Following are few samples of 45 min and 15 min integrations, respectively, with a 254mm APO and Canon 60Da of smaller dimmer objects:

Note: All images are cropped.

NGC 7048 (Planetary Nebula in Cygnus)
Minkowski 4-11 (Planetary Nebula in Scutum)

#31 Astro Boy

Astro Boy

    Sputnik

  • *****
  • Posts: 34
  • Joined: 15 Apr 2012
  • Loc: Lake Bathurst , NSW , Australia

Posted 14 August 2012 - 04:59 PM

I take it you didn't use any noise reduction software , the images show a lot of promise but a bit noisy for my taste .
The only way I can get images I'm happy with is using Topaze Denoise and Adjust , I find them both quick and easy to use .

Zane

#32 Renae Gage

Renae Gage

    Apollo

  • *****
  • Posts: 1120
  • Joined: 13 Dec 2005

Posted 14 August 2012 - 06:15 PM

I just read the fine print...254mm APO? Wow! Congrats!

I think most imagers would notice an increase in agility with that kit under dark Aussie skies.

#33 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 14 August 2012 - 11:57 PM

I take it you didn't use any noise reduction software , the images show a lot of promise but a bit noisy for my taste . The only way I can get images I'm happy with is using Topaze Denoise and Adjust , I find them both quick and easy to use.


I would agree, some smaller/dimmer ones are bit noisy; planning to redo those. Given short integrations, I have been using in-camera noise reduction.

#34 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 15 August 2012 - 12:01 AM

I just read the fine print...254mm APO


Actually, those image samples are from a suburban/red zone; not a perfect spot but still worth the try.

#35 Starhawk

Starhawk

    Space Ranger

  • *****
  • Posts: 5546
  • Joined: 16 Sep 2008
  • Loc: Tucson, Arizona

Posted 15 August 2012 - 09:38 AM

Partly- the DSLR format isn't going to have an infinite horizon for sensitivity improvement. The quantum efficiency of the professional cameras is not going to appear in DSLRs without a technology breakthrough. Because of this, they really need fast optics (previous post tells how much energy you get on the detector versus f/# of the optical system).

The conventional photographers are the driving reason for the cameras, and their needs are driving development. The internal automated routines are often impossible to turn off. So, there is a lot of hidden effort with DSLRs to bypass behaviors which affect the data.

So, I kind of feel the DSLRs are operating in a niche. If they are going to completely supplant other cameras, they are have to grow in ways their main user base is not strongly interested in. For example, the up and coming application in the mainstream world is HD video. While it's nice to see Canon throwing a bone to the astrophotography crowd now and then, the cameras they are doing it with aren't even as sensitive as the 3rd party mods.

So, I like the sentiment of the OP, but it seems to rely on something we haven't seen, yet.

-Rich

Where the DSLRs do well is high signal.

But this comes at a cost- the very low back focus.../-/...DSLR has built in filters and a compact light form factor, so this is possible.

...going wheels up without a PC following the telescope around is awesome.


You bring up some important points. And thanks for your brightness factor details; I presume the 'factor' is in support of the premise of this discussion?



#36 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 15 August 2012 - 05:56 PM

Partly- the DSLR format isn't going to have an infinite horizon for sensitivity improvement. The quantum efficiency of the professional cameras is not going to appear in DSLRs without a technology breakthrough.


Understandable. That would mean CCD is in the same boat as well, correct (horizon for sensitivity improvement I mean)?

#37 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 15 August 2012 - 06:06 PM

So, I kind of feel the DSLRs are operating in a niche. If they are going to completely supplant other cameras, they are have to grow in ways their main user base is not strongly interested in. For example, the up and coming application in the mainstream world is HD video. While it's nice to see Canon throwing a bone to the astrophotography crowd now and then, the cameras they are doing it with aren't even as sensitive as the 3rd party mods.


Very well said; I have asked this before in other threads, what are your thoughts on, hypothetically speaking, Canon offering an 'a' version of 5D Mark III? I wonder if that would be any closer to 3rd party mods.

Other thought I have is why would Canon leave 60Da hanging in terms of sensitivity compared to mods? Why not go all the way and meet or beat the mods if you (Canon) are going to do it anyway? Daytime argument...I don't get it; I never use mine for daytime; if I did, there are better choices.

#38 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 15 August 2012 - 06:12 PM

Link...


I recall you did amazing single 15 min exposures through your 24"; wondering would you like to may be try stacking/combining a few samples of both, larger/brighter and smaller/dimmer DSOs to take the 'Agile' discussion further? I am thinking may be doing ~5 minutes of integrations for each object and posting results here? That's if you are up to it.

#39 Jerry Lodriguss

Jerry Lodriguss

    Vendor

  • -----
  • Vendors
  • Posts: 3566
  • Joined: 19 Jul 2008
  • Loc: Voorhees, NJ

Posted 15 August 2012 - 06:46 PM

I don't know if you would call this "agile" or not, but here's a shot of the Milky Way and Summer Triangle shot on a fixed tripod:

http://www.astropix....ixed_Tripod.HTM

18 x 8 second exposures at ISO 6400 at f/2.8.

Jerry

#40 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 15 August 2012 - 09:46 PM

18 x 8 second exposures at ISO 6400 at f/2.8


Definitely agile... 2.4min integration. Very nice wide FOV.

#41 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 16 August 2012 - 06:40 AM

...Where the DSLRs do well is high signal...

...C11 with HyperStar... This is the configuration DSLR cameras made possible

...they really need fast optics... how much energy you get on the detector versus f/# of the optical system


I agree with most of what you said. Only part I am bit leery about is your example regarding C11 with HyperStar; my concern being added obstruction with the camera in an SC. Plus such a setup may be more suited to video. I am for unobstructed APO instead when it comes to DSLR agility.

#42 Starhawk

Starhawk

    Space Ranger

  • *****
  • Posts: 5546
  • Joined: 16 Sep 2008
  • Loc: Tucson, Arizona

Posted 16 August 2012 - 09:01 AM

Hyperstar works beautifully. The f/2 focal length of a C11 is only 560mm. So, stars are pinpoints even with the central obstruction at that focal length. This is why this is the configuration used in giant scale for the Catalina Sky Survey's 60" asteroid finding telescope on top of Mount Lemmon. The only peculiar part is what happens with cables- they cause diffraction spikes, so you can pick your spikes by moving them around.

-Rich

#43 Starhawk

Starhawk

    Space Ranger

  • *****
  • Posts: 5546
  • Joined: 16 Sep 2008
  • Loc: Tucson, Arizona

Posted 16 August 2012 - 09:08 AM

I don't know if the Canon cameras are the end-all. That's why I'm using a Pentax K-5.

Far less aggressive native IR filter inside. Very large dynamic range. Bigger detector than Canon's APS sized offerings. Last, but not least, ISO 51,200.

And at the end of the day, if I was using the same camera as everyone else, I'd be doomed to produce equivalent results.

-Rich

#44 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 16 August 2012 - 01:45 PM

I don't know if the Canon cameras are the end-all. That's why I'm using a Pentax K-5.

Far less aggressive native IR filter inside. Very large dynamic range. Bigger detector than Canon's APS sized offerings. Last, but not least, ISO 51,200.


Could you post some 'Agile' pics you may have taken with your Pentax with specs (duration, ISOs, etc.)

On a side note, you keep mentioning ISO 51,200; who shoots at that, except may be for framing, and stuff? Wouldn't you get a ton of noise? I hardly go over ISO 400-800; same is the case for most pics posted on DSLR forum.

#45 caseyfinn

caseyfinn

    Lift Off

  • -----
  • Posts: 1
  • Joined: 04 Jun 2012

Posted 26 August 2012 - 05:38 PM

Hi all. First post. I am new and have been reading for months now. My plan has evolved like so many peoples have. I copied the idea from a web search. There are some folks in the Middle East with 5 huge Cannon lenses (300mm f2.8 I think) connected straight to some really high speed ccd cameras. My guess would be they had each camera with its own filter.
Now I plan on mounting my 8"lx90 on my Atlas GEM with two Canon dslr's (one is a t3i-still shopping for the second camera) on-top. One gets a combo of a Canon 70-200 f2.8 L + 1.4x teleconverter and the other a Canon 300mm f4 L IS + 2x teleconverter . The guiding is still being worked on but currently a Meade DSI looking through the LX90. My idea is to be able to have two good lenses and one telescope trained at the same object thus reducing exposure time. How this is going to work is still not know. It sure is fun trying. :roflmao:
Casey

#46 petemumbower

petemumbower

    Surveyor 1

  • -----
  • Posts: 1680
  • Joined: 12 Jul 2009
  • Loc: West Michigan

Posted 14 September 2012 - 02:24 PM

Ok, I am now lost on the purpose of this thread or the concept. Is it about avoiding dedicating multiple nights to a single object to go deep or overcome LP? My takeaway is for a minimalist approach to time/processing. To me this approach is more about getting more objects a night than for a better result. Granted I find it tough to stick with a single object for multiple nights, the end result is always more satisfying to me when I put in the hard work. Now my processing at this point is very minimum since I am still learning that part of this hobby.

Again, correct me if I am wrong, but I am not seeing the concept here very well.

#47 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 14 September 2012 - 03:02 PM

Is it about avoiding dedicating multiple nights to a single object to go deep or overcome LP


Idea is to produce good quality astro photos with less time and effort, both imaging and post-processing. I am not saying we should cut down on sub-exposure length necessarily (although, that may be part of the equation), but mainly cutting down on total integration time. This may be possible with better technology (e.g., 60Da and other newer modified/optimized cameras). As I stated in my first post, goal is to use new and evolving technology in faster/better manner.

There are variables to this equation and agile imaging/processing will vary from person to person and equipment to equipment. For example, a half hour integration time for a DSO on a 254mm APO could be one hour integration time on a 130mm APO; a half hour integration time for an a modified/optimized camera could be one hour integration time on an unmodified one and so on… I am also suggesting reducing processing time/effort and NOT going out of the way to make images look pretty to the point they start looking synthetic. Some amount of processing is always needed, but over processing can be avoided.

@Falcon said in another post “there is an "ideal" exposure length for sub exposures. Go shorter than that and you are losing out on faint detail, go longer and you gain very little even after adding a lot more exposure time.” Agile effort in my opinion is partly about finding that minimum optimal time for a sub as well as finding minimum total integration time (of all subs combined).

Hope this helps. I am looking for ideas and would like to develop some kind of consensus that this is something worth pursuing and that it could be a worthy goal. In short, idea is to use newer technology in creative ways which results in reducing time and effort it takes to take images and process them. I’ll post some example images in coming posts and processing methods used.

#48 petemumbower

petemumbower

    Surveyor 1

  • -----
  • Posts: 1680
  • Joined: 12 Jul 2009
  • Loc: West Michigan

Posted 14 September 2012 - 06:43 PM

Thanks mmalik, I understand what is being presented here. More of a best practices via equipment/technique that will produce good results with minimum effort.

Which I am all for, but unfortunately does become a issue of throwing money at the various pieces of equipment. Like the latest camera, mounts, fast optics, filters, etc... But it does give a good reference point to what state of the art can produce.

#49 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5357
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 25 December 2013 - 03:21 PM

Thought I'll chime in on the subject of "diminishing returns" being discussed in the other... thread. I am also of the view, as I have discussed above, that there is a way to maximize your return with "logically" optimal effort, and that effort has to take into account both acquisition and processing related happy medians; cross those thresholds and you'll find yourself struggling and spending countless hours acquiring as well as processing. Regards

#50 Samir Kharusi

Samir Kharusi

    Surveyor 1

  • *****
  • Posts: 1996
  • Joined: 14 Jun 2005
  • Loc: Oman

Posted 26 December 2013 - 07:32 AM

Cutting down on the length of subs, use:

1. OTA with a very fast focal ratio to yank the Skyfog Peak well away from the Read Noise.
2. Camera with very low Read Noise, to guarantee good Stacking Efficiency (i.e. so that the Skyfog Signal is overwhelmingly larger than the Read Noise)
3. Camera with large pixels so more photons fall on each pixel and the signal generated by the Skyfog (and of course the target object) is huge.
4. A mono astroCCD if you are willing to settle for a B&W final image.

To minimise integration time:
Use 1., 3. and 4. above.
Settle for a small final image. An HDTV screen is only 2 megapixels (1080x1920) and even the larger monitors we use today on our PCs do not have more pixels. So anything over 2 MP is perhaps redundant. But "retina" displays may well be coming down the line tomorrow (4k TV anyone?).

How do the above work out? Using a current Canon crop DSLR and an OTA at f2 (Hyperstar) will require half-minute subs for satisfactory Stacking Efficiency at any site on Earth. If you use a Canon with larger, but (current technology) pixels, you can of course cut down the length of subs substantially.

Integration time depends on the eye of the beholder. If all you want is an image no larger than 900 pixels x 600 pixels and you are shooting from a dark site then you might well be very happy with less than an hour's worth of integration time. Of course if you are looking for large prints, then no integration time is ever going to be "enough".

How can we speed up everything further? Use a mono camera (also include the IR, so only mirror scopes need apply) and cut both the length of subs and integration time by a factor 5x. That ought to be quite "agile"! 6 second subs and 12 minutes integration time.

Even more agile? Find/fabricate a camera that fully utilizes the image circle of a Hyperstar (27mm diameter) but has only 2 megapixels (recall that is the best that can be displayed by an HDTV or most of today's PC monitors). Each pixel would be about 14 microns square (or about 10x the area of a pixel in a Canon crop DSLR). If it has a Bayer Array (for agile color imaging) then each sub can be as short as 3 seconds) but if you are ready to settle for mono then each sub can be only half a second! Again, integration time can also be shortened 10x (6 minutes for Bayered RGB, or ONE minute for mono). That would be awesomely "agile" :roflmao:

In brief, with existing cameras and OTAs: 30sec subs and one hour integration time using a current model Canon crop DSLR is about as "agile" as it gets. Cut down the size of the final image to 600x400 pixels and you can cut down the integration time to half and still be very happy with the apparent SNR. Use an f2.8 OTA and you need to double up. Use a slow OTA, say, f6 or slower, and you are no longer in agile territory.

Is there another way that does not require a Hyperstar? Almost certainly do-able with items available in the market. Steps to find such equipment:
Find an astroCCD that has around 2 megapixels of recent technology. It will be quite small and I expect that it will be sufficient to use it with C-mount cine or video lenses. These lenses have small image circles and are consequently very compact compared to 35mm camera lenses. These C-mount lenses also tend to be very high quality and have very fast focal ratios. I have come across some with f1.0 but of course they cost a lot. This route really makes sense only to those with access to such lenses in their other line of work. Secret to "agile" imaging is always to start with very fast focal ratios at a dark site. An f6 is 9x slower than an f2 OTA. Suburbia requires 15x to 40x as much integration time as a dark site. Specs like Quantum Efficiency, etc have trivially small variations in comparison.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics