Hi guys. Ive been playing with dslr on a heq5 pro mount taking images of m42.
I took some images from backyard (around 1hrs data. Mostly 1 and 2 min subs with a few 30 secs subs thrown in at 800 iso). About 5 dark frames for each exposure were added.
I was quite happy with the result as its my first proper imaging attempt on a decent mount.
However ive added another hours worth of data (similar composition of subs) and really did not see any improvement over the first sessions image.
Is there a limit where adding more exposures becomes futile?
I also had the experience of shooting an hours worth of subs on m42 from a dark site but after stacking (with dark frames, no flats/bias) the image was worse than my exposures under light polluted skies. Came out very grainy like it was noisy. Even though the individual subs looked good. Temp of sensor was the same between both sessions around 36 degrees on my unmodded dslr.
I only use lightroom to edit after stacking so nothing too fancy on the editing side. But what would cause that kind of issue on subs under dark skies? Do i need more exposures or darks to cancel the noise?
1. Sensor temperature, the lower, the longer max. exposure.
2. ISO value. Most optimal ISO is the highest with an acceptable dynamic range, which is 800 for many lower end DSLRs but 1600 for DSLRs like EOS 6, 5, Nikon FX bodies and Nikon D500 or Sony A7.
3. Tracking accuracy / polar alignment. For me, I get decent images without guiding when I just set BQ Octantis (not Sigma !) in the center of the polar scope field, 600mm FOV (full frame) and 4 minutes per frame.
4. A rough guideline is that the real exposure value of a stacked inage is sqrt(number of frames) * exposure per frame. So 10 frames of 1 minutes is NOT equivalent as a single 10 minute exposure, but just a single 3.16 minute exposure with one third of the noise. Noise is also inversely equivalent to number of frames, compared to single frame of that exposure.
5. Most important: avoid light polluted areas, unless you use decent filters like UHC or Halpha. But Rockhampton is not that bad I think.
It very much depends on the camera you are using, and your ability to keep it on target accurately. You mention ISO so i have to assume your talking about a DSLR, which has very different performance characteristics to something like a cooled, ultra low noise CMOS camera. When read noise and thermal noise cease to be a factor, then cumulative exposure time can be equivalent to the total exposure time, without discounting.
2. ISO value. Most optimal ISO is the highest with an acceptable dynamic range, which is 800 for many lower end DSLRs but 1600 for DSLRs like EOS 6, 5, Nikon FX bodies and Nikon D500 or Sony A7.
Nikon FX bodies from the 5100 onwards are best down low, ISO 200 as they are pretty much ISO invariant. If you increase the ISO you're just needlessly decreasing dynamic range.
I have noticed the same thing, 30min to 60min worth of exposures is the sweet spot, after that the difference / improvement is subtle. As for single subs I expose for the background, to not let it get too bright. For me in Mackay it's 2min at iso 1600, unmod canon 1100d.
A few more darks might help with the noise I try and use about 20, I have a dark library and reuse previously shot darks.
Suggest you read the 5-part “Fishing for Photons” series by Craig Stark. It’ll clear up the mystique! It basically all boils down to maximising the signal-to-noise ratio (SNR). Everything else is detail, and a means to an end (at least for the capture and pre-processing stages!)
For the purposes of getting a pretty picture there is a limit and its entirely subjective. If you dont see an improvement you've reached your limit. It will differ widely for everybody so any "rules" you see are just ballpark guidelines that should get you to an ok picture. If your processing skills are lacking you reach your limit fast, if you think a final image is what you get in camera you are way below what your limit should be. Processing is king.
For the purposes of taking scientific measurements there's almost no limit. The more data you collect the more precise and certain results can get. But plotted your values curve towards the precise value but won't ever reach it as no equipment is free of noise and the noise threshold is the true signal limit to contend with. You can only supress noise with more data and better processing techniques, never exterminate it. Its called noise Reduction for a reason.
Thanks guys. So basically there is a "soft limit" where adding additional exposures is not going to achieve the same "increase" in details / colour/ etc.
I must say if I exposed at 2min iso 1600 from home my sub would be completely white. even 2min at 800 iso is probably a bit over done tbh. But I might try adding some subs of different exposures with longer/ shorter sub lengths.
And given my processing skills in DSS, lightroom and Photoshop are still being refined, I'd say my images could probably use alot of improvement in that department. (I only found out about gradientXterminator last week which has helped alot!).
I was curious about one thing. If anyone would mind having a look at my raw data and putting it through their processing work flow to see the end result.
I am considering getting Pixinsight or Astropixel processing. But given I'm really just a beginner not sure If I need to go outlaying the funds to purchase that software just yet.
Im wondering if you are somehow getting additive noise in your workflow
What is the calibration process you are using ?
Dark noise in a DSLR is difficult to control, because each image may have a different amount of dark noise and that dark noise can be related to heat whoch often flows from one side or corner of the CCD chip and migrates its way across the frame - meaning each successive light frame has more noise and different noise in different places than the last - so dark frame subtraction might be difficult
Im not familar with your camera so cant advise on whats best or how your camera behaves, but obviously a chilled astro camera is better becasue it can control this to a very manageable level.
But selecting the correct statistical process for your calibration, stacking, master darks etc can be critical.
You say no Bias or Flats - anyone here familiar with the levels of Bias noise for that camera ? . . . and the best process.
Going longer using an chilled camera allows you to capture the faint nebulosity without the big penalties of dark noise (which gets larger with exposure time and temperature) - the problem with unmodded DSLRs
This blows out the stars but you can correct this in your processing workflow and by using the right statistical tool effectively use the values of the stars from the shorter frames.
If you like astro imaging and think you can see yourself going further - then I think its a very wise idea to bite the bullet and get PixInsight now - the learning curve is worth the effort, but its a vastly different (better) path than Photoshop !
You dont chop a tree down with tooth pick or even a pen knife ! - Use an Axe or get a Chainsaw - having the right tools for the job makes life a lot easier and will save you valuable time in the processing as well as shorten your learning curve. - But its is still a leap of faith.
There are plenty of really easy to follow tutorials for OSC and DSLR astro imaging - Harry's PixInsight Tutorials used to be the good oil, but there are others now. You can just follow them step by step and get useful results.
Gradient removal and white star balancing is almost a doddle in PixInsight.
Are you checking the seeing/tracking/guiding quality of your subs ? - check the FWHM of your subs - get rid of the bad ones so you arent introducing extra noise into the workflow by a few rampant bad subs !
Good luck - I am sure you will work out what you have done and use those same subs to extract some good images
Hi rally. There is no real processing work flow for me.
I stack the lights darks and i generally take bias frames too but didnt this time round.
I tried adding bias frames into the stack after by taking them at same temp onthe sensor but this just prevented dss from stacking them. (Ended up with a screen of lines). But it will stack fine without bias frames.
But after stacking in dss (all settings are default) i adjust the colour slider and saturation so i can see whats going on in the image then move into lightroom where i just play with the sliders till im happy.
I do plan to keep experimenting with the data ive got and see if i can get something better. But for 3 hours of data on orion i thought i would of ended up with something 'more'. My best pic of my procesing attempts is attached. Only 2 hours of data as ive got 2 hours of data from in town and an hour from dark skies. (Havent stacked the dark site ones in yet.)
Firstly, as long as your darks match your lights, biases are unnecessary,
as darks contain bias information. Secondly, the colours in your image are
all wrong, as you may already be aware from seeing other people's images.
A few basic DSS pointers.
When you get the post stacking image you will see three tall narrow
coloured dotted line cones, and another dotted line rising steeply from the
bottom of the histogram. When the image is finalised the tall cones will
intersect the rising line somewhere around 10-12mm from the bottom
of the histogram frame. Using the RGB sliders move the upper one so
that the right leg of its tall cone intersects the rising line as described above.
Do the same thing with the other two sliders.
Check the image to see if it is too bright or dark,[ too dark, and a lot of detail will be lost, too light and some detail will be washed out].
Move all three tall cones a little to the right or left as necessary. When you
are happy with the image fiddle with the RGB sliders until all three cones
align and present as one cone. This is essential in order to prevent the
whole image developing a colour cast. You can move any one of the sliders
a tiny amount if you want the image a little bluer, redder, or greener.
Note the tick box that allows you to move the sliders individually or
all together. I always move one at a time, as they don't all move
precisely together when set to do so.
When happy, shift from RGB to luminance. You can use the three sliders to
bring out or subdue detail in the image. Don't move the sliders that are set to zero. The middle one will be set at about 33.3 or thereabouts, move it
the smallest possible amount in both directions and see if either adjustment
improves the amount of visible detail. Do the same with the other two
sliders. If you want to darken the sky a little, move the top one from the default 80 to 90 or a hundred.
All these adjustments will help you get the best image you can until you
become proficient with processing in PS, Pixinsight, or other software.
Hope this helped.
raymo
Hi Rick. I have attached a light of each of the sub lengths I took from both the dark sky and my backyard in town. Should be 2 min, 1min & 30sec subs.
And I had an inkling my problems had something to do with the tinkering in DSS before I export it out to Lightroom/ PS. Even in DSS the image generally has quite a green colour cast to it. so I try and move the green slider slightly lower than the red and blue to compensate. As for the luminance side, I rarely if ever touch that.
And I thought i'd read somewhere that bias frames weren't really necessary as the darks contained that info already. thanks for confirming that raymo.
I've also added the DSS image right after stacking. This was my lights and darks taken from dark skies. 1hrs data worth of 2min, 1min and 30 sec subs. Have adjusted the rgb sliders to brighten the image as it was very dark. Green colour cast is quite strong as well even though it didn't show in single subs.
If you enlarge the histogram image you will see the green poking
out of the bottom right of the cone, and the three cones are aligned; if
you move the green cone a bit to the left, that green patch should disappear
and the green cast along with it. The black cone should not show any colour
protruding from either side.
Remember biases are only unnecessary if the darks match the lights.
raymo
Looks like you're getting good data on the single subs. Are you taking raws or jpeg's? Here is an experiment with stacking only no post processing at all.
I ran 3 x 30sec iso 1600 subs through DSS and saved the result without any processing at all, just the stacking.
1 single sub, 2 DSS screen grab, 3 Stacked result, 4 and 5 noise reduction.
The whole stacking thing is not about making the image brighter but more about reducing noise so you can "stretch" the image more through processing to bring out more detail.
Start out simple and work your way up to complicated, I'm still at the simple end myself, but learning something new all the time.
