This was my meager effort for the 17th Snake Valley camp. While aeroplanes weren't flying through the FOV, clouds were pesky. A few dark frames were taken between clear patches.
Unguided.
Image was taken with a cooled Canon 1000D at a regulated -5C, but is only 19 frames at 3 mins iso800, 200mm 1.6crop factor, f5.7.
Camera is full spectrum mod with an Astronomik clip-in UV/IR filter.
Processed in Pixinsight - manually.
Second image post processing in StarTools
View from a distance - was processed on a laptop, so it might look a little overdone on a good screen.
That is very nice indeed, especially given the total integration time. Any difficulties when using your cooled camera out in the field? (I am considering such a mod for my 1000D at some point in the future)
Thanks Richard and Bo. I am amazed by the data too. More than I suspected. I have an iso1600 set as well - on your recommendation Richard - yet to process.
No problems with the camera - it performed as designed/predicted. A small amount of condensation, for a short time until the sensor defogger came to temperature. I noticed this when focusing on Sirius with a diffraction mask.
The preset worked as planned - air temp was <16C it set and ran at -5 for hours.
However, I am considering placing a small capacitor mosfet gate to ground, because the dark frames are showing very slight differences - my hunch is very minor electronic interference in the PWM line.
They are beautiful Rowland and I love the M42 colour and the extra detail in same in the second photo. Although to my aesthetic sense the lack of detail in the M42 in the first doesn't matter, it's still beautiful.
Thanks Trevor. It's just a beautiful region of the sky - Oh for more data. Working on that right now.
Richard I just checked my images. The condensation shows up in images taken just prior to dark. Minutes after reaching setpoint of -5 and not during focus. All gone a few minutes later.
This is a combination of iso1600 and iso800 frames, 40 in all, approx 2 hours - different apertures, f5.7 and f6.3. I played around with these over 2 nights
No dark frame calibration - bias and flat only. I did several test runs and frankly, the difference was negligible. Hot and cold pixel rejection was no problem even without the darks. I think the result is better - not sure what's going on with the darks?
The first post processing task was to resample the image for display, which reduced noise levels considerably and then process, instead of the other way round. The full size image is saved, so nothing lost.
I will revert back to an aperture ring to get rid of the spikes in future.
Here is a link to the final version. Different orientation and much better noise control/reduction.
Preprocessing in Pixinsight - Post processing in StarTools. I noticed that the the Integration tool selection, "Average with Scaling" produced odd results with this set and reverted to "Average" I also increased the High pixel rejection value with Linear Clipping for the cleanest result.
Image was first reduced by half with the StarTools Bin module, which further added to better noise reduction.
If anyone is interested, Star Tools workflow; as follows
EDIT:
Develop/AutoDev
Crop
Bin 50%
Develop 99.5%
Wipe defaults
Develop 99.5%
HDR Reveal - Detail size range 256 - default is 64
Life - Isolate - 4 layer inverted star mask - Power of Inverse
Colour - 150% saturation - Top and Low End Saturation ~6.0
Turn off Tracking and accept default noise reduction.
+ HDR Optimize Brighten Dark; and
Colour Saturation, with 4 layer exclude purple mask
Thanks John and Carl. 1000D is surprisingly good. I'm surprised too. I think Richard Muhlack's 20 hour effort with a 1000D and the quality of the image shows what can be done with it.
Looks really good Rowland.
You've got quite a bit of the brown dust.
I like the final version, colour balance looks spot on.
Nice to see the star tools work flow.
Was going to ask if you used the wipe module as it looked like there might have been gradients but looks like you already did.
Perhaps a bit more contrast? Should make the dust pop out .
2 hrs is a decent integration time for a dark site.
Is it a bright star at the bottom right?
Also helps to crop before developing to remove stacking artifacts which you mustve already done.
Hi Alistair. Thanks. I guess it could do with more contrast, but I am satisfied with it as it is for now. Areas of low SNR are poorly affected in an image like this - it needs more data to give it that extra push.
My motto is, work within the limits of the data, don't push it too much - get more. I'd like to say it's free, but that's not really true of this hobby. Photons are however.
This image started out at the standard 1000D 3906 x 2602, was cropped and then reduced by half to 1800 x 1095, which is a good noise reduction strategy, particularly as the image will be displayed at near that resolution. I left crop out of the workflow by mistake - will fix.
Used wipe - there is a degree of glow in the image which is intentional - it's light and airy spacey.
There are 1 or 2 bright stars in the bottom right corner and also the water mark.
Am just trying to understand the concept, I've read about scaling down an image to reduce noise, but in your case, you've cooled the 1000d to reduce noise, so why scale it down? I believe the darks show very little noise at -5C?
Is there a lot of difference between the scaled and native res image? will be interesting to see the difference in terms of noise.
Processing, contrast, etc is all subject to one's taste and preference, so understand your approach.
Reason I mention a full res image is in case you wanted to print an A3 poster. would look great on the wall and contrast with strong blacks will help with Glossy prints!!
Alex Cherney recently took a superb image of the Crux area with a diffusion filter for star colour. wonder if that will help with reducing the spikes but preserving star colour. not sure how it works, but thought I'd mention it.
Hi Alistair. Thanks for your feedback. It's always good to get a critical appraisal and improve on things.
I did a few trial runs and found that the dark frames were not helping - I am looking at that to see what gives. This is bias and flat calibrated only.
The noise is read noise from what I can tell. A function of exposure time in this case. The subs need to be longer to fill the background with more signal. Dark sites typically need longer exposures and I have under estimated on this occasion - to some extent I have attempted to compensate with a higher iso set.
Cropping heavily removed the dithering overlap and reducing the size of the image, which is intended only for forum display, further reduces the effects of read noise and speeds up post processing time. Once I get more data and happy with the result will I consider a print - early days yet.
The iris spikes are interesting if not overdone - on other images I used an aperture ring. This is pretty much a test run for the cooling system - and I'm not convinced of the effectiveness of aperture control at the front end of the lens. Don't know. Maybe there's nothing wrong with it.
Must admit though, I'm surprised by the effects of the darks - perhaps this is a case of read noise overwhelming noise levels at -5C where dark current is typically low.
I'm still a learner. Largely experimental. However, the strategies I'm using and more experience should produce better results - needs a bit of refining - still not entirely happy with the result.
Pixinsight did a good job of pixel rejection with a little help - rejection rates were upped a bit producing a smoother image than previous tries.
Are you able to post a dark frame? Surprising that its contributing to noise.
Is read noise a function of exposure time? I thought it was per read event, that is noise generated whilst reading out values, hence longer subs to reduce read events. I could be wrong.
I know you arrived at the -5c value based on a lit of trial and calculation but looking at graphs showing noise vs temp, -5c and -15c have a marked difference in noise. Ive seen this with my sbig stf8300m. -5c is borderline unusable but -15c is.
Have you tested pushing the cooling to -15 and comparing 5 min darks? Will be worth testing.
And finally, startools does a very good job of controlling and reducing noise when turning tracking off. This produces a smooth image as well. Have a go with the beta release, Ivo mentioned it handles develop and noise better.
Cheers
Alistair
Last edited by alistairsam; 12-03-2014 at 09:03 AM.
Are you able to post a dark frame? Surprising that its contributing to noise.
Is read noise a function of exposure time? I thought it was per read event, that is noise generated whilst reading out values, hence longer subs to reduce read events. I could be wrong.
I'm just guessing about the noise issues and need to look at this a little closer. I will happily post my findings once I have assured myself that the problem is not operator generated.
I dont understand how dark sites need longer exposures.
The sky background values are much lower at dark sites and dso's easily pop out of the background whereas from light pollutef sites, you need a number of long exposures to get the dso to reveal any sructure. I would've thought the snr from dark sitss would be higher.
For example, I took a 45 sec test image of m83 from the asv dark site at heathcote and I was surprised to see the amount of detail revealed , whereas the same from the city would be washed out by lp.
The dark site vs light site is counterintuitive - there are a number of discussions about this. I understand the converse to be true.
I know you arrived at the -5c value based on a lit of trial and calculation but looking at graphs showing noise vs temp, -5c and -15c have a marked difference in noise. Ive seen this with my sbig stf8300m. -5c is borderline unusable but -15c is.
Have you tested pushing the cooling to -15 and comparing 5 min darks? Will be worth testing.
-5C is a practical level of cooling, as is 0 and 5C. These presets work within the capability of the design. I have no intention of attempting temperatures below -5.
We may be at risk of making comparisons here. While a cooled CCD has capabilities that exceed a cooled DSLR, fundamentally, they address different needs for the same purpose and therefore demand slightly different strategies in several respects.
And finally, startools does a very good job of controlling and reducing noise when turning tracking off. This produces a smooth image as well. Have a go with the beta release, Ivo mentioned it handles develop and noise better.
I am familiar with ST noise reduction and have used the beta release frequently.
Cheers
Alistair
I wish I had the time to discuss at more length.
Clearly, testing shows that there are a few issues to be resolved. I suspect these are user generated, rather than the equipment.
spacey.
