How Can I Improve? ?

[JohnSadlerAstro]

Hi,

I've been into astrophotography for several years now, but I can't seem to make much improvement between images. Looking at the images in the Deep-Sky section with clean, white galaxies and perfect smooth backgrounds has got me thinking about what I'm doing wrong.

I know I can achieve better images with the setup I've got, and I really want to improve in time for the summer nebulae. Is there any advice anyone can give?

This is a diagram of my setup. My 1000d is modded. I use PHD drift align, and with that running my guiding is to within 2 seconds. I use APT for dithering and image collection, DSS for stacking, and GIMP 16-Bit for processing. A quick note at this point -- ive heard lots about StarTools and other programs, what is the best choice program which wont cost a few hundred pounds?

Collimation has always been a bit of a difficult one for me, I've tried using both a Cheshire and a laser, but I never seem to get properly coma or gradient free images.

 

These are the kind of results I'm getting--the main issues I can see are horrible backgrounds (dithering doesn't seem to be able to remove the dark frame scratches) and really grimy colour, but there is also coma and gradient/vignette which refuses to be corrected by flats. (I take my flats through some white fabric pointing at notepad open on my laptop.)

A big issue for me has always been imaging time--where I live the council turns off the lights each night at midnight. For some reason, although I will often have the setup running from midnight until 3 in the morning, everything seems to go wrong after an hour. For example, the guiding crashes, or a glow appears in the images and wont go away, or hundreds of aircraft seem to pass overhead and streak everything . 

As the setup became more and more complicated, it began to take more and more time to set up: last imaging session it took 2.5 hours from opening the garage door to starting the first proper exposure. This makes everything far more difficult and time-consuming.

Does anyone have any tips or suggestions on how I can solve these issues, or just improve my imaging in general?

Thanks for looking

John

[Owmuchonomy]

Hi John, there are so many answers to your key question.  I am not qualified to answer all of them but 2 things jump out at me.  The first that I am qualified to answer is your set up time.  I really struggled with this until I built my obsy.  I can safely say that was the most significant change to my whole  astronomy life.  It cuts the set up time to a fraction of that with other setups.  Secondly, I would suggest to make a big step change then it may be worth going the mono imaging system route and the necessary filters.  It doesn't necessarily increase your capture time as I am sure @ollypenrice will testify.  Other answers I will leave to those with considerably more image processing desire and time than I have.

[vlaiv]

Two tips come to mind (one is "free" and one needs quite a bit of cash to spend on):

1. Throw more imaging time to each image - this will help with noise issues and is always recommendation (but often not feasible)

2. Switch to cooled / set temp camera. This is obviously costly change, but it will help with much of problems you are talking about. You will be able to do proper dark correction and as a consequence flats will correct properly also. Noise will be reduced significantly.

There is one thing that you can do to change appearance of your images: Do a slight change in workflow. Use 32bit processing for your images (again Gimp is fine for that, but load in 32 bit fits files instead 16 bit png/tiff) and learn to do custom white balance / star color calibration. Also have a play with different color models for your processing workflow, you might try to do something like this:

Using gimp, load in RGB file. Decompose RGB file into LCH color space components. Use L as luminance and do stretch on it alone. Transform original RGB into RGB ratio - this means divide each or R, G and B components with sum of components, and again transform it into LCH. Multiply two L layers (one stretched and one from RGB ratio) and recompose with C and H components of RGB ratio image into final image.

Also you might want to do heavy denoise on RGB ratio image since it will provide color only.

[Jkulin]

I have to set up every night and I wanted to reduce my setup time down and this is my work flow and what I did to achieve it.

I couldn't locate an observatory in the garden because of the location of my home and the trees, so the only option I could do with out upsetting my wife was to build and install a permanent pier on a rough 0.75mx0.75m concrete block.

I then looked at how to power it all up quickly and safely, so I purchase two 13.8v power supplies and made up long enough cables to power my laptop and mount etc.

I fitted an extension socket to my LAN outside.

I made up my power cables so that they were permanently attached to my mount and similarly with my USB hub. I purchased Chroma USB cables to the correct length to cut down on fitting more than necessary.

All I have to do now, attached mount and two power leads, slide the OTA that I want to use onto the rail and connect my laptop, it's not perfect but I can be up and running in less than 15 minutes and packed away in less than 10

For processing I realised fairly quickly that for me PI was the easy option but I didn't realise how long it would take before I could produce okish images, I'm still a long way off, but I consider it better than Nebulosity 4 that I was previously using.

The only other thing that has improved my images was switching to a Mono camera and FW rather than my Nikon D800, cooled cameras make so much improvement to the noise.

HTH

[JohnSadlerAstro]

Hi,

I did think of getting the EQ5 Pier Mount, and fitting it under one of those make-your-own-compost bins. The main snag is talking the family and neighbours into allowing me to put an eyesore in the front garden!  

Although, with some strategically placed potplants, it might be "Entirely Invisibleᵀᴹ"  

John

[alexbb]

I set up each time I image, though I don't disassembly the mount, nor the scope/camera. I have an USB hub permanently secured with tape on the mount and the cables too. Also I marked the legs' positions and now I place the mount in the same place.

Each evening I bring out the mount, I put it in the same place, put the counterweight in the same position, put the scope in the same position, connect the cables to the cameras, bring out the power source and the image acquisition mini pc. Connect power and pc and fire up everything. Then I wait 1-2 hours for scope to cooldown. I then start the camera's cooling and do a 2 star alignment, a polar adjustment (since I put the mount in the same position, in the beginning I'm off less than 1 degree) and another 2 star alignment. Focus, goto, adjust until satisfied and then start imaging.

The part before cooldown takes ~5 minutes, the second part takes longer, perhaps 15 minutes, also because I let the camera cool slowly to 30C-35C below ambient temperature. As I'm imaging relatively close to the house, the temperature doesn't drop more than a couple degrees and I don't really need to refocus.

[almcl]

John

you mention StarTools, so as a very much non-expert user I had a play with your M51 jpeg image using it.  Result below - I think it might have produced a better  esult using an unstretched, original .fits file.  There were some unusual artefacts on the right hand side which, for speed I just cropped out, but I think there's quite a bit more data to be pulled out.  You can download a fully functional copy of ST from the website, for free - save is disabled (but screen grabs are possible) and there's no time limit, but you can see if it meets the need.  There is a bit of a learning curve and it's not for everyone but there are some good tutorials on how it works now and the creator, Ivo, is usually pretty helpful to newcomers.

In the image I looked at I can't see coma (didn't look very hard, in truth) although in the Horsehead Alnitak may be suffering from focus tube intrusion - I fixed this on my Explorer 200P with a cutting disk - but the puzzle I don't understand is why your flats aren't dealing with vignetting?  Maybe a topic for another thread?

 

[JohnSadlerAstro]

Hi,

The image I posted had used IRIS to remove the worst of the vignette--I'll restack and export as a FITS and link it, if you want

John

[happy-kat]

How many flats do you typically include in your stack?

Are you including darks these might be adding noise.

Worth trying dark flats with your flats.

[almcl]

46 minutes ago, JohnSadlerAstro said:

Hi,

The image I posted had used IRIS to remove the worst of the vignette--I'll restack and export as a FITS and link it, if you want

John

OK, I'll certainly have a go.  

[JohnSadlerAstro]

I'm not sure if the Fits worked, I think the program crashed half way through, but the TIF is 16 bit the same as my camera.  

Stacked.FTS

stacked.TIF

[almcl]

The .fits file appears incomplete, so I've used the /tif.

See what you think?  For speed I cropped the image quite a bit and binned it 50%.

 

[JohnSadlerAstro]

Wow!

How do you get it to look like that with Star Tools?

All I can get when i follow the steps in the tutorials is a horribly overstretched gradiented mess !

John

[almcl]

The workflow was

Autodev (to see what's there - no intention of keeping the result)
Crop (to remove edge artefacts)
Bin (normally to reduce oversampling on my rig but just to speed things up here)
Wipe
Develop  (84% Digital Development, 10 (ish) Dark Anomaly Filter)
Mask (Auto, Selection Mode Highlights>Threshold set the threshold to 47%)
Decon
Sharp (Amount 342%)
Clear Mask
HDR (Optimise hard)
Contrast (Defaults)
Mask the nebula
Life (Isolate, strength 75%, mask 39.9)
Colour (Cap green to yellow, then play around with red and blue sliders to taste)
Switch off tracking setting grain size to 12 px in the wavelet denoise option

At this point I used a new star mask and the repair module to make the stars a bit rounder but perhaps not necessary.

 

If it helps, I'll post the full log?

[JohnSadlerAstro]

Hi,

Thanks for listing the steps, that's really great I'll try out using Star Tools a bit more, and see if I prefer it to GIMP & IRIS. The result above certainly looks very different to what I'm used to from those 2 programs, but I definitely like it.

John

[happy-kat]

If your data has any weaknesses then don't use Autodev other than to look at for later cropping reference but don't keep it. If you keep it with not ideal data it will lead to poor results.

[JohnSadlerAstro]

!!!!!!! 

<Star Tools insta-buy> (well, would be insta-buy if I had the money!) And yes its embarrassing to still be on pocket money.  

Definitely a licence will be my next astronomy purchase. Meanwhile, I will mess around with old/new data and learn the ropes!

Thanks soo much for the processing details, @almcl I followed them through with some slight taste changes and wow.

Processing is indeed what part of the problem was with my images. (And this M51 image was a pretty mediocre lot of full-moon data.)

John

[carastro]

Well I just had a "go" at your Horsehead, just using the Jpeg posted in this thread (never ideal), and it's come out a lovely image using Photoshop.

I tried using GIMP when I was on holiday last year as I didn't have PS on my mini laptop, and found it incredibly difficult to use and could not produce any good results with it.  

Why don't you have a look out for a second hand old version of PS on sale on somewhere like Ebay.  Anything from CS2 onwards will be fine.

Anyway, this is how it came out.

Carole

[carastro]

I have now examined several of your images and your flats appear to be OK to me.  There were some gradients in the M51 image, but they did not look like Vignetting.

Carole 

[JohnSadlerAstro]

Hi,

As you say, GIMP is awfully difficult to get results with -- it's quite likely that its GIMP causing the trouble/exacerbating any vignette or gradient. I'll reprocess the 4 images above with Star Tools tomorrow, and see how they turn out.

John

[david_taurus83]

Hi. I too have to set up from scratch every session. The council don't do me any favours with turning street lights off though! I can be up and ready for dark in about 20 mins and maybe another 20/30 mins once dark to polar align, focus, get on target and start imaging. What is the most time consuming part for you?

 

Ps. Your pictures look great to me! Maybe try a different processing software? Have you calibrated your laptop/pc screen?

[almcl]

Looks like a result, John!

Well done.  

Good that Happy-Kat and Carole have joined in, H-K has come up with a very clever technique for reducing star bloat in individual colour channels when using StarTools, and Carole's use of Photoshop shows there's definitely more than one way to process the data!

As Olly P said recently, the processing can be the fun part, once you have the tools.

[happy-kat]

That's kind but I can't the credit for that the thread was started by another member then Ivo added to it and I took what they had done and tweaked the workflow further to suit my image.

[JohnSadlerAstro]

Hi,

The part which takes longest with the setup is the centering. It takes soo long . The main issue is that I have a non-GoTo mount, so I cant change the slew speed off my laptop. (which is connected to the setup from inside). I essentially go back and forth giving the handset a jab, then back in again to see which way the object went, take another image, then back out again and so on. It can take up to 45 mins to get the finding and framing right. OK, I can probably do some more work getting the finders aligned precisely for better results but it is quite annoying. I don't know if it would be practical to do the centering from indoors, but using the manual guide tool of PHD?

John

[wimvb]

You already have a nice set of images. Image processing takes at least as much effort as data collection, probably more. Imo, you need to invest in this part of our hobby; "there's no such thing as a free lunch".

I loaded the M51 image in PixInsight, and had trouble flattening the image. There was a gradient on the right that could be removed by using division as the correction method, which suggests a vignetting issue. But since it only shows on one side, I wonder if the secondary mirror is set up correctly. Otherwise, collimation looks ok to me.

There was also a brighter patch in the background around M51 that looks like dew build up on the optics (not sure that it is though, as stars weren't affected by it). I removed that with repeated iterrations of DBE subtraction. There is still some of it visible in the image, as well as a dust bunny.

I use this method to create flats, which works ok for me with my 150PDS and ASI174 cooled mono camera:

I put an embroidery hoop (my wife does needlework while I do my thing ) with 4 layers of white fabric on top of my scope which points at zenith. On top of that I put a 30x30 cm FLOALT led panel from IKEA at its dimmest setting. I expose at the lowest gain setting of my camera and typical exposure times are 0.2 - 1.5 seconds (L and B). The target average intensity is 25000 ADU in a 16 bit image, ie just below half maximum. I take 30 flats per filter and calibrate them with dark flats (the ASI has amp glow which is removed with darks).

Here's the processed image. After cleaning up the background, the noise looks as expected from a DSLR: mainly colour noise. But also a bit of a pattern to it (walking noise). You probably need to increase dithering. There's more to be pulled out than one would think. The faint outer halo of M51 is just above the noise floor, as are a few faint fuzzies. But the image would benefit from more data.