Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Diagonal "brush stroke" artifacts?

astrophotography dslr beginner
  • Please log in to reply
14 replies to this topic

#1 Ohgren

Ohgren

    Lift Off

  • -----
  • topic starter
  • Posts: 4
  • Joined: 03 Jun 2020
  • Loc: Apache Junction, Arizona

Posted 05 July 2020 - 04:03 PM

Looking for some post-processing help.  I keep getting diagonal "brush stroke" artifacts in the stretched images.   The direction of the lines are not in line with apparent star motion through the image so it does not seem to be related to tracking error (i.e. there is very little horizontal or rotational change throughout the subs).  Attached image is a cropped section from the NA nebula taken on 6/23/2020

 

Using an un-moded Canon 80D with 400mm f/5.6 canon prime telephoto and an unguided ioptron skyguider pro tracker.  80 subs @ 30 seconds, 40 offsets, 25 darks, 25 flats.  stacked in DSS, stretched using curves in Gimp. 

 

All comments appreciated!

 

Second attempt-01-smaller cropped-smallerer.jpg



#2 17.5Dob

17.5Dob

    Fly Me to the Moon

  • ****-
  • Posts: 6,403
  • Joined: 21 Mar 2013
  • Loc: Colorado,USA

Posted 05 July 2020 - 04:30 PM

Dither.....That's a "Text Book" example of Walking Noise aka Correlated noise..It's caused by the slow drift in your tracker over time. DSS is stacking the stars and your hot pixels/noise are getting smeared into streaks.


  • Ohgren likes this

#3 Ohgren

Ohgren

    Lift Off

  • -----
  • topic starter
  • Posts: 4
  • Joined: 03 Jun 2020
  • Loc: Apache Junction, Arizona

Posted 05 July 2020 - 06:44 PM

My first impression was that there was not enough X translation in the tracking error to cause this.  I went back into DSS and looked at the dX and dY values for the subs.  Sure enough, the star motion from the lowest dY to the highest dY is right in parallel with my "brush strokes".  I guess my options are to

1. Learn how to hand dither on the fly.

2. Upgrade to a guide scope and PHD2.

3. Explore the possibility of enhancing the noise to an "artistic effect" smile.gif

Thanks for your help, Dave.



#4 BQ Octantis

BQ Octantis

    Mercury-Atlas

  • *****
  • Posts: 2,985
  • Joined: 29 Apr 2017
  • Loc: Red Centre, Oz

Posted 06 July 2020 - 02:30 AM

G'day Ohgren,

 

Fixing walking noise in post is a tall order. But "hand dithering" is easier than it sounds. Since walking noise is fixed pattern noise smeared in the direction of right ascension drift, you just offset your pointing by 10's of pixels in declination ≥3 times (equally spaced in time) over a session. You might also try using twice as many darks and closer matching the temperature of the darks with the lights to reduce the residual fixed pattern noise that is being walked. Finally, make sure to use sigma-kappa reject stacking to get rid of the outlier hot pixels.

 

Cheers,

 

BQ


  • Ohgren likes this

#5 calypsob

calypsob

    Fly Me to the Moon

  • *****
  • Posts: 5,529
  • Joined: 20 Apr 2013
  • Loc: Virginia

Posted 07 July 2020 - 03:09 PM

My first impression was that there was not enough X translation in the tracking error to cause this. I went back into DSS and looked at the dX and dY values for the subs. Sure enough, the star motion from the lowest dY to the highest dY is right in parallel with my "brush strokes". I guess my options are to
1. Learn how to hand dither on the fly.
2. Upgrade to a guide scope and PHD2.
3. Explore the possibility of enhancing the noise to an "artistic effect" smile.gif
Thanks for your help, Dave.


Use apt as your image capture prgm and you can dither without an autoguider

#6 DanH.264

DanH.264

    Explorer 1

  • -----
  • Posts: 57
  • Joined: 24 Jun 2020
  • Loc: AZ, USA

Posted 08 July 2020 - 12:56 AM

I find that more exposure and less stretching cuts down on walking noise, even if that means using a higher ISO.


  • Ohgren likes this

#7 the Elf

the Elf

    Mercury-Atlas

  • *****
  • Posts: 2,553
  • Joined: 06 Sep 2017
  • Loc: Germany

Posted 08 July 2020 - 12:47 PM

Dither is usually performed in both axes. An astro tracker has got only one motorized axis, so automatic dither will not give you the result you expect. Manual dither means you need to move the ball head on top of the tracker in dec. Some 20+ slightly different positions eliminate the problem. On the long run you need a 2 axis device, i.e. a mount to get rid of it.

The 80D is not a bad camera and should not be that noisy. The short exposure time adds to the problem. If you use this camera with longer subs you should not have that much noise. Which of course does not work on a tracker.

For the time being better processing might be a good way to deal with it. If you have the bandwidth to upload all your subs, bias and flats I could give it a go in PI just to figure out how much can be removed by processing. (No need for the darks, the effect is very small with such a short exposure time unless you had tropical temperatures)


  • Ohgren likes this

#8 Ohgren

Ohgren

    Lift Off

  • -----
  • topic starter
  • Posts: 4
  • Joined: 03 Jun 2020
  • Loc: Apache Junction, Arizona

Posted 08 July 2020 - 09:35 PM

Dither is usually performed in both axes. An astro tracker has got only one motorized axis, so automatic dither will not give you the result you expect. Manual dither means you need to move the ball head on top of the tracker in dec. Some 20+ slightly different positions eliminate the problem. On the long run you need a 2 axis device, i.e. a mount to get rid of it.

The 80D is not a bad camera and should not be that noisy. The short exposure time adds to the problem. If you use this camera with longer subs you should not have that much noise. Which of course does not work on a tracker.

For the time being better processing might be a good way to deal with it. If you have the bandwidth to upload all your subs, bias and flats I could give it a go in PI just to figure out how much can be removed by processing. (No need for the darks, the effect is very small with such a short exposure time unless you had tropical temperatures)

@the Elf,

Thank-you, that is a very kind offer.  At present I don't have sufficient cloud storage capacity (or upload bandwidth) to accommodate all the light files.  I did upload the 2GB TIF file that is the output from DSS here.

A couple of possibly relevant notes:

  1. I was hoping to enhance the resolution of the "wall" feature in the North America Nebula so I used the 2x drizzle in DSS.  I'm not sure what the noise implications of this decision are.
  2. The images were collected in central Arizona on a typical summer night when night time temps ranged from 90 down to 80 degrees F. so, yeah, I guess that would count as "tropical".  I collected the dark frames after the lights (as the night was cooling) so it occurs to me that thermal noise reduction from the dark frames may have been ineffective.

In the not-to-distant-future I hope to add autoguiding which will allow me to take longer exposures.  Cooler temps will, as you say, no doubt help as well.



#9 the Elf

the Elf

    Mercury-Atlas

  • *****
  • Posts: 2,553
  • Joined: 06 Sep 2017
  • Loc: Germany

Posted 09 July 2020 - 01:08 AM

Link is locked for me. Sent you a request.

 

Many people use 2x drizzle and in most cases it makes no sense at all. If you have optics that resolve more than the sensor and if you have an imaging scale that is coarse compared to your seeing and if you have enough subs drizzle helps to bring out more resolution. The first condition does not hold if you are using a camera lens because these are designed to have resolving power that matches the sensor and does not outperform it. The first condition also does not hold if you use an entry level scope. Only if you have well corrected optics like a Tak and an older CCD camera with 10 micron pixels the optics outperforms the sensor.

The second condition only holds if you have an imaging scale of 3 or 4 arcsec per pixel and your seeing is perfect and the object is high in the sky. Yesterday I aligned an image of the lagoon nebula taken at 420mm focal length 2 years ago to an image at 1100mm taken this summer. At my loc the nebula is about 10° above the horizon. Guess what, there was no more detail in the long FL than in the short.

I tend to say 95% of all amateurs do not need drizzle. You don't need it because the Canon lens is not made to outperform the Canon sensors but to match it. All you do is quadruple disk space and processing time. Sorry.


  • Ohgren likes this

#10 sharkmelley

sharkmelley

    Gemini

  • *****
  • Posts: 3,346
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 09 July 2020 - 01:42 AM

I tend to say 95% of all amateurs do not need drizzle. 

On the other hand, I think anyone using a DSLR or one-shot-colour camera should be using CFA Drizzle (a.k.a. Bayer Drizzle) just to cut out the colour interpolation performed during de-bayering.  Of course that also requires dithering during acquisition.

 

Mark


  • Ohgren likes this

#11 DubbelDerp

DubbelDerp

    Viking 1

  • *****
  • Posts: 846
  • Joined: 14 Sep 2018
  • Loc: Upper Peninsula of Michigan

Posted 09 July 2020 - 09:04 AM

Not that it helps in fixing this in post processing, but the walking noise is essentially the result of imperfections between your lights and darks. Matching the temperature as closely as possible between the two will drastically reduce the residual noise in the light frames which leads to this walking noise. Since your 80D has a temperature sensor in the camera body, you can take a look at the camera temperature in the light frames by a program that lets you see the EXIF data in the RAW files, such as exiftool. This might let you get a better set of darks if you can set up in an appropriate temperature.

 

I've recently started guiding and dithering my Skyguider Pro in RA-only, and it does make a big difference in the amount of walking noise, IF you've matched your lights and darks within a couple degrees C.


  • Ohgren likes this

#12 ChristopherBeere

ChristopherBeere

    Ranger 4

  • -----
  • Posts: 303
  • Joined: 24 Jul 2019
  • Loc: London UK

Posted 09 July 2020 - 09:51 AM

Yeah that's the dreaded walking noise.

 

You can sledgehammer it to an extent in post with agressive cosmetic correction but I usually just chuck it in the bin.

 

The only real fix is proper temp matched calibration, get the tracking sorted and dither heavily.


Edited by ChristopherBeere, 09 July 2020 - 09:52 AM.

  • Ohgren likes this

#13 DanH.264

DanH.264

    Explorer 1

  • -----
  • Posts: 57
  • Joined: 24 Jun 2020
  • Loc: AZ, USA

Posted 09 July 2020 - 01:38 PM

I'm in central AZ also, where the temps can easily drop 15C on a clear night.  I recommend shooting half your dark frames before lights, and the remaining half after.  Use the average of all the darks to subtract from your lights.   

 

The only way to get a closer temperature match is to shoot shorter groups of lights and darks to process together, and average the results.


  • Ohgren likes this

#14 the Elf

the Elf

    Mercury-Atlas

  • *****
  • Posts: 2,553
  • Joined: 06 Sep 2017
  • Loc: Germany

Posted 09 July 2020 - 02:16 PM

Here is a video I made some time ago how to build up a dark library at daytime putting the camera in the fridge. Most Canons put a temperature tag to the file. Once you have a set of darks you can use them from disk.

 

https://www.youtube....h?v=-6FXbs8fu_Y

Mind the time table in the description to jump in for a specific topic.



#15 the Elf

the Elf

    Mercury-Atlas

  • *****
  • Posts: 2,553
  • Joined: 06 Sep 2017
  • Loc: Germany

Posted 10 July 2020 - 01:51 AM

Downloaded your file and tried several things. The best I found was this:

- split the image to channels

- save each channel

- use cosmetic correction, auto detection, highs 3, low 0.03 -> fills up a lot of the dark parts with brighter pixels

- channel combination

 

For a near to final exposed image (like Pleiades or Orion) this might have worked. For the dim Ha clouds the reduction was by far not good enough.

Any kind of noise reduction failed. Again, if there was a bright structured eye catching object one might have treated the background and save the object. Here it does not work.

Sorry, I wish I had more for you. Dither is probably the only cure. You might try longer exposure times if your tracking is good enough and lower the ISO to avoid saturation. I don't think this brings out the Ha parts. It will give you a star only image. Probably not what you want. I doubt better darks will clean it all up. It will become better but if you stretch until you see the Ha clouds you see the lines again. Don't know how deep you want to go into it and how important portability is for you. If you get a mount with 2 axes and find a way to dither it unguided or add auto guiding and dither you will have a good result. Quite some $$$ to spend.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: astrophotography, dslr, beginner



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics