Horizontal banding noise on Canon 450D

[Penguin]

I have noticed some annoying horizontal banding that is totally ruining my shots and was wondering if any one could diagnose it. It's not an artifact of the stacking, I can see it in the raw subs and stacking just makes it worse even though it is also visible in the darks: surely subtracting the darks should then remove it? I have seen it before but its not normally got in the way, the last few sessions it has really dominated my images.

Last night, I took 1.5 hours of 3 minute subs of M106 at ISO800 and the stacked result looks awful, as you can see from the PI screenshot! I am taking lights, darks and flats (one of each attached and also in the screenshot). I have not grabbed any bias frames for ISO800 or I would have used those as well. I am stacking with DSS, following all its recommendations blindly, and then doing further processing in PixInsight. Someone on SGL recently pointed me at a tutorial for stacking in PI instead of DSS which was interesting and very long but didn't seem to resolve this issue and the process took too long to be worth it for the improvement it gave in the image.

I have seen a couple of other threads talking about banding. Some explicitly mentioned Canon SLRs but none actually resolved the issue. There was mention of dithering, which looks like an absolute ball-ache to do but if it is likely to fix my issue I would give it a go.

I am using an unmodified 450D through a Skywatcher 150P on an HEQ5-Pro, unguided. I am using Stellariumscope/ASCOM to control the mount and APT for the camera. Really, I want to understand why it happens in the first place, how to prevent it happening if that is possible, and/or how to remove it in processing. If anyone can help, I'd be hugely grateful.

Many thanks,

--- Alistair.

 

L_0467_ISO800_180s__20C.CR2

D_0434_ISO800_180s__24C.CR2

F_0482_ISO100_0-3s__20C.CR2

[Davey-T]

If you want to spend more money I think Noel Carbonies actions have an action to sort them, never used it though.

Dave

[Ouroboros]

In Pixinsight select Script then Utilities then Canon Banding Reduction. Always worked well for me. I also use an unmodded 450D. Good luck. 

[bottletopburly]

Star tools has a banding module vertical and horizontal, try for free

[Atreta]

3 hours ago, Penguin said:

There was mention of dithering, which looks like an absolute ball-ache to do

I don't know if it will get rid of the banding, but dithering is very easy with apt.

you only have to connect it to your mount, click  on the guide button and choose the guiding method. If you're not yet using phd2 all you have to do is choose apt dithering and auto dithering on. You can also choose to dither after any number of images. If you're using phd2 you choose the phd guiding option and that's it.  

[Ruud]

Topaz Denoise has a banding option. See here: http://www.topazlabs.com/denoise There is a free trial so you can try if it's worth the cost.

[Atreta]

There is also a debanding option under nik`s collection dfine2, i did  a test with your screenshot image.

It works as a phothoshop filter and best of all, it is free.

https://www.google.com/nikcollection/

 

[wimvb]

+1 for cbr in PixInsight. Use it before dbe/abe. Set highlight protection to 1. That should solve it.

[Penguin]

Thanks for these suggestions, I will definitely try the Canon Banding Reduction in PI on that image. I don't have Photoshop at the moment, so those plugins are a backup option. Will they work with PS Elements or would I need to subscribe to the full Creative Cloud?

I have had a quick look at the dithering options in an APT manual and that does indeed look very do-able. I like a setting that you can configure once and then forget. Not knowing much about ASCOM though, will it work alongside my current use with Stellarium through StellariumScope? I guess I will just try it and see what happens.

A quick Google for "canon banding reduction" has bought up loads of interesting links. That's my lunchtime sorted then!

[Ouroboros]

Good luck. As I say, CBR in PI works for me. 

You also asked what causes horizontal and vertical banding noise. I found an answer here on a photography site. Whether it's the right answer I don't know, but it seems a reasonable explanation to me. 

What I have noticed is that the problem seems to have got worse over time. Whether this is down to sensor ageing, I don't know. 

Incidentally, dithering makes no difference to the sort of horizontal banding noise I see. Maybe that's because the banding noise is low frequency ie changes slowly down the image. Whereas dithering seems good at removing small scale, high frequency noise. 

[wimvb]

As @Ouroboros noted, banding is caused by sensor readout. When you see it in a cmos image, the exposure is too close to the read noise, and the image is basically underexposed. The only sure way to get rid of it is increasing the exposure, or adjusting the ISO/gain.

[gorann]

18 hours ago, Atreta said:

There is also a debanding option under nik`s collection dfine2, i did  a test with your screenshot image.

It works as a phothoshop filter and best of all, it is free.

https://www.google.com/nikcollection/

 

Interesting, so I just downladed Define 2 for Mac to my Mac (I have PS CS5) and it was a lot of megabytes being installed but where do I find it when I open PS?

 

[Stub Mandrel]

Three minutes is along exposure in this warm weather and will accentuate the banding.

Noel's 'Astronomy Tools Actions'  will work remarkably well.

When I started I used an ancient 10D that was even worse, Noel wrote me a custom version of the script that worked wonders.

[Thalestris24]

I think I'd be inclined to check it could be being caused by electrical interference eg proximity of usb cable to mains cable. Where it's possible, prevention is always better than cure

Louise.

[wimvb]

28 minutes ago, Thalestris24 said:

I think I'd be inclined to check it could be being caused by electrical interference eg proximity of usb cable to mains cable. Where it's possible, prevention is always better than cure

Louise.

Absolutely correct. When the READ pattern of the camera shows through, the image is underexposed. But the banding can also be caused by electrical interference, or ground loops. The banding will then show up even if the exposure is long enough.

[Paul67]

Nik collection can also be installed with GIMP, I did just that a couple days ago.

[michael8554]

Hi Alistair

I've never had horizontal banding on astro modified 450D, 1000D, and 500D, usually 10 minute exposures at ISO800.

But my latest 600D does have horizontal banding, at 10 minutes ISO 800, and 5 minutes ISO 1600.

My first two sessions with the 600D were on the recent hot evenings, sensor temperatures were up to 28C according to EXIFLog.

But my 500D at 30C sensor temp last July hasn't had banding.

I've always had the DC power lead in the same loom as the camera USB, but I agree with Louise that this could be a source.

FitsWorks (free) has a "Lines to Equal Values" filter that removes the banding (also has useful gradient removers etc).

Michael

 

 

[Atreta]

4 hours ago, gorann said:

Interesting, so I just downladed Define 2 for Mac to my Mac (I have PS CS5) and it was a lot of megabytes being installed but where do I find it when I open PS?

 

it should be under the filters menu, maybe mac has it on a different place?

[Penguin]

Interesting, I tried the CBR script in PixInsight on my stacked image and it has significantly improved the banding, but of course that has then allowed other problems to show through, notably some linear diagonal noise and also a little vertical banding. I wonder of running CBR on the individual subs prior to stacking might help further. As to the diagonal noise, would dithering reduce that?

I've noticed some people saying that the 3 minute exposure might be causing more banding than shorter subs and that the warm nights might also be a factor, both of which make sense as I normally only do 2 minute subs for fear of star trailing, and I have only really noticed bad banding recently and not so much on the colder nights earlier in the year.

Also, would taking many more subs also help? I tend to only take an hour or two of lights and then only stack around 2/3 of those after discarding ones with planes, satellites and star trails. Maybe I will have another go at this target on the next few opportunities.

Anyway, here's what CBR did (unstretched tif and a quickly stretched jpeg)

 

M106 flats darks-CBR.tif

[wimvb]

20 minutes ago, Penguin said:

I wonder of running CBR on the individual subs prior to stacking might help further.

Put the images in an image container, and apply CBR to that

21 minutes ago, Penguin said:

As to the diagonal noise, would dithering reduce that?

Yes, this is sometimes called walking noise. It's due to leftover stuck pixels from calibration (although it can also be caused by flats). Dithering in both RA and DEC will prevent it.

24 minutes ago, Penguin said:

I tend to only take an hour or two of lights and then only stack around 2/3 of those after discarding ones with planes, satellites and star trails.

More data will always improve your image. But if there is a systematic fault, such as banding, then no matter how many subs you take, you won't get rid of the problem.

You can safely use all subs in PixInsight stacking, if you use large scale pixel rejection. Don't set the rejection parameters during image integration too aggressive, since that will increase noise.

[Penguin]

Hmm,

I have created an image container and added all my raw images but when I go to scripts->utilities->CanonBandingReduction, it says "No active window. Terminating script". So how do I apply the script to the images in the image container?

[wimvb]

8 minutes ago, Penguin said:

Hmm,

I have created an image container and added all my raw images but when I go to scripts->utilities->CanonBandingReduction, it says "No active window. Terminating script". So how do I apply the script to the images in the image container?

Sorry, my bad.

You have to put CBR in a process container:

• open any image and apply CBR to it (set the settings the way you want them in your sequence of images).
• Open the image history, select the image that had CBR applied to it, and pull the process container (small triangle on the left side of the history panel) into the workspace. This creates the process container icon with the CBR script in your workspace.

You can now close the image, no need to save it.

• Next, open a new image container and load it with the images you need to apply CBR to. Also set an output directory.
• Now pull a new instance of the image container (small triangle) over the process container icon.

This should do the trick.

[Marci]

22 hours ago, Atreta said:

it should be under the filters menu, maybe mac has it on a different place?

S’in the same place on Mac. Photoshop on Mac and PC have pretty much 99.9% identical user interfaces.