Background streaks, Help!

[DaveS]

Hi

I've found that I'm getting diagonal streaks in the background of my recent pics.

I've tried stacking with light, dark, flat and bias frames using median, sigma-kappa and entropy weighting. I've tried stacking without flats, without darks and without bias. I've tried stacking just lights. Nothing seems to work.

I've used DPP to look at the original raw files and the 16 bit TIFFs I use for stacking but can't see any trace of streaking.

The exposures are 32 x 150 secs lights and 16 x 150 secs for darks. Flats are another problem. I don't have an EL panel so I'm using the white tee-shirt method. At first I thought it was the weave showing up but it's there without flats. I'm also having inverse vignetting problems as well, but that's another topic.

Kit used as in my sig, with the CLS filter (Also in place for the flats)

Any ideas?

[steppenwolf]

I am not entirely sure why you are getting this but I suspect a poor download from the camera if you are using a USB lead. If you are using a lead, try changing it or removing any USB hub in the system between the camera and PC.

I do know how to fix it to some extent though - rotate the image 35 degrees counter-clockwise and run Noel Carboni's 'Horizontal Banding Noise Reduction' action on the image. Its a kludge though and you do need to get to the source of the issue really.

[DaveS]

Thanks

Because the current version of DSS doesn't support 550D raw files I've been opening the card in DPP (Via a USB 3) card reader and converting the files to 16 bit TIFFS, transferring them directly onto the computer.

The thing is, I'm not seeing the streaking on any of the raw files or TIFFS when I open them in DPP.

The JPEG I posted was somewhat exaggerated to show the problem. but it's still there.

[Jessun]

What does the stack look like without darks or bias? First sub vs last drift wise?

That should reveal how you've been drifting due PA error. My money is on that the hot pixels when left in the image will seem to wander in the same mannor as the streaks. At least that's what I found when using a DSLR making the mistake of not dithering and not allowing the camera to cool a bit between exposures. It's like all pixels became a tiny bit hot, and stacks would present this as streaks.

/Jesper

[DaveS]

It's there in the basic Light frame stack as well (Using Entropy Weighted). I'm looking at building up versions adding Flats, then Darks and Bias frames. I'll post images when I have something to post.

[StuartJPP]

There is a beta version of DSS available here which supports more recent camera models (works with my Canon 1100D...

Latest beta available 3.3.3 beta 51

http://tech.groups.y...DeepSkyStacker/

[DaveS]

Thanks, giving it a go now.

[DaveS]

Well, it's done a better job of stacking, but the streaks are still there. Argh!

Here is the full frame as stacked, with only some excess red taken out, and a cropped version.

I looked at my original Leo Triplet image and that has streaks as well, but this time they're running horizontally.

[steppenwolf]

That's useful that you have the streaks running in different directions! As suggested by Jesper, check a test star from the first and last images from each session (M51 and Leo Triplet) and determine whether or not this test star has drifted from start to finish. If it has, does its direction of movement match that of the streaks? If it does then this could be noisy pixels smearing the view. If not, we need to look further.

[Mark72]

Have the same issues Dave,

Pretty much gave up on this. on PixInsight forum there I was suggested the following solution. Have not tried it yet:

"If it is shown in each frame, then try taking super flats. Just point near to the zenith, far away from deep sky objects, and take images as long as your light frames. Use a very high dither distance. Later, integrate the image using and aggressive rejection to high values, to delete all stars (of course, do not align them)."

Mark

[IanL]

I am not entirely sure why you are getting this but I suspect a poor download from the camera if you are using a USB lead. If you are using a lead, try changing it or removing any USB hub in the system between the camera and PC.

The contents of the image download will not be affected by a USB cable per se. When the image is transferred from the Camera to the PC it is digital data and the USB protocols have error correction built in. A dodgy cable will not add noise to the digital image; at best it will slow down the transfer, at worst you won't be able to transfer. If you do get the image across, it will be exactly as digitised on camera.

Some types of camera may be susceptible to externally induced noise prior to or during the analogue to digital conversion process. A noisy cable can certainly lead to poor images on cheap webcams, but I haven't really noticed any problems with DSLRs. That said, it may affect the fixed/non-fixed pattern noise described below, but I haven't really investigated whether it does. In any event I just try to separate all my electrical and data cables as much as possible and use ferrite chokes to cut down the amount of induced noise between cables and hopefully prevent them acting as conduit for pumping interference in to the analogue side of the camera's process.

It is good practice anyway, since you definitely can have data transfer slowdowns, failures and disconnects due to electrical noise interfering with digital cables. (We moved a bunch of people in to an office that already had all the network cabinets cabled with shielded twisted-pair cabling by the previous occupiers. Shielded patch and fly leads were a lot more expensive in those days and so we just used the normal unshielded cables to hook all our PCs and servers up. Cue several months of network problems and disconnects. When we discovered there was a giant police radio transmitter on top of the building next door we wised up and switched the UTP over to STP cables and the problems went away - lesson learned there ).

:

I do know how to fix it to some extent though - rotate the image 35 degrees counter-clockwise and run Noel Carboni's 'Horizontal Banding Noise Reduction' action on the image. Its a kludge though and you do need to get to the source of the issue really.

:

That's useful that you have the streaks running in different directions! As suggested by Jesper, check a test star from the first and last images from each session (M51 and Leo Triplet) and determine whether or not this test star has drifted from start to finish. If it has, does its direction of movement match that of the streaks? If it does then this could be noisy pixels smearing the view. If not, we need to look further.

I'd concur that this is likely to key the problem. I suspect that the streaks are an interference pattern caused during stacking.

For one thing many Canon DSLRs suffer from what they call 'Canon Banding'. You can get an idea of what it is as follows:

- Take a bunch of bias frames, i.e. dark frames with the camera covered that are the shortest exposure you can set (usually 1/4000th of a second). Now stack them together without any kind of aligning. This gives you a bias frame, and if you stretch it in your image editing package you will usually notice a pattern of streaks on the image running along the long axis of the frame. (Technically this is what is known as 'fixed pattern noise', and is caused by the camera electronics when it is capturing and digitising the image).

- If you now take another bias frame (or maybe a longer exposure dark frame) and subtract the stacked bias frame, you will see that the streaks reduce but don't go away completely. Each subsequent frame you take and subtract will have a similar but different set of streaks. (I don't know what the technical term for this is, if there is one. I guess we could call it 'non-fixed pattern noise', since it has an obvious pattern but it varies from frame to frame).

- When you stack an image with banding (or any other semi-regular noise pattern), there are parts of the frame that consitently tend to be a bit darker and other bits that tend to be a bit lighter. Depending on how and how much the frame drifts between each exposure, an interference pattern will build up as some dark patches consistently subtract from other dark patches, some light patches consistently add to other light patches, and some dark and light patches consistently cancel each other out. In the end you get noticeably brighter and darker patches, streaks, trails or other artefacts aligned along the dominant axis of the drift like you seem to have.

- You don't say if you are guiding these images or not. If you are guiding and not dithering (or only dithering by a few pixels at a time as I would) I'd expect any Canon Banding to appear as a pattern of fine-scale 'streaks' along the long axis of the frame, exactly as per the bias frame or bias-subtracted frame above, along with a larger-scale pattern of 'bands' on the same alignment (which are basically made up of lighter bands where there are more/brighter streaks and darker bands where there are fewer/darker streaks).

- Your image suggests that the stars are drifting from the top left corner towards the bottom right (or vice-versa) between each frame, either because you are not guiding at all , or more likely because your image is dithering predominantly along the same diagonal axis between frames. The reason I say this is that in that case the banding streaks would then form a diagonal interference pattern as you stack due to the process described above.

- As suggested, check the first and last frames in the stack to see if the stars have moved in the diagonal direction the the streaks. If so, that is pretty definitive in terms of diagnosing that this the cause.

There isn't a 'root cause' solution for the banding problem that I have ever discovered. It is inherent in many models of Canon DSLR to a greater or lesser extent and so you just have to adjust your acquisition and processing techniques to compensate for it.

Things that may help:

- Guide your images and make sure there is no unplanned drift in any direction. I'd have thought you were already doing this if you have 150s subs though.

- Then dither your images randomly, you want the image to drift randomly up or down and left or right between each frame, not keep going right and down or left and up as seems to be happening here. (If you are using APT or BYE and PHD for guiding, you can auto-dither between exposures).

- This random motion between frames will reduces the build up of the interference pattern. You may need to experiment to find out how much to dither by for best results with your camera. I tend to dither by a few pixels at most, and end up with a classic canon banding pattern which I can deal with in software (see below). You may find that a much larger dither works better for your software, though be aware of any limitation in your stacking software if it cannot cope with stars moving a long way between frames.

- PixInsight has a 'Canon Banding' script which works wonders on these types of issue, but seems to function best where you have a pattern of streaks and.or larger scale banding running along the long axis of the frame. It's easy to apply to the stacked image and just this weekend I have taken images that had banding similar to interference on a badly tuned analogue TV, run it through the script and the banding is gone. (It doesn't produce a smooth, noise free-background, but it does produce a background which has a consistent level of noise across the whole frame, making it possible to use other tools to reduce the noise without leaving artefacts).

- I have to say though that dithering and using the banding script doesn't eliminate the issue completely, you tend to get a large scale 'grain' of darker 'holes' in a brighter overall sky background as the dithering will (overall) make the bright and dark parts of the banding move within a circle of a given diameter and some larger scale residual pattern tends to result. It is far less noticeable than the diagonal streaks you have now and easier to smooth out using noise reduction tools, etc.

[DaveS]

Hi

Been away today and only just come back to my thread.

I'm not guiding, only tracking. I've got an autoguider on my shopping list, but haven't jumped yet because:

1) My finances are in a bit of a parlous state ATM and I expect some big bills in the near future. And,

2) With London's dire LP 150 sec at iso 1600 is the most I can give without the frame totally washing out.

3) I'm also budgeting for the Astronomiser conversion for my 550D at some point soon.

I have thought of guiding and pulling the iso down, say to 400 which should increase my DR and colour depth while decreasing noise, but I would have to increase exposure to 600 sec per sub.

I'm using APT and have seen the PHD connection in the desktop so I may go that way.

Yes, I've noticed that drift in my stars. I think the drift on the earlier image with horizontal banding was more up-and-down.

I may have to print out this thread so I can read it properly away from the computer.

[Psychobilly]

Try stacking without the darks and see what happens....

Peter...

[DaveS]

Here you go, no darks, streaks still there.

[Psychobilly]

I Used to get bad streaking if I used mismatched darks so stopped using them..

I tend to just use lights, bias and flats and dither out the hot pixels...

Peter...

[DaveS]

I had a look at my M42 image and I couldn't see any streaking.

I'm wondering if the reason I'm seeing it now is because the images are so close to the background that to see them at all I'm also seeing any background artefacts exaggerated,

Now I'm thinking about how to avoid them in the first place, perhaps by pulling the iso down to 400 and pushing the exposure times up, with more subs and dithered guiding.

Problem is, I costed out a guiding package from FLO and got a figure the wrong side of £500 so it will have to wait a while.

[Jessun]

If we assume the streaks are indeed a representation of your drift, you could at no cost polar align a notch better next time, move your telescope a little bit - as often as possible, ie between each sub, and let the camera cool between subs. The trick is to force some randomness into the mix.

The data you have so far will be a hard nut to crack in processing as far as I can tell, but adding more data over a couple of nights should reduce the worst effect using some variant of a sigma routine on the combined stack.

Guiding does add to the overall cost, but can certainly be done on the cheap. I have used several webcams and PHD with great success. A webcam can easily be attached to a finder scope. I'd throw out an estimation here, but £50 should get you going guiding. And it's worth it!

/Jesper

[DaveS]

OK, thanks for the advice.

The next clear night I'll have a go at getting a better polar alignment, though I'm not sure how to drift align without a view to the East or West. I can do an azimuthal alignment on an equatorial star near the meridian. But beyond that I'm stumped.

If I get a game-controller hand-set I should be able to shift between subs, esp if I increase the interval between them. I dunno, roll a die and shift depending on number? The only thing is I've been able to set up an imaging run and go away but if I have to sit with it, shifting every 2 min or so...

I'm not sure how I'd get the cost of a guider down to £50, allowing for a web-cam, finder 'scope and WO finder rings.

[Jessun]

Just use a standard second hand finder scope in its attachment shoe and gaffa tape for the camera once you figured out where it comes to rough focus. No need for fancy rings. OK, sometimes it will be a struggle to find a guide star with a web cam, but that's life for many even with expensive OAG's depending on how it's all bolted together.

Roll a dice is as good as anything! It's just about not having a continuous trail in the final stack. You wouldn't know from the onset how you're drifting from session to session, so go in all directions. The only downside is that you'll crop the final stack, since all subs don't perfectly overlap anymore. It's still worth it!

/Jesper

[IanL]

I'm not sure how I'd get the cost of a guider down to £50, allowing for a web-cam, finder 'scope and WO finder rings.

Fifty quid may be a bit ambitious but 500 quid is way over the top, what were you being quoted for?

An Orion ST80 plus a QHY5 would cost 300 quid brand new (you may or may not need a set of rings for about 30 quid, mine said OTA only but came with a carry case, normal clamshell rings and a short dovetail). Second hand would be a bit cheaper, look for a reasonably short focal length 60/70/80mm achromat; it doesn't have to be top quality so long as it is fast enough to produce a reasonably bright image on the guidecam. The QHY5's tend to hold their value second hand and will go for 140 quid plus come what may.

You could probably shave a new setup down to nearer 240 quid by getting a QHY5 and a finder. Ridiculously (in my view) similar spec finder guider packages cost about £350 for something not as good as an ST80 in terms of performance (except weight), and if weight is the issue you can put your own kit together for 100 quid less that the pre-packaged finder-guider kits.

Second hand gear is always worth investigating as people are often selling off finders, ST80 refractors, rings and dovetails, webcams and guide cameras.

I'd avoid any kind of adjustable finder rings for guiding purposes; you do not need anything more than a rough alignment between the guider and the imaging scope to successfully guide, and adjusting screws, o-rings, etc. are just a source of flexure you can do without. If you are using a non-long exposure webcam it may be an issue getting a guide star; they are okay for really bright stars but a bit hit and miss if there isn't one near your target. You'd be better going for a proper guide cam from day one if your budget can stretch to it.

[nytecam]

Looks like progressive diagonal drift of field across sensor during the session due to polar missalignment and/or drive errors. Usually sorted by flats applied to each image.

[Jessun]

Camera Microsoft Lifecam Cinema £25.00:

http://www.ebay.co.uk/itm/Microsoft-LifeCam-Cinema-720p-HD-/281089764426?pt=UK_Computing_ComputerComponents_Webcams&hash=item4172416c4a

Finderscope £24.99:

http://www.ebay.co.uk/itm/Skys-the-Limit-6x30-finder-scope-with-90-degree-angle-/160993441830?pt=LH_DefaultDomain_3&hash=item257bf4f826

Gaffa tape £0.99:

http://www.ebay.co.uk/itm/50M-SILVER-GAFFA-GAFFER-TAPE-DUCK-CLOTH-TAPE-48mm-x-50M-X-2-Rolls-/190820381795?pt=UK_Packaging_Materials&hash=item2c6dc7f463

I'm sure it can be done even cheaper !

That aside, IanL, focusing the attention on a dedicated guide setup still falls short of the £500, so don't give up on guiding.

/Jesper

[DaveS]

I was working on the principle of "If a things worth doing, it's worth doing properly"

So:

Skywatcher ST 80

Starlight Express Costar

ADM dual Vixen saddle

ST4 cable

Came to just over £500 inc delivery.

I could try mounting the ST 80 on top of my Megrez 90 to save £135 I suppose, but I dislike piggy-back mounts.

And I have a dislike of flea-bay for some reason.

Gaffa tape's no problem, I have it by the mile.

O, and I've just posted a pic of my set-up in the Members Equipment section

I am not giving up on the guiding by any means, just that I may have more pressing calls on my funds.

[Jessun]

I was working on the principle of "If a things worth doing, it's worth doing properly"

Fair shout. That's a guide setup hard to rival.

I used a couple of webcams to get started and was buzy for an evening or two just gutting them and getting it all to work. It was a joy!

/Jesper

[DaveS]

I think my next step is to have a more solid go at polar alignment while debating my finances.

The data set I already have is, I think, a write-off still not too big a loss, as it's only 80 min.