strange pattern

[ClausDuerr]

Hello everybody,

what could cause the strange pattern?
Photos done with an Canon 600Da, lights and darks matching.
darks look ok, lights too...

anything I should consider will integrating?

thanks alot
Claus

[Phil Leigh]

And the darks have the same ISO, temperature and duration as the lights?

Did you use the Batch Processing Script or did you do this all manually?

[Carlos Milovic]

How do your master bias looks like? How long are your lights?

[NGC7789]

I get this sometimes. I use a modified Canon 50D. It is my understanding that it is caused by linear drift between frames causing a regular pattern noise to form lines. I am not clear why normal calibration does not remove this noise. Apparently the correct solution is to drizzle during capture to randomize the drift and therefore the noise does not add up into lines. My equipment is not capable of drizzling but I don't don't get this problem every session so I'm guessing that depending on the polar alignment and position of the target in the sky maybe I get a hit or miss drizzle effect.

[pfile]

this is the infamous canon banding noise. check the following:

are flat subs well-exposed?
how many bias frames? need something like 200 bias frames to minimize injection of more banding into calibrated subs
similarly with the darks, the more the better as this pattern will be in there as well.

if you still see this, then you can try using Georg's CanonBandingReduction script.

generally this script is used on integrated images, but if your project was a multi-night image and your camera rotation was not the same between nights, it will be impossible to remove the banding as it will appear at odd angles. in this case one strategy is to simply integrate each night's work and run CBR on the sub-integrations, then integrate together each night's work.

another strategy is to use an ImageContainer and a process icon for the CBR script to apply the CBR to each sub exposure. this can be dangerous though as the CBR script can put large-scale banding into the image to the left and right of bright features like M42. there's a "protect from highlights" slider that can help mitigate this. another issue is that there are circumstances where the script parameters are not properly saved into the process icon, so you may need to edit the code of the process icon by hand to make sure that the settings you chose for the strength and highlight protection are truly in the process icon for the CBR script.

if you ended up with the large-scale banding in the integrated frame due to not enough highlight protection, you can actually use the CBR script to remove it again. however you have the same issue with multi-night in that case; even though you have removed the banding from each sub, you'd still have to integrate night by night to get rid of the artifacts caused by CBR.

rob

[ClausDuerr]

Hi everybody,
sorry for the delay answering :-)

iso,temp and duration has been the same. temp was slightly different.
exposure time was 58secs.
dark scaling factor was 0.2, which is terrible but Vicent and Georg told me that there is some problem with the algorithm (so i should not worry to much about this k-value).
attached you can see the bias and dark as a screenshot.
all has been done in the batch process.
"anti-banding" only had little positive impact on the picture.
drizzling really might be a good idea, but my polarie can´t do it :-)
i didn´t do any flat (didn´t care about vignetting etc).
i did 100 biases - not enough?
it was only 10 darks...

what i also tried was to only have lights and biases (skipping the darks): result looks nearly the same...

the CBR script had some problems next to m42, as you mentioned.
I did send my files to georg and vicent. georg is just living 5km from my place, maybe it´s better to visit the masters :-)

thanks alot for your help.
I let you know if there will be any solution :-)

Claus

[Carlos Milovic]

I think that with 1min exposures you are barelly above the readout noise. I strongly recommmend you to take longer exposure, or leave the background very dark. Or, you just should get a bit "insane", and take more than a hundred lights, and twice as many darks (just skip the bias if you are not using flats, and the temperature matches).

[pfile]

yes of course carlos is right, 58s is way too short and so your signal is buried in the read noise.

58s is good for M42 itself so just crop the image to the bright part of M42 and you will be fine 

or when you get around to making longer exposures, you can use these to fill in the core of M42.

rob

[vicent_peris]

That's exactly the problem, Claus. 58 seconds is an extremely short exposure for narrowband. You should expose *much* longer... Maybe 30-40 min.

The problem that arises now is that maybe the Polarie is not going to handle this exposure. I think that mount is for broadband exposures.


Best regards,
Vicent.

[ClausDuerr]

so next time i will take much longer frames - thanks alot for your help :-)

usually 5-10 minutes with a 100mm lense are ok with the polarie when well polar-aligned.
which is pretty simple on the northern hemisphere.

i leave on saturday for chile - how tricky is polar alignment using a small finderscope with the octan stars?

[Carlos Milovic]

Did you see that we are holding a PI workshop in Chile this saturday?
Polar aligment is always the same: star drift. This is the more accurate method.

[ClausDuerr]

sure star drift would be the best :-)
this would take quite some time because obviously the Polarie is a mobile "pocket" mount, so I have to do it every time again...

Carlos, I would love to be there - but we are arriving on Nov 24th in Santiago but fly on immediatelly to Calama.
on our way back (SCL-Puerto Montt) we could meet for a coffee and a chat :-)
just send me your contact details to clausduerr@yahoo.com

hasta luego
Claus

[georg.viehoever]

Here a summary of my findings:

- Looking at Claus's data, I think that PI Dark Frame Adaption gets it wrong. When I switch it off during BatchPreProcessing, the result looks much better. I dont know why this often does not work properly for DSLRs. I think it is definitely something that the PTeam should look into.
- I think Claus needs more darks, otherwise the darks introduce noise. I usually use at least 30.

In addition to that, some general advice for us DSLR users out there:
- For narrow band, longer exposures (58 seconds are not enough) or -if that is not possible- simply more lights are needed to get faint structures away from the camera noise.
- Finally, you need to cover the view finder and switch of LiveView. These are frequent sources of "light pollution" within the DSLR.

Georg
Georg

[Phil Leigh]

I'm pretty sure it's because of the internal processing that Canon DSLR's do on exposures longer than ~10 seconds... this processing cannot be overridden and means that the linear relationship between light and dark frames required for accurate dark subtraction is broken. If you take darks that match the lights and do not use dark frame optimisation that is the best you can do in my experience.

This was why I went for a -30C cooled "Canon". I no longer use darks at all (just a superbias and flats) and my images have improved massively. The cooling means there is very little thermal noise in the light frames and flats and so darks are not needed.

[papaf]

I found different results while working with a friend's data coming from a 350D (Rebel XT for the americans): darks matched lights quite well, as they were taken the same night. However, they got scaled by nearly 50%. This, though, produced the least amount of noise visually, the image was really sensibly better, even if it still showed the amp glow on the corner.
Not scaling the dark got rid of the amp glow, but the noise was far far worse across the entire image, showing the dread plaid effect.