Should we try something new on pier 14?

Hi Ray,

this article is pure gold. Personally I will give it a try on some older datasets using an RGB extracted L and mixing them back to the RGB image. If this works as Pete mentioned we can save lot of integration time.

Reading further about QE of the cameras, and the unbalanced subs we're giving to the submissions, I believe a crib sheet will be a good starting point. But as an IT guy I would recommend that the submission form should cover this. 
Let me share my thoughts on this.

In the submission form we should be forced to use RGB only, set sub length and total integration time. Then it should calculate optimal sub count (not sub length) for each channel based on the QE. 

The same approach can be adopted for narrow band filters. 

If we need to do retakes on a specific channel, there can be a option to overrule this. This might not be ideal for everyone but i believe it can worth a try and helps the less experienced or lazy members like I am.

Does anyone know how RGB subs are taken? Is it per channel until it reaches the count like RRR,GGG,BBB meaning sequentially ,or is it taken RGB,RGB,RGB and so on? If the first we may consider to switch to the second, this can saves us retakes on a specific filter when high clouds appear. 

regards

Michael

Our system on p14 is automated and images in 30min blocks due to the specific cool down characteristics of fast telescopes like this, so depending upon the moons size and position it may image R G or B or even narrowband, it may also change filter as the moon moves further over to the west, or if the moon sets part way through the night.

eg If the Moon sets for example it will at the end of a  30 minute block our system may have been imaging HA but it would choose another filter that is more susceptible to it's influence to take opportunity of the moon being below the horizon

I have zero problems doing a crib sheet as i have shared many but as its the members pier to input and use we will leave YOU to input the job

HTH

Steve

Hello Michael, 

Thanks for the suggestions, hopefully there will be a few more responses.

To answer something you asked about, it will  only ever image per channel and why we can end up with unequal numbers of usable subs due to clouds. 

Another suggestion along the lines of something you mentioned on the other post. Split the total integration time per job on those rare targets which require a huge amount of imaging time.

This should only mean that it is split over 2 narrowband submissions plus rgb if required. A slight overhead for Peter who enters our job submissions to the queue admittedly. It wouldn't necessarily need to be split 50:50. So for example, 20 hours of narrowband could be submitted as 12 hours initially. Upon eventual receipt of the data and inspection thereof, then and only then should the remaining 8 hours plus any retakes be requested. If at that time the target will be setting soon, you could just request the retakes and optionally an additional hour or two. This strategy would go some way to help avoid receiving datasets with incomplete numbers of subs for filters, should imaging time on the target run out. This is just an insurance policy to cover long spells of bad weather. 

What is a long job and ought to be split? It's not possible to quote an exact figure and the 12 hours mentioned above as an initial submission is open for debate. If the total time for example was 13-14 hours and there's plenty of time available to capture the target then you might not split it. Anything much above that might be inadvisable. 

Sorry for the lengthy explanation. 

Cheers and CS,

Ray 

Lot of thanks for explanations to both of you. I've learned a lot and will considerate this for my next submissions. In the meanwhile i started to program a little which helps me doing that.

Now lets wait for other members share there thoughts.

Warning: Contains controversial content.

Attempting to get support for this suggestion is going to be near impossible. Not only would this require support from pier 14 syndicate members but Steve as well.  The suggestion will no doubt make me less popular than someone who passed wind in a lift/elevator full of people. Mission Impossible? So I'm having to resort to a bit of frivolity and psychology.

A discussion over the phone yesterday with Steve on this topic went the the usual way.

I'm informed that what most of us are doing is causing many of the problems,. That was followed by a long spiel on what we need to be doing. Admittedly much of it does make very good sense, but then the usual boast came. "I could get twice the amount of imaging done" he said. Remember that.

If you've read the RGB vs LRGB post you'll have seen that Steve mentioned he would be posting a crib sheet to assist us when planning a job. This has been done before and yet here we still are?

We all signed up on the basis that this is our pier and we can, within reason, put in what we like. Steve's point is that we don't always get that right. You have every right to do so and put in jobs exactly how you want them. Simply vote against this.

Suggestion: (pier 14 only)

After the test jobs have been completed the job queue will be inspected by The Man himself and those deemed unsuitable will either be adjusted or removed. Its unlikely that any job would be removed, with the possible exception of his pet hate, mosaics with more than 2 panes. More on that later. You would need to look at the job queue to see if the hours match what you submitted to see if he reduced the integration time. 

If in future you think you might be submitting a job that on the face of it requires an excessive amount of imaging time, then as many people already do, put an explanation in the notes at the bottom of the submission form.

What we need to do to keep Steve happy is by submitting jobs ahead of the time they rise.

One other thing for now is that despite Steve's best efforts there will arise situations where after your job gets to say an arbitrary 70% completion, you see no further progress after a  few weeks. I'd suggest sending in a help desk request to have what's been collected downloaded. On RGB jobs the moon phase needs to taken into account before doing so as this might be part of the cause. Inspect the data asap and as a minimum submit retakes for bad  and missing subs. You would need to request just enough subs by filter to approximately make the numbers equal to allow the creation of an image. If you put in the notes that these are retakes, and these should only be just enough as previously mentioned, then I hope that when good old Steve sees that these are retakes, he will find it in his heart to override things and get these done.

Remember guys, he reckons he can double the amount of imaging time we get compared to now. So if by some miracle the members accept this but Steve doesn't, then I think we will all know that he can talk the talk, but not walk the walk!

Tip:

if you desperately want Steve to approve an exceptionally long job or a larger mosaic, you might want to consider sending a bottle of fine whisky to 802, Kingsbury Road, for the attention of Mr S. Asbury, with you name clearly marked as the sender. That ought to help oil the wheels and you may get a surprise when you later look at the job queue. Works for me! :)

Now I should set up a poll, but not yet having done so I'm in fear that if I try I'll lose this entire post!!! I'll have a go later on a separate one.

Cheers and CS, 

Mr. Unpopular :(

[Note to Michael regarding the post just ahead of this one. Whilst I was typing this rival to War and Peace, your post appeared. I wasn't risking looking at it in case I lost what I'd written so far. I'll have a look in a short while. Again thanks for your keen interest. Regards, Ray ]

Let me get this straight Ray, you want me to do even more work with the added potential to also fall on my own sword ? :) :)

Me and my the big mouth strikes again!

Steve

Steve, 

LOL. 

I already tend to submit RGB and NB of the same target separately (I don't do it often) as that seems to me to be an obvious hack to get useable data more often. I have also in the past separated L from RGB - again maximising the chance of getting data to play with.

Being able to check how a job is going (data-wise) would be excellent, although I suspect a massive disruption to the work flow.

With all due respect to Michael, I am sceptical about too much automation and forced choices. It is hard to get right and I have suffered many times from IT systems that delivered what the IT guys planned, but were unusable by the real problem owners. Great care is needed.

One piece of advice I was given that has worked well is to find targets that are rising in the east (NE - E - SE) late in the night. We all have a tendency to target what is overhead, which is not the best way to maximise imaging time.

But let us continue to help all of us make better choices of job (cf LRGB/RGB debate) and maximum pier time. I want to learn more about how to chose and implement imaging jobs.

I have been out of it for a while (partner's cancer diagnosis and treatment) and have a lot of existing data to process, but I hope to be back in the saddle soon.