Problem with Flats

Ok Plan of action

Re-Do Flats up to 80 seconds with the following exposure times (Along with corresponding darks to subtract from the flats):

2,4,8,16,20,24,28,32,36,40,44,48,52,56,60,62,64,66,68,70,72,,74,75,76,77,78,79,80

My existing Darks for the DTC are still OK right?

If I can get the "Overscan" data, I will redo both flats and darks, but if that is not available just re-do the flats? If I re-do my darks, should I do them at +15 rather than cooling them and do the same thing with BIAS?

Simon

Note step 3 - I qualified that when pasting in a copy of I and J we paste was values only - not formulae - that way the computed values are just copied as numbers and the dependency on other cells are broken.
It just makes for a simpler new spreadsheet for working purposes.
It's sometimes useful (for me anyway) to avoid clutter that I'm not immediately interested in.

Mark C.

Hi Mark

But surely since we're copying values only, those values are skewed because of the "Offset" values that are in the old spreadsheet and because I do not have any "Overscan" data I do not have any offset data, would I be correct or am I barking up the wrong tree?

Simon

In no general order of importance:

Mark - the data at 82 seconds for shot B certainly does appear suspect.

Just out of interest, are you still plotting on log-log scales?

Overscan - my understanding (limited) is that since this region is not affected by photons (it is masked in some way) it gives raw data regarding only noise generated by the electronic circuitry (including heat I suppose). So it is "pure" data. A bias frame will include any random noise associated with any given pixel, presumably including any charge it has retained from a previous exposure plus anything generated by a photon. Of course, if your shutter if light-tight and you know not one photon has gotten through, the Bias frame should be the same (or at least close) to the pure electronic noise. Should being the operative word. The overscan region gives you a definite. Otherwise I'm guessing that manufacturer's wouldn't bother going to the trouble of putting it there.

Of course, even it is there (as is the case with Simon's chip), the camera manufacturer can simply ignore it and give the user no ability to access it through a driver or image software. The better manufacturer's of course *do* provide the means to access it. Richard Crisp, whilst he has a 'certain way' of putting things across (that upsets a lot of people) has always been very good at calling out these issues and giving people at least real data on which to form an opinion about a particular product. I digress - like you, I don't know for certain but I'm sure what I've said above is based on something I read some time back rather being randomly generated by my brain

Simon - assuming you *can't* get access to the overscan region, stick with the darks (for the DTC) as they are. No need to reshoot.

As regards the lights (flats) perhaps the better strategy is this:

1. Put on your original number of sheets of paper and add maybe 10 to 20% more. This should extend the exposure time - perhaps to 100 or so seconds - that will lengthen the curve and reduce the need for at least some images.
2. Start at 80 seconds and perhaps go up in 5's - to see where you hit the full-well point (plot the data on the spreadsheet and you should see that clearly even just looking at the numbers).
3. Say that turned out (with the extra paper) to be around 100 seconds. I'd then go for

2, 4, 6, 8, 10 (to get the lower end)
15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85 (mid range)
88, 90, 92, 94, 96, 97, 98, 99, 100, 101, 102 (top end)

I'd adjust that list depending on where I'm seeing the full-well - it might actually only be 90 seconds - adjust accordingly. If it was more than 100 seconds, I'd take off a sheet or two of the paper.

I'm really still not sure about the bias frames. Perhaps use the same ROI as the darks/flats and use the data in the "offset" column - i.e. you substituting the overscan region data ('cos you can't get it) with the next best thing - which is the bias data. Or another way, the bias data *is* your offset data. But as I say, I'm not certain about this... Mark?

Obviously if you *can* get the overscan region, you're on the pig's back in terms of getting better quality data. If that is the case, reshoot everything. Shoot the darks at about 15 deg C. Shoot the lights at your usual imaging temperature and shoot the light-darks (cal frames) at that temperature and duration. And you don't need to shoot bias frames at all.

Hope I got that all right and you can understand it...

I think we all deserve a pint...

Dave

I have the info from my Master BIAS (250 BIAS Frames Stacked) for the region 2249x1351 at 140x140 which is the same region as my Lights and Darks, and this gives me the following:

Mean: 353.204
STDEV: 1.646

So how do I translate that now to my offset? Obviously if that is the correct thing to do

Simon

I made (at least one) error I though I'd taken data from a separate bias col - I took I took it from from overscan - which is the bias - and if you take bias frames you can populate that col.
I'm not sure why the value in col N is being subtracted - or even what it is.

Dave is right about overscan - but I think it may be more than shielded - I think that they may be minimal non-functioning pixels which allow for bias alone to be captured - ie not even dark current - need to read up myself on that.

Either way overscan is the preferred way of getting bias - but it can be done with zero-sec (or nearest) dark exposure.

In response to Simon's last post - bias has to be measured.

While overscan is the text-book method - it is not necessary.

Simon, you can calculate the gain from mean value of a pixel region and its std dev - but not the bias.

Dave I was tryng to plot log-log - but it didn't come out well.
Dave - youre right where the spread sheet has overscan and offset - just do bias frames of same region as other data.

I will revisit the spread sheet and make more sense of it and post back - but not tonight...
Simon try to do the new sheet and copy as directed - despite it maybe not having your overscan - and one subtraction that I'm not comfortable with - it will give you an idea of some of what we have been talking about.

Mark C.