Problem with my flats...! HELP - Olly, Rob H....., anyone!

[TakMan]

Hi everyone, I could do with some help!

To my shame, I did not take my flats at the same time as when I shot M45 - see Deep Sky thread currently...

But, all was not lost as I purchased one of those Electro-luminescent Panels from Earlsmann as Olly had shown in a thread pre-Christmas!

I carefully broke down my set-up, taking care not to move the camera, and a few days later, over the Christmas break did the flats in the garage.

30 of them -20c as I had for the light frames (with the Hutech IDAS LPF in place in the True Tech wheel - did check it said the correct position!). Captured with the focal reducer fitted - f3.6.

I have a piece of perspex masking-taped to the panel as suggested (partly to protect the connection as well as aid light diffusion).

Now the first question! to get a 20,000 average (ish) I was only shooting at 65 milli seconds.... I'm starting to wonder if this was too quick and perhaps I should have added 3-4 sheets of A4 paper to slow it down to a second or so.... I don't know why I say that - perhaps to even out the light....??? What do the rest of you shoot at...?

I've added 2 files of a single flat: both the same file, but one to show that the dust bunnies are in the same place as those on the stretched lights that follow.... you can just make the top one out!

I've really struggled with processing M45. Don't mind trying to sort the nebulosity from the main stars - that's all part of the fun of it, but had this pronounced gradient from the top has been an unexpected nightmare....

Even though I've got the IDAS V2 fitted, I believed that with 5 minute exposures (at f3.6), that perhaps this gradient was light pollution getting through?..... we suffer from distant moderate pollution if that helps....? M45 was fairly high up though, and I've seen other peoples shots of this using 8 - 20 minute subs....

Followed the instructions in Nebulosity 2 (Mac), not aligned, just stacked....

3 examples follow.

The first was a basic adaptive stack (flat frames applied) in Nebulosity 2. cropped, then a DDP stretch to show the problem.

Second is a drizzle stack of the same files (flat frame applied) and the same DDP stretch to show the problem - but 1.5 times bigger this time!

Third is a drizzle stack of the same files BUT WITHOUT ANY FLAT FRAMES, and the same DDP stretch. I think you'll agree that this does not suffer anything like the previous 2. I presume that the darkening around the bottom and the right corner show the reason for requiring a flat frame...? But by not applying that master flat, the gradient has mostly gone - HELP, HELP, HELP - where am I going wrong!

Sorry to burden you with all that!

Any help would be greatly appreciated...

Damian

[brianb]

to get a 20,000 average (ish) I was only shooting at 65 milli seconds.... I'm starting to wonder if this was too quick and perhaps I should have added 3-4 sheets of A4 paper to slow it down to a second or so.... I don't know why I say that - perhaps to even out the light....???

Your example "flats" have a marked gradient across them, this could be caused by the short exposure as the wells will continue to fill whilst the CCD is reading out unless you have a mechanical shutter. If you can't reduce the brightness of the light source, try a filter of some sort ... a few sheets of paper will reduce the light level to a hopefully manageable level.

[TakMan]

Thanks for the help brianb.... The SBIG ST4000XCM one shot colour camera does have a mechanical shutter I believe.... does this help..?

Ohh, and the light panel was just resting on the top of the scope....

I've got another question to.... my Kodak chip 4021 I think is 'only' (sorry Olly!) 15.2 x 15.2 mm wide - surely well within the TAK FSQ106ED's 44mm imaging circle (reducer spec) - would I see any vignetting....

Damian

[steppenwolf]

Now the first question! to get a 20,000 average (ish) I was only shooting at 65 milli seconds...

My most recent set of FLATS taken with my EL panel were of 175 milliseconds duration - longer than yours but a long way from seconds so I don't think this is the issue.

Am I right in assuming that this image was taken with a one shot colour camera? If so, did you apply a boxcar filter to the FLATS to destroy the Bayer Matrix?

[brianb]

The SBIG ST4000XCM one shot colour camera does have a mechanical shutter I believe.... does this help..?

Not if it's not being closed at the end of the exposure. I don't have this camera & therefore don't know how to control it but I'd be surprised if the mechanical shutter is normally used as it will cause vibration. Naturally this is of no consequence when making flats.

[steppenwolf]

Your reply to BrianB has answered part one of my question - so what about the boxcar filter to the flats?

[TakMan]

Steppenwolf - not sure about a 'boxcar filter'... sorry.

Use MicroProjects Equinox Image software (Apple Mac) and just choose 'Lights' and name them myself so I know they are the flats. Then process in Nebulosity..

Does this help...?

damian

[TakMan]

Just thought of something else...

When I take lights, I convert them to colour (after normalising), before I align/combine/stack - so the demosaic happens there...

I don't do this with the 'light' flats though, just combine and stack before selecting them in the dialogue box (with darks, bias frames etc) to process the real lights....

Does that make sense...?

Damian

[steppenwolf]

Unfortunately I am not acquainted with your processing software. A Boxcar filter applied here destroys the Bayer Matrix which means that no colour can be derived from the FLAT data thus avoiding colour casting from the flat. I FLAT frame my OSC data before I de-Bayer it - can you also do that? If so, try blurring the non de-Bayered FLAT frame to destroy its matrix before applying it to non de-Bayered LIGHT data and then de-Bayer the calibrated light frames.

This should stop any colour cast imparted by the FLAT.

Secondly, how accurately did you hold the EL panel on the front of the telescope? Could the EL panel have 'bowed' over the front of the telescope? I held mine flat by pressing down on it with a piece of heavy cardboard to ensure that this couldn't happen. If the panel had bowed, it would no longer give a 'flat' light output.

[steppenwolf]

Sorry, we are doubling a bit!

[TakMan]

The panel should be flat. It's only A4 and masked all around to a 7mm piece of perspex. White hardboard securely masked to the back with the transformer stuck on the top!

[steppenwolf]

I agree, your panel is flat then!!

[TakMan]

Sorry for doubling....

So I think so far I'm doing the correct processing (?) my 'light flats' are still RAW 'colour' un-demosaic files when Nebulosity asks me to process them with the RAW 'colour' un-demosaic 'proper' Light files (along with any bias and dark masters)....

then i 'normalise', then convert to colour, then align and stack.....

perhaps it is the duration of my flat light frames then...?

Damian

[narrowbandpaul]

you have the 4021 sensor....this is interline transfer technology, and does not require the mechanical shutter. It will have an electronic shutter. The FLI version of the same sensor allows you to hold the shutter open while exposures are controlled by the electronic shutter. This permits very short flats if need be.

it also will remove any issue of shutter shading.

try using this mode for the flats...or use sufficiently long flats so that the integration period is much longer than the shutter opening time

[TakMan]

Thanks for more info narrowbandpaul!

I don't recall having the option in the software - it just does it's thing, sorry...

Just took a look at the SBIG site:

Mechanical shutter for automatic dark frames

Electronic shutter for exposure times to 0.001 second

any help...?

Thanks everyone! WOW

damian

[steppenwolf]

So I think so far I'm doing the correct processing (?) my 'light flats' are still RAW 'colour' un-demosaic files when Nebulosity asks me to process them with the RAW 'colour' un-demosaic 'proper' Light files (along with any bias and dark masters)....

then i 'normalise', then convert to colour, then align and stack.....

I'd still like to see the result using a version of the flat that has had its matrix destroyed.

Paul makes a good point regarding shutter shading, this could lead to uneven lighting of the sensor.

[TakMan]

OK...... so what's my best approach to do this boxcar filter 'steppenwolf'?

Photoshop CS3, as I can't find anything so far in the nebulosity manual!

Damian

[TakMan]

Guys, thanks for all the help so far....

Due to my long commute, I'm off to bed now, so many thanks to everyone that has or may reply later.... I'll do my best to add more tomorrow!

Sweet dreams....

Damian

[ollypenrice]

No, 'boxcar filter' is a new on on me too but I am guessing that it does the same thing as the 'shift' function I apply in AstroArt when making a master flat. I use the Shift (of 0.5 pixel on both axes) to remove the Bayer matrix. This is very important but I'm afraid I only know about it within AstroArt and not in other programmes.

I do sometimes get a gradient when using the light panel. To avoid this I use it in a dark place and I also use an added filter of perspex glued to a polystyrene sheet with a hole to fit snugly over the dewsheild. I suspect stray light can sneak in without this.

An FSQ106 will have no difficulty whatever in evenly illuminating a 15mm square chip. Your flats show about the same fall-off as mine at native focal length. My own FSQ is the smaller one but Tom O'Donoghue images here a lot using our Atik 4000 in his own 106. It is still worth taking flats but the fall off in illumination is minor. When using the reducer in our 85, however, the illumination is much less even and the flats more important. Tom's earlier FSQ doesn't accept the reducer so I have no experience on that front.

I think the issue is this debayering and maybe stray light.

Oh, just a thought but your flats are being applied the right way up? My camera software gives you the option to invert and this foxed us once!

Olly

[RobH]

Hello Damian,

The boxcar filter is in Maxim DL's calibration section.

You can aways download the free 30 day trial and have a bash in that.

Are you bias subtracting your flats?

If you don't, they may well overcorrect as this throws out the calculation.

Remember also that the top left of your image contains nebulousity, some of which is a brownish colour, so don't try to get rid of this

I use an EL panel these days, and have 3 or 4 sheets of paper in front of it. My exposures are still very short though, quite a bit less than a second.

Cheers

Rob

[ollypenrice]

Good point about the bias frames Rob. I forgot about that.

Damian, if its any use I can send you a colour camera workflow I wrote for guests at Les Granges, but it uses AstroArt. Go direct to lesDOTgranges@hotmail.co.uk and I can attach it if you do.

Olly

[TakMan]

Thanks again to you all!

OK...

The flats were taken in the garage with the lights off and the laptop screen brightness lowered. Perhaps I'll add a shroud next time just to be doubly sure!

I do actually have some more flats taken at 1.3 seconds (I added some sheets of paper as Rob suggested - again, I don't really know why, perhaps I just thought that 65 milli - seconds appeared a little too short...), for my first go at the Horse Head - the camera has moved though by a degree or two, but I think it may be worth trying to see if these have the same effect.

Olly, a good point about making sure they were applied the right way up(!) I believe that there is this option, but I kept it unchecked/checked for each, so don't think that is the issue - as wouldn't the dust spots still be there if one frame had been inverted.....? Thanks for the link by the way... will take a look!

So, to start with I think I'll stack the light flats in Equinox Image (camera acquisition software) to see if that handles the files (rather than nebulosity), I'm just going to have a look at the pixel values across the 65ms and 1.3s flats to see if I can detect this shutter shadow..... I'll post them later for you to browse/advise on.....

Maxim is a slight problem being a Mac user.... I'll see if I can borrow a spare laptop from IT here and download the free trial to see if that helps... means learning PC/Microsoft though..... ;-) and that sounds more scary than processing...!

I did bias subtract the flats with a newly created master (done at the same time) just so you know. I did not take any darks as I felt that 65ms was short enough(???), I did take some though for the 1.3s one, just before you all frown!

Speak later.....

Damian

[RobH]

Hello Damian.

I never bother with flat darks as on really short exposures bias frames fulfill the same role.

Macs are great, and I won't get into the childish Mac vs PC debate, but you will find that a lot of astro software isn't written for Mac, so it might be worth you getting a secondhand PC laptop for astro stuff...there are plenty of secondhand XP ones around for next to nothing these days.

Cheers

Rob

[TakMan]

Thanks Rob - you're right about most of the software......

Attached are two shots showing rough measurements of pixel values for the two flats. First is the 65ms one (and it does show a significant variance in values). The second is the 1.3sec flat. Although there is a difference across the frame, it is not as pronounced as the first.... is this significant(?) and possibly validates brianb's and narrowbandpaul's concerns......

I tried to get as close as possible for each one by using the edges of yellow sticky notes attached to the screen and took for the average around each....! Just trying to help!

It may be not this bayer matrix issue possibly then......

Any more thoughts before I head off to the IT dept!!!

Regards,

Damian

[TakMan]

If you all feel this is not the route of the problem - I have just got a laptop PC for the weekend.

I'll download Maxim...

Any chance someone could forward a basic tutorial of boxcar filtering and stacking the flats just so I can see...please!

Thanks in advance,

Damian