Flats - some clarification please

[Psychobilly]

Nope...

[roundycat]

I fiddle with my dew shield as well due to height constraints in the obs when the scopes are pointing straight up. It makes no difference.

Dennis

[ollypenrice]

Agree with Peter and Dennis.

There's something odd about those flats.

Olly

[martin_h]

I found that placing an EL panel directly over the end of my scope was way to bright, 4 sheets of clean printer paper toned it down to a level where I had sensible exposures.

[IvanT]

I found that placing an EL panel directly over the end of my scope was way to bright, 4 sheets of clean printer paper toned it down to a level where I had sensible exposures.

I've placed a white T-shirt over the panel which is sandwiched between two pieces of plywood (8" diameter circle cut out of one face) and it seems to produce better flats though I still don't see any dust mites etc. in the flat.

[martin_h]

I've just had a look at your posted flat in Photoshop, and alltho the histogram is central for red and green the blue is off the scale!

I also loaded it into Nebulosity2 and looked at the pixel sats, and the mean is around 38k but the max reading is pegged at 65535

I dont know if this is causing your problems, but to me it seems as if the blue channel is over exposing and washing out the detail.

Perhaps someone with a bit more knowledge could confirm this?

[martin_h]

I've just noticed that you set the WB to tungsten...this could be causing the blue cast, altho colour isnt usualy an issue in a flat perhaps in this case the maxed out blue channel is messing with the average????

Dunno - just a thought?

[IvanT]

Thanks Martin. I wasn't sure what WB to select for the flats. For the subs I used a custom WB & that flat you've looked at may be based on that. I've tried so many different combinations I'm loosing track.

My latest flats seem a bit better though there is some banding visible. I used my light panel with the T-shirt, Auto WB, ISO 800, exposure duration 1/13 (I think it was), just long enough to place the histogram 2/3 up the scale. I've been told that exposure duration is typically between 1-3 seconds, if I go over 1 sec it's off the scale.

Here it is, the circular banding is quite apparent

The flat you've analysed looked blue to me too so I tried sheets of white paper in front (looked grainy) then the T-shirt. It didn't occur to me to looked at the histogram of the flat in PS to try and see what's going on. Thanks for the tip.

[roundycat]

I am not a DSLR user so take my comments on their use with a pinch of salt.

In the first place you should set your white balance to whatever 'normal' is. In terms of Kelvin, 6500 should do it. Certainly not tungsten (it will be trying to set an amber-ish colour to white so it will appear blue) and in any case Photoshop can take care of most imbalance in the colour.

As regards exposure, there is a lot to be said for setting the same background level as for the light frames. But only in the math. The idea is daft but I take my flats with an average adu of about 9,000. In Histogram speak that would be about one-seventh of the way up from the bottom in 16bit. If your camera is 12 or 14 bit the disparity will not be as large. It sounds as if a lot of your flat is saturated.

Dennis

[Psychobilly]

I normally shoot my DSLR flats in Manual mode and under expose them by -2/3EV ......

I use an EL panel with a couple of layer of opalescent perspex in front of it one is immediately in front of the panel and the other is 2 inches in front of it...

The flats are raws so white balance adjustments arent applied... unless DSS is soing it...

The individual flats have a colour cast (from the LP filtersa) but the master flat thats produced is a grey scale image...

Peter...

[martin_h]

Your flat still looks to even and clean.

This is my master flat loaded in to Nebulosoty2 - the capture program. My EL panel was about 18" away from the scope, it has 2 layers of opal perspex over it and 3 layers of printer paper, the exposure was 3.5 seconds.

You can see the pixel sats in the box...the mean value for the whole image 22809 and the histogram is 1/3 in from the edge....and you can se the crud in the light path...

[IvanT]

the exposure was 3.5 seconds

Hmmm, I think my 1/13 s exposure might be a tad fast then to catch the imperfections. I'll play tonight with some plain printer paper in front to see if I can get the exposure duration up. I was also only stacking 10 flats, I'll up that to 30 or so.

Thanks

[Psychobilly]

Dont get too hung up on the exposure time... most of my flats with the Megrez72 and the reducer are around 1/80s .. depending on the camera you may also need to make sure the exposure is long enough to allow for any shutter related variations...

The EL panels are driven by a "switching" waveform which caused variation across it.. if you take a very short exposure you will see banding...

If your optics and sensors are clean then you wont see the "bunnies"... any variation show up best when stretch 16 bit image to the extremes...

What I did to find the optimum flat exposure for my setup was take a series of flats covering a wide range of EV's (initially +/- 3EV) and they applying the master flats from each of these to a set of lights and compared the results... then repeating with finer steps around the EV Values that gave the best results...

Peter...

[martin_h]

Yes - my exposure was that way because I have a shutter on the CCD so I need 1+ secs, all I do is adjust speed to get an average pixel reading of 22-25K ADU.

But the bottom line is what are your flats doing to your image? if the combined calibrated image is looking ok and reasonably flat, the flats must be working. Do a stack with and without flats and compare.

[IvanT]

Can't believe I'm happy to say this but... I'VE GOT DUST BUNNIES.

There are a least 3 in here that I can see.

Two things I think were causing me problems other than the learning curve:

1) My camera had a custom W/B Blue bias offset configured I didn't realise was still set from way back when I first got the camera and played with the settings. I know WB doesn't affect the RAW but would a bias offset impact it?

2) I was trying to set values based on the histogram on the camera rather that looking at the pixel stats in Nebulosity. The master flat above is composed of 30 1/8 ISO 800 frames stacked in Nebulosity, the camera histogram is off the scale in exposure and RGB, related to my recent removal of the IR filter maybe?

I played with the settings in manual mode (previously had been relying on Av mode) until I got an image mean value of around 25 ADU. I also diffused the light panel with a sheet of high grade paper (almost like cardboard) and in front of that a white T-shirt.

Going to have a play now to see how it applies to the subs but thought I'd post this in case someone else has similar difficulties in the future.

Please please please, tell me it looks like a flat

[roundycat]

just for the sake of mentioning it, short exposures are not too fast to catch the imperfections. They all travel at the speed of light!

Dennis

[IvanT]

Now why couldn't you have told me that at the start

Does make me wonder though why lots of imagers shoot flats of 1-3 mins

[martin_h]

That looks more like it

[RobH]

Does make me wonder though why lots of imagers shoot flats of 1-3 mins

I've never heard of flats being done with that length of exposure

Where did you get that info?......it's completely wrong!!

Rob

[IvanT]

Ooops, I said mins, meant seconds. I guess it all depends on the light source being used.

Sent from my iPhone using Tapatalk

[RobH]

Ooops, I said mins, meant seconds. I guess it all depends on the light source being used.

Phew!!

That's more like it, although as you say it depends on many factors....light source, filter, type of camera etc.

Rob

[Psychobilly]

Thats more like it....

Congrats