Are my flats doing the business?

[broadsword]

Hello all,

I have recently started using flats after a few months of struggling with bad vignetting and the like. Overall, I've been very impressed with how they 'flatten' everything (farily logically), but I'm now facing new challenges in processing and I don't really know whether I'm doing the flats wrong or should focus on other possible issues.

To illustrate, first image below is a stacked image of M101 taken with lights (31x120s), darks, bias and flat frames. The flats were done indoors using a white-T-shirt-and-plasma screen-with-a-white-powerpoint-slide method and were 20x1/6s at ISO 200. The image has had some work in DSS (colour balancing and increase in saturation). As I said, a lot better than the non-flat days, but still a bit of vignetting in the corners.

To show the probelm I'm facing, with the 2nd image I have just done one overly exaggerated curves stretch in GIMP. I'm very pleased with the faint details in the spiral arms which have been captured, even with my slow and high FL scope, but also present are some major orange regions plus a blue band near the bottom. I just can't remove these in GIMP processing as they have the same pixel value as the faint spiral arms, and I don't want to lose these. Also I guess the flats seem to have drained a lot of brightness from the galaxy part of the image.

So, I'm wondering whether I need to start messing about with the how I do the flats, or start looking for a good gradient exterminator...

Any advice/goes at processing these very welcome.

.

[Jessun]

I would expect a good flat to a bit deeper into those corners. I often do a few different sets of flats, each set with different exposure. Depending on target one will work better than the others.

The band can't be cured by flats. It's within the CCD, and other weapons must be used. A gradient removal tool is hard to live without for me, and the best one is PixInsight's DBE tool, which allows you to exclude the galaxy from the removal.

/Jesper

[broadsword]

Hi Jesper,

Interested in you comment on experimenting with flats. Is it trial and error, or do you move a certain way exposure wise to get more out of the corners, for example.

I'm steeling myself to part with the 150 euros or so for Pixinsight. Although I might try IRIS first....

[andrewluck]

Nice picture Neil. Once you've sorted the background issues you'll find you can push the processing rather further to reveal a lot morre detail.

The flats are undercorrecting for some reason. Difficult to see why from here but there are a few general tips.

1. You must be shooting in RAW mode. Jpeg isn't going to work.
   
2. Make sure that your light source is evenly illuminated. Difficult to achieve in practice but do the best you can.
   
3. Expose for a histogram that has the bulk of pixels in the centre of the histogram. Auto exposure will generally give a good enough result.
   
4. Take a lot of flats, certainly more than 20.
   
5. Subtract bias frames from the flats
   

I'm away for a week now bu if you upload a flat and lightframe to Dropbox or similar I'll take a look when I get back.

I've been using PixInsight since Christmas and I wouldn't part with it. It has a 45 day trial licence and you will need all of those days to start to tame it. The learning curve is pretty painful when starting out. If you can get over this then the results are very good and, in many cases, less work to achive than using Photoshop.

Cheers

Andrew

[Jessun]

Is it trial and error

Yes, for me at this stage it still is. I use a low setting on my flat panel, and take maybe 2, 4 and 6 second exposures. It's quickly done, and then I do different calibration and stacking runs with these sets and finish with a quick stretch to see which one worked best. Sometimes the flats seem to overdo it leaving inverted looking dust bunnies for instance.

I think internal reflections can play a role in how well flats work at different intensities, but as I will happily admit - I don't know the true inns and outs of it. I know flats are a pixel by pixel division, but how to put that theory to perfect practice I honestly don't know.

/Jesper

[broadsword]

Jesper, Andrew,

Some great points and questions, guys, I would like to respond to them, I'll do them separately if I may.

Nice picture Neil. Once you've sorted the background issues you'll find you can push the processing rather further to reveal a lot morre detail.

The flats are undercorrecting for some reason. Difficult to see why from here but there are a few general tips.

1. You must be shooting in RAW mode. Jpeg isn't going to work.
   
2. Make sure that your light source is evenly illuminated. Difficult to achieve in practice but do the best you can.
   
3. Expose for a histogram that has the bulk of pixels in the centre of the histogram. Auto exposure will generally give a good enough result.
   
4. Take a lot of flats, certainly more than 20.
   
5. Subtract bias frames from the flats
   

I'm away for a week now bu if you upload a flat and lightframe to Dropbox or similar I'll take a look when I get back.

I've been using PixInsight since Christmas and I wouldn't part with it. It has a 45 day trial licence and you will need all of those days to start to tame it. The learning curve is pretty painful when starting out. If you can get over this then the results are very good and, in many cases, less work to achive than using Photoshop.

Cheers

Andrew

- I always shoot in RAW, I just can't upload them as such due to the file size limit, hence the JPEGs.

- I think the light source perfectly even, as it is a TV screen showing a white slide. However, I was wondering how close to put the scope objective lens, as the other lighting in the room might be getting in (maybe I'll just turn out all the lights. Have I just confessed to a major boob?).

- I was trying to get the histogram past the halfway point, as I thought I read to do this somewhere. Great point on using the autoexposure, I'll try it.

- I'll increase the flats in number.

- Ah, now then, this is where my (non-existent) level of knowledge comes in. I haven't a clue what I'm calibrating against what, I just load up the lights, darks, bias and flats into DSS and follow the default settings, and hope it is doing it right. Do I need to delve into this more?

Thanks for the offer re. looking at a flat and lightframe, I'll post one here when I get home. Dropbox is a bit new to me, is it straightforward!?

I also used PI for the 45 day free trial. What with the steep learning curve, lack of images due to the weather, and limited time to play I only did a couple but they worked well. I wanted to look at IRIS before buying PI, wondered how good it is for gradient removal?