Using different exposures in one image

[OK Apricot]

Hi guys, 

I'm new to imaging and a little confused by the use of different exposure times in one image. Take this screenshot for example:

Are these different exposure lengths all put together at the stacking stage? Are they stacked separately and merged in processing? 

Another question - how do we go about flats over multiple sessions on one target? If I had 3x3x3hrs on a target, do I take flats after every 3hrs or do one set of flats at the end and stack them all in the same run? 

Sorry if a silly question, I've tried searching and reading but I'm not sure if I'm finding the answers. 

Edited October 20, 2022 by OK Apricot

[scotty38]

You could do it both ways. I guess the most obvious thing might be a short exposure for stars and a longer exposure for the nebulosity, stack them separately and then combine however needed in processing. Having said that there are a number of ways to skin this cat.....

As for flats it depends, if your optical train, or more importantly i guess the dust etc, has changed/moved NOTICEABLY then you would need new flats.

I can rotate my camera and filterwheel but don't need new flats. The dust, in my case, would seem to be on something that's not moving, I'm assuming my flattener so doesn't cause me any issues.

[ONIKKINEN]

You can combine any combination of exposures, provided that they are not completely out of the same league (like 10s and 600s) when stacking. Choose a stacking software that does normalization of the input subs, and weighting of the subs based on their measured noise levels. This way you will make the most of different length subs and be greeted with the best image in the end.

If you do end up taking very different subs, like 10s subs and 600s subs, normalizing them will be difficult and i would say not worth the effort. In this case you would merge them in processing later. Could be used to create a nonsaturated star layer for example, if your 600s subs had saturated all the stars.

You will need to calibrate each different dataset with their own calibration frames though, including (especially) flats. Ideally you will take flats at the end of every night and use those flats to calibrate that nights data. If you have a permanent setup, or if your camera remains attached to the scope after use, you may find that reusing the same flats works ok. If you are imaging with an imperfect scope like a newtonian or one of the more affordable doublet refractors with suspect focusers, you will need to take flats every time. At least you will want to.

[OK Apricot]

8 hours ago, scotty38 said:

You could do it both ways. I guess the most obvious thing might be a short exposure for stars and a longer exposure for the nebulosity, stack them separately and then combine however needed in processing. Having said that there are a number of ways to skin this cat.....

As for flats it depends, if your optical train, or more importantly i guess the dust etc, has changed/moved NOTICEABLY then you would need new flats.

I can rotate my camera and filterwheel but don't need new flats. The dust, in my case, would seem to be on something that's not moving, I'm assuming my flattener so doesn't cause me any issues.

This would lead me on to my next question. I don't know how you go about combining different images. I've tried looking fit tutorials but I'm not really any the wiser. I use Photoshop CC. Say I can calibrate each stack correctly (deep exposure for nebulousity, shorter for stars) and produce two images to process in photoshop - how do I actually blend these together? I downloaded the free trial of Starnet 2 and am trying to get to grips just with that - once the stars are separated I have no idea what to do about processing the stars separately. I've tried doing a select and mask, but can't seem to copy that mask to a new layer. Even then, I don't know how to merge those layers. 

Thank you for your advice on the flats and stacking though. I'll see how I get on. 

[scotty38]

Sorry can't really help with PS, probably need someone like @ollypenrice to give you some tips there

[ollypenrice]

10 hours ago, OK Apricot said:

This would lead me on to my next question. I don't know how you go about combining different images. I've tried looking fit tutorials but I'm not really any the wiser. I use Photoshop CC. Say I can calibrate each stack correctly (deep exposure for nebulousity, shorter for stars) and produce two images to process in photoshop - how do I actually blend these together? I downloaded the free trial of Starnet 2 and am trying to get to grips just with that - once the stars are separated I have no idea what to do about processing the stars separately. I've tried doing a select and mask, but can't seem to copy that mask to a new layer. Even then, I don't know how to merge those layers. 

Thank you for your advice on the flats and stacking though. I'll see how I get on. 

When you say, 'combine different images in Photoshop' I guess you mean different finished images, possibly of different exposure times?

But question 1 is, Why complicate the issue by taking different exposure lengths in the first place? I have combined different exposure lengths on M42 and... that's about it. I did not find it necessary for the Andromeda galaxy, for instance. Manual aligning in Ps is slow and labour intensive. You might try its automated image alignment, though. I find it works well for daytime photography when I want to combine a set of  focus-bracketed macro shots; I've never tried it on astro images. (The path is File-Scripts-Load files into stack.) This will give you a set of layers with each image aligned - we hope. Now, what do you want to do with these layers? You could try Photoshop's automated layer blending. Make all layers active and then follow Edit-Auto-blend layers. Choose stack. I've not tried this for high dynamic range, only for focus bracketing.

If you want to be clever, as is necessary for M42, you can take very different exposures. I used 10 seconds, 50 seconds and 15 minutes. The tutorial I used to point out to people has been deleted but this one is similar. https://www.astropix.com/html/processing/comp2.html   What I do is not quite the same. I paste the short exposure on top of the long, create a layer mask and copy the long exposure onto it. I then greatly increase the contrast on the layer mask  and give it a huge Gaussian blur. Maybe 4 or so. This allows the short exposure to pass through the mask only where the mask is burned to white. I think it is the most natural way to do the blend because the long exposure itself largely creates the mask. As you work on the mask you can see its consequences on the image as it will end up. The beauty of Photoshop!

While I have not used multiple exposures on normal targets in the past, I'm thinking of doing so since I've started using StarXterminator in Ps. This lets you remove the stars, process what's left very hard, then replace the stars. I'm wondering if it might be good to shoot a set of short exposures for the stars and use these on the starless image at the end, but I haven't tried it yet so it might or might not be worth it.

Olly

 

Edited October 21, 2022 by ollypenrice

[Elp]

Regarding combining results from different sessions, I stack each session separately with their respective calibration files, then stack the stacks 100 percent with no rejection. DSS recognises the exposure time of each stack, I would have thought this is correct.

For star separation in PS or (GIMP), goto select > colour range, select "highlights" in the dialogue which pops up this will select the stars (also note it will select other bright objects of similar white intensity, if it does once the selection is made click on the lasso tool, hold alt and drag around the objects you don't want selected, you don't have to complete a whole circle around the object just near enough a circle and release and the software will auto complete the circle (in GIMP you have to manually complete/close the circle selection)), you'll see the stars are now selected. Now you want to go to select > modify > expand to make the star selections slightly larger to accommodate for star glow, I input something like 4-8 pixels, then you want to soften the edges of the selection so they blend better with the new layer so go to select > feather and input a pixel value half that of the one you expanded with. Once done, you just right click and "layer via cut/copy", and now you have a new layer with just the stars, which you can select all, drag onto a new document or however you want to duplicate it. GIMP methods for the above may be slightly different but it's very similar.

Personally I find manual star alignment easy in PS as I do it all the time having shot with a mono camera, use one layer as a reference on the bottom of the layer stack, the layer you want to align on top with around 50 percent opacity and move it manually, note similar stars which are out of alignment and do a edit > transform > rotate (it'll enable you to do scale/rotate/resize/skew at the same time anyway), at the top under the menu bar you'll see fields where you can type numbers for more precision, edit in tiny steps until they're aligned, press enter to confirm. Repeat. Once you're close select move tool, press the arrow keys on the keyboard to move the layer by 1 pixel increments until you're aligned. To confirm set the opacity of the top layer back to 100 percent and keep toggling the top layer on and off and you'll see the minute misalignment, again use arrow keys to move until you're there. I find maybe I'm like 1 pixel out, but when they're layered as an RGB you don't notice. I did this last night with a narrowband image and it took around 5 minutes to align the three layers. It is however extremely difficult to impossible to accurately do this with a star and starless image only, you can use the image with the stars as the reference image to align to then delete it once you're done.

[OK Apricot]

@ollypenrice, @Elp, thank you both for your replies - plenty in there that should keep me busy for a little while. I will be back soon I'm sure!