Photoshop (and other image processing) Tips.

[gregbradley]

I posted this on another site but it became quite detailed so I thought it may be helpful to others.

It assumes you use CCDstack and Photoshop and use CCDsoft or Maxim DL for controlling the camera and taking the actual images.

It is based on using processing techniques written up in Ron Wodaski's book The Zone System and it is also based on DVD tutorials by Adam Block (one of the best) and Tony Hallas (also very good).

If you find it confusing or want something clarified feel free to PM me for clarification:

Image acquisition. Save to a file named after object. Make sure the filter wheel is labelled correctly so red is red and not blue etc.
Image LRGB with luminance say 2 hours of 10 minutes subs 1x1 and RGB 10 minutes 2x2 and 30minutes each minimum.

CCDstack
open all the luminance
dark subtraction and flat fielding
normalise using a bright and dim area in one selection box in the image.
data reject hot/cold pixel removal
register (I use the CCDis alignment plugin - its worth the $99)
combine - median.
save in file above with a new file called masters. Call it master luminance (number of subs)
I then run a 40 iteration deconvolution on luminance usually unless its a nebula.
save that.

Open all reds.
Do the same as above but usually I don't do deconvolution on RGB unless one is a bit bloated like blue.
Save each as an Red, green and blue master.

Open master deconvolved luminance and the masters for each of RGB.
Using luminance as base register (if you took Ha you do that now as well so they are all registered).
Save them all replacing the original masters - now all masters are aligned.
Create colour and create an LRGB at this point. Some may only make an RGB colour here. I use LRGB here. Sometimes I need to normalise RGB only if I get weird colours.
I select desaturate background as its often bright red or blue or something weird.
Save colour image as a tiff (all above saved as 32 bit floating tiff files).

Open the LRGB and luminance in Photoshop.
I work on the luminance with levels and curves.
then often a duplicate layer set to soft light and opacity to suit to increase contrast.
Often high pass filtering to increase sharpness and detail.
correct any gradients first up using Gradient Xterminator or gradient methods as in Adams tutortials.
Save luminance.

Do the same for the LRGB but you are more working on the colour, saturation, the stars, the star sizes, the areas of interest selected out and tweaked etc.
Now add a new layer and add the processed luminance this gives LLRGB really and you will see it make the image more luminescent.

Do final tweaks using masks usually for noise control (near the end of the processing) and final enhancement of details and colour.
Save final image. I save each version with V1 V2 suffix so I can revert to an earlier image if things go wrong.
I do a clean up of any background artifacts using the healing tool (there are often a few odd coloured pixels that get through or satellite trails etc).
I reduce the final image to 8 bits, 1200 x 1200 or thereabouts and save as a jpeg maximum settings for posting on the net.

You need to know how to do inverted masks, layer masks and levels and curves and know the basic tools of Photoshop and how they work.

That's it - piece of cake!! phew!

Greg Bradey

[DavidTrap (David)]

So can you tell me how to do it in PixInsight and how to use all the tools in clear easy to understand language?



I'll try to digest what you've written later!

DT

[cventer]

Greg,

Good post.

Be great to have this worflow for OSC Cameras as well. Many beginners use OSC be it QHY or DSLR

Also a workflow using the typical free or cheaper tools like DSS and Nebulosity would be awesome.

[gregbradley]

Originally Posted by DavidTrap

So can you tell me how to do it in PixInsight and how to use all the tools in clear easy to understand language?



I'll try to digest what you've written later!

DT

I am a beginner when it comes to Pix Insight. So I can't help you there. I intend to learn it soon though.

Originally Posted by cventer

Greg,

Good post.

Be great to have this worflow for OSC Cameras as well. Many beginners use OSC be it QHY or DSLR

Also a workflow using the typical free or cheaper tools like DSS and Nebulosity would be awesome.

I have done OSC and DSLR. I used Images Plus which is popular but does cost around US$200 or so.

Here's what I did for that:

1. Create a dark library for one shot colour cameras that are cooled. Pick a couple not 10. Make a standard exposure length and a temperature you can achieve almost all year round or 2 temperatures - one for summer and one for the rest of the year. Shoot darks by simply setting the controlling software to dark and shoot it at night or in a darkened room with the lens cap on or at night.

You can do this with a DSLR as well. Its a bit trickier with DSLRs because they are at air temperatures and of course that varies not only throughout the year but throughout the evening. They will be less noisy on cold winter nights and most noisy in summer.

It is also best to shoot darks with a DSLR and use a DSLR with the power adapter rather than using batteries if you can. Batteries heat up with use and change the noise of the camera. The power adapter does not.

With DSLRs you also need to standardise your ISO as well as the exposure length.

ISO of 400 or 800 is usual. Above that you are only amplifying what is already there so much like boosting it afterwards with software. You didn't really gain anything except stretch what was already there.

DSLRs in light polluted areas will pick up sky glow faster than dedicated astro cameras so work out an ideal exposure length for your area.

3-4 minutes at ISO800 may be close to ideal in a typical urban setting.
You may also need a light pollution filter. Survey shows Hutech LPS-2 or IDAS are the ones to use.

So take about 12 darks or more. Why so many? Because there are random artifacts that the combining process removes so the more the merrier in that regard.

use sigma reject combine for making master darks out of the 12 you took.

Now take a flat field. Put a white t-shirt over the end of your scope and make sure there are no wrinkles. You may need a large elastic band to keep it taut.

At dusk point your scope to a clear area of the dusk sky and take an exposure to about 1/3rd your DSLR histogram and very short exposure.
For one shot colour that is around 20,000ADU. Take about 6.

With flats make sure your scope is in focus and the camera is in the same orientation that you will or did take your images at. If you change that they will not be valid.

Also make sure if you are using a one shot colour camera that is cooled that the camera is cooled to the temperature you will be imaging at.

Also make sure you take fresh flats if you clean your camera window or filters as flats remove the effects of dust shadows in your images.

Take some flat darks (3 to 6) meaning take a dark image of the same duration as your flat exposure.

Now use your astro software to create master darks and master flats.
Images Plus does that. Use median combine and subtract the master flat dark from the flats to make the master flat.

Ok so now you have flats and darks for your camera.

2. Take your exposures with exposures planned as above (exposure time is determined by light pollution, type of camera - DSLR shorter, one shot colour longer so DSLR maybe 3 minutes and one shot colour may be 10 to 15 minutes. Take lots like several hours. Have patience, good images need long exposures - like 3-20 hours worth. Its better to stay on one object for several nights than take 10 short images of several objects none of which will turn out that great (easy to say hard to do).

3. Now you have your images, your matching darks and your flats.

So you callibrate the images. That means removing the dark noise, correcting the vignetting (shadowing in the corners) and the dust shadows (called dust donuts) and uneven illumination of your scope and camera with flats.

With DSLRs and one shot colour for that matter, you can use adaptive darks. This means as long as the dark image is accompanied by using a bias (very short exposure of the read noise at the same temp as imaging, say 6 shots sigma reject combined) then the darks can be applied via software to different temperatures and exposures. This is handy if you did say a 5 minute exposure and only have 10 minute darks.
It is also handy for DSLRs that do not have temperature regulation for their sensors.

4. Now you have subtracted the dark noise and corrected the dust and uneven illumination with your flats.

Your image is probably pretty dull and hard to see. The image needs to be stretched/boosted to be able to see it. In Photoshop that is the levels/curves tool. In Images Plus it is the DDP tool.

Boost the image so you can see the image but do not clip the data. Clipping means to cut off some of the data - usually it is done when trying to make the background dark and you cut off some of the faint background data with it so watch the histogram and make sure the
histogram forms a nice bell curve and you do not see it hard against the left with part of it being a vertical wall on the hard left.

Next step is to align the images. The images are aligned. Images Plus has some great alignment tools for that. Try out a few and see which ones you like best.

Some split the image into luminance and red green and blue here. They do that so they can process the luminance separately and differently to the rgb. RGB is usually processed to enhance colours, adjust colour balance to get rid of a colour bias. Luminance has all the details so it is usually sharpened, increase the contrast etc.
You recombine these at the end. I am not sure if you need Photoshop to do that with its layers ability or if you can do that in Images Plus (you probably can).

You don't have to do that step and it depends on the image. In an image
where you want maximum details you'd do it (like a galaxy or detailed nebula) but on a diffuse nebula it won't gain anything so its not worth doing.

This is a basic overview to help you get started. Ron Wodaski's books are good and Images Plus is the king software for one shot colour or for DSLRs and there is an excellent group that offers advice and tips for it.

I hope this helps.

Greg.

[jase (Jason)]

I originally produced this for Lightbuckets members who were just starting out with RGB filtered imaging. I thought I'd post it here as well given I didn't want such information to go into a blackhole, never to be seen again.

Other the overall work flow (which is a couple of years old now I should add), the focus of attention was on normalising of data. This is a very important step in a workflow, but I found the term is often used loosely, hence there can be confusion as to the definition. Normalisation is a processing activity that equalises data for two purposes. These are;

Normalisation for data rejection
Normalisation for colour combine

Normalising for data rejection typically occurs after you have registered the subs and hot/dead pixels have been removed. The act of normalising the data to the same level (equalisation) across the sub frame set is vital in determining what data is considered outlier i.e. what data does not conform and should be considered noise or another form of erroneous data and be rejected. If the sub frames are not equalised prior to the data being combined, the data rejection algorithm will fail to determine what is "correct" or normal value of specific pixel. Depending on the tool you use, normalisation is automatic. In MaximDL there is a small check box in the combine dialog box asking if you want to normalise the subs before combining the data. In CCDStack, the normalisation process is more of a manual task were you have the opportunity to select in the image data what constitutes as being a highlight (bright area) and background (dark area). Once normalised, the average values of the pixels amongst the subs should be relatively close to each other.

Normalisation for colour combine is performed on the R, G, B master frames. That is the sub frames per colour channel that have already been combined to produce a noise free master frame. The act of normalising the R, G, B master frames is to equalise the background colour between the frames so that the correct colour weights are applied. For example, you may have blue filtered data that was collected when the moon was rising, as such the background is brighter then the R and G filtered data. Without normalisation of the background, the resulting colour combined image would have a strong blue cast across the image. Normalisation simply alters the pixel values across the three filters master frames (R,G,B) to similar values (again equalises).

The attached processing 101 flow chart highlights where normalisation is performed. The first set is the data rejection i.e. prior or during the combine algorithm executing. Then, the colour combine normalisation that occurs only on the RGB master frames.

The work flow listed is dated though is proven to work having produced several images using the flow in the past. Of course should you have any questions, please PM me or establish a new thread for discussion.

Cheers

[solissydney (Ken)]

Which version Photoshop are you referring to? All of them?
I use photoshop 7

[sanusense (Sanusense)]

it will be easier if you are using photoshop 7, very user friendly.

[salimant (Anthony)]

thanks jase