Masked Stretch FTW! M101 Widefield, Unmodded DSLR with star colour to the cores!

[IanL]

This image was taken on my first night out following the street light switch-off in the nearby towns (between midnight and 5:00am), helpfully the moon was also on the opposite side of the sky by this time. As you can see, I have also also managed to capture a fair few of M101's companion galaxies in this image, including NGC5474 just left of top centre.



This is also the first image I have processed using the new 'Masked Stretch' process.  It has made a massive difference to retaining colour in the cores of the stars and the galaxy, so much so that I may well have to go back and reprocess some of my earlier images which suffer from a lack of star colour (a common problem with DSLR images).

Imaging: Sky-Watcher Evostar 80ED DS-Pro,  Sky-Watcher 0.85x reducer, Hutech IDAS LPS P2 2", Canon EOS 500D (Unmodified), APT
Guiding: Orion ST80, QHY 5, PHD guiding, Sky-Watcher NEQ6, EQMod, AstroTortilla
Processing: PixInsight 1.8.01
Date: Jan. 12, 2014
Lights: 30 x 600 seconds ISO400
Darks: 109
Flats: 102
Bias: 330

[Warhen]

Fantastic Ian. I love the MS script as well! You used the script as the primary stretch, correct?

[ajbarr]

I guess this is the same question Warren asked but I assume you use masked stretch on the linear fit file and then continue to stretch it with histogram transformation?

Thanks

Albert

[pfile]

Fantastic Ian. I love the MS script as well! You used the script as the primary stretch, correct?

warren, did you miss it? 1.8.1 has a process version of maskedstretch. really fast.

rob

[IanL]

I used the new masked stretch process (not the earlier script) as the first stretch instead of a standard histogram transformation. It does result in less contrast between structures, but these are easily recovered using HDRMT and a few extra histogram/curves tweaks.

The star colours don't leap out immediately, but using curves targeting saturation brings them out very nicely indeed.  No more messing around trying to push colour back in to the star cores which tend to get blown out by a normal HT.

[Coco]

Glad I found this post, was getting frustrated with DSLR star colours..

Interesting why 600 seconds at low ISO? I thought the norm was ISO 1600.. its definitely working however..

[pfile]

if you have a lot of light pollution often times you'll blow out the image after just a couple of minutes at high ISO. so if you want to expose longer you have to take the read noise hit and move to a lower ISO (though i think the unity gain on these newer canons is much lower than ISO1600)

rob

[IanL]

Read noise on my 500D is somewhere in the region of 10e at ISO400.  Unity is somewhere between 200 and 400.  There might be a marginal read nouise improvement at a higher ISO but I've not had the inclination to measure it.  I go by the results I get in my images, and longer exposures at this ISO seem to work best for my conditions and equipment.

I know people who swear that hundreds of 60s exposures is the way to go, and others that think fewer longer exposures work best.  At the end of the day there is so much contradictory advice (and much of it not supported by basic theory) that you could spend a year reading it and be none the wiser.

For my 2p's worth, basic sampling theory says that the number of minutes of exposure will govern the SNR, so fewer longer exposures will introduce less read noise (assuming you have sufficient exposure for dithering and pixel rejection to work effectively).  Of course you then have the whole photon shot noise thing in the background regions, but I have yet to fix that properly in my own mind.

[pfile]

well in fact if your exposures are so long that the sky background completely swamps the read noise, then it kind of does not matter. that's the theory behind sky-limited exposures. so i guess when i said "the read noise hit" it does not really make sense.

as you said the bottom line is the only thing that puts more photons on the sensor is exposure time. that's how you drive sub exposure SNR up.

rob