Thanks for the feedback. To answer some of the questions -
I started off taking 1600 ISO shots of 30, 60, 120 & 240 secs and placed them in the marked folders. However someone suggested that ISO 1600 wasn't the best setting to test noise, so I then took 800 & 400 ISO equivalents of each. Because the film speed was being 1/2'd in each case, I had to double the exposure times to compensate so that the dark frames had the same overall exposure. These make up the other 2 files in each folder.
My high speed noise reduction was set to default. However I've previously tried all the settings including "disabled" , which provided minimal difference.
Also for reference I do of course have the camera's in built dark frame subtraction routine disabled as this will be handled by DSS.
These examples are straight JPG's from camera in fine quality mode. Not RAW converted to JPG. I know RAW is better, but I've already been emailed some dark frame samples in native JPG format from a Canon 40D that's virtually all black bar a few pixels @ around 2 mins. Look at my ISO 1600 2 min shots (or the 800, 400 ISO equivalents) and you'll see heaps of noise.
Shooting temp was around 24 degrees c. I shot these indoors with the Aircon set as this temp, no lens, just a cap over the lens opening, and subdued lighting.
I've heard a few comments that the Canon 40D was the last decent camera, before the pixel density on the newer (same size) chips caused these noise issues. That seems to make sense, and seems to be backed up by dark frame samples I've been emailed. However the one niggle is the article link I orginally posted which specifically tests the 500D's dark frame abilty with long exposures up to 5 mins, and shows very cleans results.
David...this noise doesn't seem right. Something is not right here wit your noise subs.
The canon have always been touted as lower noise compared to most others and I thought the 500D to be a step better than mine yet they appear a step backwards by a quantum.
Have you contacted Gary Honis to compare or Canon?
If you refer the link with samples on my orginal post, I ran tests @ 800 & 400 ISO (doubling the shutter speed in each case to provide the same overall exposure). Same noise.
Now try taking 19 or so of those ISO-400 dark frames and median combine them.
Don't stretch them or alter them in any way. Take them straight from the camera (ensure that the viewfinder is covered with the viewfinder cover which is attached to the strap when taking them) and median combine them in the astrophotographic processing software of your choice.
Post your resultant master dark frame -- no processing done on it.
For reference I have been using Darks, Flats & Bias in DSS, however I was starting to get artifacts (blue speckles in final image) on longer exposures over 1 min. I traced this back to the the orginal quality of my darks frames, hence my interest in comparing the initial dark frame output quality between difference DSLR's.
Below is example of post processed image (100% crop) from DSS, before processing in Photoshop.
1. are you taking your dark frames immediately after shooting your light frames?
2. how many dark frames are you typically stacking?
You simply do not need to shoot bias/offset frames. The bias is included in your dark frame. It is automatically subtracted from your lights via the master dark.
Bias frames should only be used when you're using dark frames which weren't taken at the same temperature/after exposing your light frames. I consider DSLR dark frame libraries to be an astrophotographic sin. The consistency is not there, unlike a dedicated cooled CCD. And, even then, people flush their libraries once a month, or once every couple of months.
Well I can now confirm (thanks to Fiona at our DSS imaging group) that my Canon 500D IS much noisier than typical. Last night we compared 4 min ISO 1600 exposures produced by 2 x 500D's .. and the difference in noise was dramatic. I'll post examples later today.
Thanks for your suggestions & feedback. I typically take 50 lights & 25 subs, and they are always created fresh in the same session over a few hours. My experience has been that initial results shooting at 30 secs worked well, and it it was possible to see the improvement through Deep Sky Stacker of combining lots of data. It's only since I started trying guiding and longer exposures 2min+ that I've had this noise issue using the same methodology that worked before for shorter exposures.
I'm about to post further results from my 500D vd 500D comparison, and this demonstrates my issue .. eg. that even before manipulation I'm handicapped.