ELS

I took another crack at processing the image and got way better results.
I also reduced the light pollution glow by manually raising the black level on subsequent copied layers and blending them, there's probably a tool made for this but I couldn't find it. A gradient wouldn't really cut it since it's more of like a diamond shape, and also that would lack finer level control



 

Some NR and a bit more color correction:

13 minutes ago, Elp said:

Perhaps read my comment again. I could see the milky way in my single 30s images without editing out of camera.

With post processing and stacking I could reveal far more.

The point of my comment is, you maybe need to expose per sub longer than you are if star trailing isn't an issue. One of the last videos posted by the late great Alyn Wallace demonstrated this perfectly, he did the same subject tracked and long exposure with one rig, and very short exposures fixed camera with another. The latter he basically said was a waste of time compared to the former.

Yeah that's what I'm saying, I got way better results with the 13s exposures than the 8s ones despite having nearly 2x the total exposure time with the 8s ones.

50 minutes ago, Elp said:

Depends on camera really and your light pollution. With my advanced compact which isn't really designed for AP I've revealed the milky way from a bortle 4 zone with a single 30s exposure.

Perhaps you haven't processed them enough. I fiddle the curves on these images a ton. Without a ton of tweaking you can't see anything.
Try raising the black level/the lower left point on the luminance curve for each channel until you start to blank the noise level and get the starts to start contrasting against the background (I usually have to shift the mid point on the levels (not curves since it's easier to use the levels adjustment which gives a simple midpoint adjuster) until I can actually see what I'm editing. Then do the noise blanking.
And then just tweak the curve afterwards to get the contrast in the parts I want. usually having the an S figure on the left side to highly contrast the low light objects.

Here, 1 is the stacked image unprocessed. 2 is just shifting the luminance midpoint down to brighten up the image and make is visible. The overlay is the color curves added for picture 3 to blank out the noise. and 4 is the "final" image and the corresponding curves applied to it overlayed

50x6s exposures at 90mm.

Don't judge the horrid state of the lens, I didn't notice it was this dirty until after I stacked the photos. Maybe it's just the solar flares creating the artifacts


Here's also 19x13s @ 18mm...
 

278x8s exposures here. But the sky wasn't as clear as on my previous stacked milky way image that had 100x13s exposures. This time it seems the dark frames worked, for which there are 50x10s dark frames (yes I know, I should've matched the exposure time)
It was hard to get the colors right, and they still are jacked. I think the problem is at such low light input the linearity is unworkable. Most likely the response curve is also very unstable in that part, not too surprising for a 15y/o sensor. So I should probably try to maximize the light input instead of just total exposure time.

 

Well, would it really change dispersion? It would be the same as in air that was 25% longer. So the dispersion isn't different for the required length of each and cancels out.
I guess it's just cheaper and easier to build a longer line than buy a bunch of very clear fuel. Which would probably also come with higher annual costs because it would be a fire hazard.

I'm more a fan of the old LPS lamps, not HPS which are what you're probably thinking of.
LPS had a very narrow spectrum, so you could just filter them out. Also they maintained the night vision of your eyes, so it didn't take 10+ minutes every time you drove through a city to get your night vision back.
Animals also minded them less.

Wouldn't it cut down the required length of the delay line by about 25%?
Is it too hard to get good enough optical performance out of a fluid? I presume it would for glass, but you could use a hydrocarbon with a high refractive index too, no?
 

Modern Sony sensors pretty much blow away all CCD options available, Theoretically CCDs could have better performance but since there has been barely any development in them since CMOS became popular, they're not comparable to CMOS options.
And when it comes to CMOS options, Sony sensors are generally the best in noise specs. And much more available than Canon sensors.

13 hours ago, happy-kat said:

If your image was expected to be wider then you could try; 

Break your 100 images into four stacks of 25

Stack each 25 and then process 

Stitch your four finished images together

Nevermind that, I was expecting it to be much more, but it is about right.

I tried adding flats, darks, biases, dark flats. And it seems to give me a less noisy result, but it's even worse in the vignetting. And I still cant get the colors right.
What I think is happening is that the noise gets amplified and since there are twice as many green channels, it looks as if the white balance is off, when it actually is not. I don't really know how I could fix that, I guess raising the black level on the green channel more but it doesn't seem to work like I was expecting it to.
And the vignetting might just be that my flats aren't right. idk.
I'll try to see if I can process chroma and luma separately so that the fainter features, if the noise is too bad get turned to BnW instead of green.
I attached the TIFF unprocessed master. another try with calibration.TIF

That wasn't my question.
I meant that since the sensor is gonna heat up anyway, wouldn't it help to warm it up before to the final temp so the calibration frames are accurate across all the frames instead of just the latter ones?

It seems like it would be a good idea to warm up the sensor before taking the exposure(s), so the noise pattern is more stable and your calibration frames are more accurate.
Does anyone do this? does it help? and how long should the exposures be to warm up the sensor enough to be close enough to the final temperature for a good result?

I get it now. Also makes sense why it wouldn't be common, since it would introduce all the problems of lenses, and would be better to avoid.

I haven't seen any design which does this, have the light cross over before hitting the eyepiece... It would seem like that would allow you to have a smaller secondary mirror, as well as allow you to use shorter focal length primary mirror.
Does it cause some wacky diffraction problems or what? Or does it have a specific name for it?

Interesting noise patterns in these.
I remember why I didn't like DSS... The C28 cluster images only got 9, 5" frames stacked because DSS has a low limit on the threshold for accepting frames, So it discarded all the other 81 frames. I was actually aiming at M33 but somehow ended up way off

There might be some galaxy in there, just need to figure out which blobs they are

Pleiades was brighter and it accepted all 21 4" exposures.

The Perseus cluster image was 100 6" exposures @ 90mm

Hard to get the color accurate on the larger stacks, seems to get very non-linear. I might have to stop being lazy and take dark and white frames and calibrate everything.


:D

Stacked 100x13" exposures @ ISO3000, F3.5, 18mm. using DSS.
The amp and driver noise is incredible, really shows the 400D's age. DSS for some reason didn't properly frame the final image and cut off a lot of it. I was kinda relying on the sky movement to get a wider shot (I pointed the camera vertically)
Apparently with enough exposures stacked together the banding cancels out quite a lot. So perhaps that's an advantage of stacked exposures over longer ones.

3 hours ago, Elp said:

You've got a result which is the important thing, but now to refine.

15s is too long, even at 25mm, you can tell with your star trails. You need to dial the exposure back.

Iso3000 is also way too high, you'll only introduce excessive amounts of noise (gain or iso is only an electronic amplifier to the received light, it doesn't increase sensitivity at all and cannot cheat the laws of physics) when I had a Canon I didn't go above 1600.

I'd advise to also get a tracker, even if you DIY your own barndoor.

You can take short exposures, but you need thousands to get any sort of result and stacking them is the way to go, DSS and Siril are incredibly easy to use (especially the former). In fact I saw a YT video fairly recently where the person was using two cameras, one I think 30s tracked, the other 10s or so fixed, the latter didn't pick up much other than stars, you'll get far more of a result with tracking. The only DSO targets I'd attempt fixed are M31 Andromeda galaxy and M42 Orion nebula as they're so bright, but without tracking you'll struggle to get any outer details.

Well with the small aperture 15" makes it so there's barely any elongation and only in the middle. And since the SNR is the main limiting factor, I'd rather crank the exposure even if it distorts a bit.
I found that the noise becomes invariant above 1600 so I maxed it out... The bloom sort of compensates the loss of DNR on brighter stars

Well I processed the andromeda pictures I took the same day, 14*4" exposures, this is the result... The color is more accurate but with this short of an exposure the galaxy just isn't bright enough to get over the noise floor.
Btw I used DSS not Sirilic for the previous (and this) picture. I also found that it has a setting to remove hot pixels. Also the automatic star mapping + offsets finally worked and I didn't need to manually set them. So I guess it ain't that bad after all

41 minutes ago, Paul M said:

Good start. The stars are sharp so not suffering much trailing. 

I'd say the next step would be to take a bunch of images like that and stack them DSS or similar software. 

The Milky Way is my nemesis, Not yet made a publishable image of it, so you are ahead of me!

 

Well it was quite clear the night I took the picture, I could clearly see it with my bare eyes even.
In a way the camera had a harder time seeing it than me
I don't like stacking because the free stacking programs are a nightmare to use in my experience...
Also apart for Siril (but even then) they don't have very good modules. I like darktable since I can enhance the very underexposed and noisy images a ton, while with the stacking programs I'm left to processing the exported one... I guess I could stack individual exported darktable images but that would be an even bigger nightmare than it already is.

Since my current images aren't much to write home about, I'll attach some other passable ones I took earlier this year instead of making a new thread:

There's a 10x8" stack @ 3000ISO, 300mm, f5.6 of Antares and Messier 4 beside it which came out pretty amazing, taken in the middle of May. Even can see a bit of NGC6144
Stacked it using Siril, then adjusted the luminance curve more in paint.net

There's a 22x10" stack of Andromeda, settings the same as before. This time stacked using Sirilic
Afterwards adjusted the lum curve in paint.net.
I did also take a shot at capturing M33 but apart for a slight change in the noise patterns, there's nothing. I'll try to take another shot now that the sky's clearer and it's further from the horizon, but I'm just aiming it by eye, and at the small FOV it's not easy to get it in frame.
In fact this picture of andromeda was sort of pointed by eye too, I could only see the star at 7 o clock, not even the one on the right.
I did take some pictures of andromeda at the same time I took the milky way one but I've yet to stack them.

There's also one I took even earlier.  I was messing around with various attempts to boost the SNR. You can see at the top the stacked one which I thought was a dud before I tried post-processing it in paint.net.

Looks like it's from 1979 or something
Canon 400D @ 25mm F4.0 ISO3000 and 15 second exposure.
It should improve a lot once I get a startracker with longer exposures. and or a wider lens.... But it's pretty descent for what it is.
I also attached the raw file if anyone's interested in looking at a black image

Unfortunately darktable doesn't seem to have a debanding module so there's that problem.

IMG_9356.CR2