[MossyRocks]

In most posts of this thread supposed to be a guide to astrophoto, 20 seconds or 30 seconds exposures times are used at ISO around 3200 and 6400. Those settings aren't best. Pentax astrotracer with slow shutter speed noise removal can deliver best quality between 2 minutes and 4 minutes of exposure, that's 4 stops lower ISO => something like ISO800.

It all depends on the length of the lens. There is no way I would be getting even a 2 minute exposure with my 300mm lens or my 400mm but with a 50mm or 28mm that seems doable without issue but astrotracer is pretty optimistic with it's estimates of total exposure time. I don't typically shoot wide field astro shots and even when I do I prefer shooting with astrotracer and stacking. I don't make use of the slow shutter speed noise removal as one will get better results by stacking frames. If clouds are going to roll by or during tear down I will go and shoot dark frames which are frames that use the same exposure settings as the ones with actual images but are of the lens cap and astro image processors like Deep Sky Stacker will make use of these too and will subtract out the systematic noise that is found by averaging dark frames. Basically dark frames are use to remove the systematic noise from dark current in the sensor and are dependent on temperature. There are also bias frames that are shot at the same ISO as the actual image frames but these are shot at the fastest shutter speed and again are images of the inside of the lens cap. You really need to shoot a lot of these and by a lot I mean 200 to 300 but they are not temperature dependent so can be shot at your leisure. These frames capture the systematic noise introduced by the signal apm circuit, known as amp glow, which is another source of noise. The last bit of noise that should be left is the random thermal and noise and random photon quantity noise and this can only be resolved by taking more and longer images so that you can drive noise down. Depending on the ISO you had to shoot at and how dim of an object you were chasing you will need varying numbers of shots. The dimmer or more washed out object you are shooting the more shots you will need to get the noise floor to a usable level below the object. A 100 shot stack will have 1/10 the noise of a single image in that stack but often that isn't enough and you really want several hours of total exposure with good images for deep sky objects. This often means shoot even more hours worth of images and then only stacking the best 25% to 50%.

As far as what ISO to use I find that I am often shooting in the ISO 400 to ISO 1600 range depending on what I am shooting and the conditions I am shooting in. In my massively light polluted backyard I have to use a lower ISO otherwise the sky glow just dominates the image, even with a light pollution filter. My goal in choosing an ISO varies based off of what I am shooting. If it is something bright like M45 or M42 I will be shooting at ISO 400 and now maybe even ISO 100 from my backyard. However when chasing after really dim things like M51 or M101. I will be shooting at ISO1600 or even ISO3200 as they appear just above the noise floor from my backyard

[biz-engineer]

It all depends on the length of the lens...However when chasing after really dim things like M51 or M101. I will be shooting at ISO1600 or even ISO3200 as they appear just above the noise floor from my backyard

Thanks for that explanation.

---------- Post added 23-09-19 at 20:14 ----------

A 100 shot stack will have 1/10 the noise of a single image

That's a lot of stacking. I found that beyond 16 stacks the improvement in noise becomes marginal with resolution limited by frame alignment rather than noise. On the Pentax K1, , it is possible to perform pixel shift exposures with tracking, and the dark frame is recording and subtracted after the pixel shift exposure, up to 30 seconds long, which give a good noise/resolution performance point. Not bad.

[MossyRocks]

That's a lot of stacking. I found that beyond 16 stacks the improvement in noise becomes marginal with resolution limited by frame alignment rather than noise.

I think here we may be going after different things in the night sky. I go after dim objects that even under good conditions are difficult to see as naked eye objects while others are not, it sounds like you might be going after things that are substantially brighter like the milky way or substantially larger.

I think there it really depends on what you are shooting and what you are using for tracking. I didn't think any of the pentax cameras would allow pixel shift with astrotracer but if you are working off of an equatorial it might work but that seems like it would require some very accurate tracking other wise it might turn into a hot mess. Even wide shots at 30 seconds using pixel shift seem like they would lead to sub optimal results as you would want to really be using the rule of 300 so a lens no wider than a 10mm to avoid star trails.

What software are you using for stacking? If you are using Photoshop then yes going much beyond 16 frames doesn't provide much benefit as it kind of sucks at alignment and you are really limited to 16 bits per channel so it isn't long before you have exaused that space when using a K-1 which gives some nice 14bpc images. I use DeepSkyStacker (DSS) for stacking images of the night sky that aren't the moon. Out of DSS you get a 32 bpc (96 bit color) TIFF. From there you want to do some initial stretching using levels in Photoshop before converting to a 16bpc (48bit color) image. I have actually taken to working more in GIMP for initial edits now as it allows more things to run with a 32bpc image so you can do some initial noise reduction, stretching and curve editing before tossing out a bunch of info. I have found that curves in GIMP don't seem to produce quite as nice of results as the ones in photoshop but I think that is more me not understanding the difference fully.

I've stacked moon images but for those I up sample them to 2x resolution first and for alignment I use the Hugin tool align_image_stack for aligning the individual images since unless the moon is really huge in the frame photoshop wanders off and gets lost. I usually do multiple iterations of alignment, first a course x,y alignment, then a fine x,y one, next a fine one for rotation, and finally a super fine x,y one. What I am trying to accomplish is a super-resolution version of my moon shot when I do this. I usually do this to make up for the fact that I want one of those impressive moon images that people take with giant lenses and telescopes but the biggest I own is a 400mm so I try to make the best of what I have.

[biz-engineer]

I didn't think any of the pentax cameras would allow pixel shift with astrotracer

What I wrote is that the Pentax K1 and K1 II allows for pixel shift with astro-tracer enabled.

Even wide shots at 30 seconds using pixel shift seem like they would lead to sub optimal results as you would want to really be using the rule of 300 so a lens no wider than a 10mm to avoid star trails.

The reason pixel shift does work with astrotracer is because the sensor moves relative to camera body, but remains virtually at a fixed position relative to the stars.

it kind of sucks at alignment and you are really limited to 16 bits per channel

Strange. I guess we aren't referring to the same things here. I think if you need more than 16bits depth per channel for your images, you are looking for imaging beyond what can be coded by raw files, displayed and printed as an image.

---

Last edited by biz-engineer; 09-23-2019 at 10:51 PM.

[slartibartfast01]

The reason pixel shift does work with astrotracer is because the sensor moves relative to camera body, but remains virtually at a fixed position relative to the stars.
.

Doesn't the sensor position reset after each shot using astrotracer? If it does then the four shots required for pixel shift will be offset from each other in the same way that repeated astrotracer shots can be seen moving across the camera field of view ultimately requiring the camera position to be reset to return the desire target to the field of view.

[biz-engineer]

Doesn't the sensor position reset after each shot using astrotracer?

The sensor position doesn't reset during between pixel shift frames. On my K1, it works the following way:
- sensor tracks star motion
- 1st ps frame is exposed
- sensor shifts 1 pixel from last sensor position
- sensor continues to track star motion from last position
- 2nd ps frame is exposed
- sensor shifts 1 pixel from last sensor position
- sensor continues to track star motion from last position
- 3rd ps frame is exposed
- sensor shifts 1 pixel from last sensor position
- sensor continues to track star motion from last position
- 4th ps frame is exposed
- sensor position resets
- mechanical shutter closes
- black frame exposure (if low shutter speed NR is enabled or auto with exposure longer than 1 second).

So I effectively get full RGB definition of seemingly fixed stars. If there is a foreground, if it only selected from a single frame if motion correction is enabled, if motion correction is disabled then the foreground will have lots of pixel shift artifacts since the sensor was not at the same position for each of the 4 pixel shifted frame.

[slartibartfast01]

The sensor position doesn't reset during between pixel shift frames. On my K1, it works the following way:
- sensor tracks star motion
- 1st ps frame is exposed
- sensor shifts 1 pixel from last sensor position
- sensor continues to track star motion from last position
- 2nd ps frame is exposed
- sensor shifts 1 pixel from last sensor position
- sensor continues to track star motion from last position
- 3rd ps frame is exposed
- sensor shifts 1 pixel from last sensor position
- sensor continues to track star motion from last position
- 4th ps frame is exposed
- sensor position resets
- mechanical shutter closes
- black frame exposure (if low shutter speed NR is enabled or auto with exposure longer than 1 second).

So I effectively get full RGB definition of seemingly fixed stars. If there is a foreground, if it only selected from a single frame if motion correction is enabled, if motion correction is disabled then the foreground will have lots of pixel shift artifacts since the sensor was not at the same position for each of the 4 pixel shifted frame.

OK so it takes all four pixel shifted images in one exposure. I always thought four separate images were taken. I guess each pixel shifted image is saved as a separate file to be combined later. Does normal pixel shift work the same way and only open and close the shutter once? I always assumed it was multiple shots.

[biz-engineer]

I guess each pixel shifted image is saved as a separate file to be combined later.

No , each frame is saved in memory , the black frame is subtracted from each of the 4 frames, and the resulting frames are packed into a single file saved in SD card (when using RAW format).

Does normal pixel shift work the same way and only open and close the shutter once? I always assumed it was multiple shots.

The mechanical shutter open and close once to avoid vibration between frames, each of the 4 pixel shift frames have their exposure started and stopped electronically.

[slartibartfast01]

No , each frame is saved in memory , the black frame is subtracted from each of the 4 frames, and the resulting frames are packed into a single file saved in SD card (when using RAW format).


The mechanical shutter open and close once to avoid vibration between frames, each of the 4 pixel shift frames have their exposure started and stopped electronically.

OK thanks, you can tell my camera doesn't have pixel shift can't you.



---------- Post added 09-24-19 at 11:58 AM ----------

No , each frame is saved in memory , the black frame is subtracted from each of the 4 frames, and the resulting frames are packed into a single file saved in SD card (when using RAW format).


The mechanical shutter open and close once to avoid vibration between frames, each of the 4 pixel shift frames have their exposure started and stopped electronically.

Now I am confused. The manual for the K1 says pixel shift and astrotracer cannot be used together.

[biz-engineer]

Now I am confused. The manual for the K1 says pixel shift and astrotracer cannot be used together.

Yes. On my K1 pixel shift and astro-tracer can be enabled and work together in M mode (max exposure time is 30 seconds), but not in B mode. Either firmware update removed that restriction, or the manual is about B mode, or the manual isn't up to date vs firmware.

[bikehead90]

Which lens would be recommended more? Wide angle or telephoto? This is a type of photography I’ve been trying my best to get involved with.

[MossyRocks]

Which lens would be recommended more? Wide angle or telephoto? This is a type of photography I’ve been trying my best to get involved with.

The correct question you should be asking yourself is what do you want to photography in the night sky?
Star fields and the milky really demand a wide if not an ultrawide lens
Constellations depending on their size would be a normal lens to a wide
Deep sky objects like various nebula, galaxies, star clusters, etc. are all things that need telephotos of various sizes. One of the largest deep sky objes in the night sky is M31 the Andromeda galaxy and it will basically almost fill the frame on my K-3 with a 400mm lens. A larger object still but it is darker is the veil nebula which is hard to frame the whole thing in the frame with a 300mm lens. A 200mm lens frames up a lot of the nebulae in Orion nicely and even will work with smaller objects but you will have to crop the image. A 100mm lens would give you a bunch of nice wider shots of things. A 600mm would give some nice shots of smaller messier objects if you crop them. If you want to go longer than 600mm you would be better off getting a telescope and necessary adapters and corrector optics as you will need a lot more than 600mm, think 2000mm to 3000mm. A big telescope also will give reasonable shots of the planets.

So to answer your question: Yes I recommend big telescopes, ultra telephotos, telephotos, normals, wides, and ultrawides for astrophotography. Buy fast primes that don't need to be stopped down much or any to be sharp.

[wings]

Wow. Thanks for this well written primer for AP.

[MossyRocks]

Wow. Thanks for this well written primer for AP.

Glad to help. lots of people ask the same questions over and over again so I just answered them and pointed out some other pitfalls when starting out.

[Pentikonian]

Thanks a bunch for taking the time to write this, and answer questions.

I spent time this evening reading through the thread and trying to digest the info. One of the projects I set for myself this year is to try some astro between Oct and Feb.

I figured I'd start with something easy to find, like the constellation Orion. Probably with the DFA*50 on my K-1.

After getting the basics down, I hope to tighten the view a bit and try to isolate the Orion Nebula.

Debating whether or not to invest in a light pollution filter. There's still a mix of LED, Incandescent, Sodium and Mercury lights here, and I doubt I'll be able to get out to dark skies for most of these attempts.