MW Panoramas: Optimizing time for foreground images

nighthiker wrote:

indigoshrine wrote:

nighthiker wrote:

Hi!

Sorry, but the usual suspect again

I just wanted to comment on the internal NR: I wouldn't use it as there are better ways to reduce noise and hot pixels:

1) The inbuild NR is a dark-frame subtraction and hence it's very time-consuming. A much faster way would be to take the foreground pictures without NR and take one or several dark-frames afterwards (with lens cap on). Taking several dark-frames would have the advantage of averaging to reduce non-stationary noise. You can subtract the dark-frame with any astro software or simply as layers in PS and 'subtraction'

2) There is also software for noise and hot-pixel removal. The modern RAW-converter will already remove most hot-pixel and the 'dust and scratch' filter in PS is a very powerful tool to remove remaining hot-pixel. To be honest: I only have that problem on old images where I took >10 min exposures for startrails (lesson learned). The modern Topaz software is very good for noise reduction and sharpening, but also not that cheap .....

My time for the MW has already ended and I will get astronomical darkness in mid-August again. Due to timing- and weather restrictions I wasn't able to take any serious MW pics during the last winter (but I got compensated by the Aurora).

Appreciate the feedback! It's tough living so far up in the north - but then again, you do have the Aurora

The internal NR is of course switched off for the sky images, I use it more for the hotpixels (due to very long foreground exposures), less for the noise. Until end of last season I would use PS 'Dust and Specles' filter which generally works well but not perfect. It always leaves some manual clean up work, also it will introduce a tiny amount of image degradation. Tried the internal NR and thought the results look really clean. I like the fact you get it all in one file and no further editing step necessary which could also degrade the quality. But ... yeah, if you have many foreground panels the time multiplies.

Topaz and NR: I bought the Topaz bundle a while back but am no longer eligible for further updates. Agree it is superior to LS/PS. But I am actually considering to change to DxO's PhotoLab/PureRaw. Had a demo license until last month and tried it on a larger range of my images, just to see what all the hype for DeepPrime was about. And the results were very impressive.

Back to the hotpixels - I will definitely try the dark frames technique. Just wondering how much of a time-saver this is going to be. As far as I remember they need the same exposure length than the normal frames and are temperature dependant. So e.g., if I take my 5 foreground panels and then 3-5 dark frames, I am roughly ending up with the same overall integration time. The time saving effect would only apply if I can reuse the dark frames for the 2nd, 3rd etc. panorama. As there can be up to an hour between two panos, I wonder if reusing is possible or they have to be taken again due to climate changes?

Thanks for the info with DxO for noise reduction. I will have a look into it!

As you need the dark frame only for the foreground, one frame should be enough which will save time. Remember that all the sophisticated processing for the stars - which requires low-noise data - is not necessary for the usually relative dark foreground.

So one set of dark frames (for averaging purposes) should normally last the whole night then? Will try that out next time.

Have you tried to split your long exposure of the foreground into several shorter intervals and stack (adding, not averaging) them in PS or so? Could that reduce the number of hot pixel while keeping the total exposure time nearly constant?

I've only tried that once a long time ago. If I remember correctly, there was less detail in the shadows than with the equivalent same total exposure time for a single image. My unscientific explanation is that some very faint signal might need a minimum exposure time to even 'make it' onto the sensor at all.