I was watching the subs come in for the Crescent Nebula the other night and after awhile I noticed that when there would be a hiccup in the guiding (it was a windy night) the stars in the image would trail but the details in the nebula seemed unchanged. It really took a disturbance that lasted for most of the exposure to degrade the detail in the Nebula to any significant degree. This was while shooting Ha at 60 sec.
Coming from a photography background my thinking was that, if a frame is degraded due to lets say, camera shake, the whole image is ruined. Applying that to Astrophotography told me that if the stars are wonky, the frame is bad. After looking at this closer I realized, that is not the case and here's why:
On a 60 sec. sub-frame I took that night, the noise floor was 491, the core of a med. bright star was 22,714 and the Nebula brightness was 550. Subtracting the noise floor gave me 22,223 for the star core and 108 for the nebula. If I calculate counts per second I get 370 for the star core and 0.98 for the Nebula. If there's say a 5 sec. disturbance in the guiding, 1850 counts from the star core will be displaced while only 4.9 counts will be displaced from the Nebula. 1850 counts that's not where it's supposed to be will be clearly noticeable and significant while the 4.9 counts from the nebula won't be.
This is of course, relative to the sub-exposure length. The shorter the disturbance as a percentage of sub-exposure length, the less the effect will be on the dim details but it will still effect the stars in a visible way. That tells me that stars are a poor indicator of image quality, especially in longer exposures. This is also what I see in stacking. If I go in and filter out the frames with wonky stars I end up with a good star field but at the expense of S/N in my subject. If I just include everything, the subject is more detailed than it was with the culled stack, even though the stars had minor problems which could easily be solved by sub-stacking the stars and general star removal processes.
The lesson I take from this is that for longer integration times (60 sec or more), culling should be kept to a minimum, if at all. For short exposures culling becomes important, giving you some leverage over the seeing and guiding quality. It's certainly an interesting finding that in my case was unexpected and changes the way I approach the work.
I would be curious about what the more experienced imagers here think.
Coming from a photography background my thinking was that, if a frame is degraded due to lets say, camera shake, the whole image is ruined. Applying that to Astrophotography told me that if the stars are wonky, the frame is bad. After looking at this closer I realized, that is not the case and here's why:
On a 60 sec. sub-frame I took that night, the noise floor was 491, the core of a med. bright star was 22,714 and the Nebula brightness was 550. Subtracting the noise floor gave me 22,223 for the star core and 108 for the nebula. If I calculate counts per second I get 370 for the star core and 0.98 for the Nebula. If there's say a 5 sec. disturbance in the guiding, 1850 counts from the star core will be displaced while only 4.9 counts will be displaced from the Nebula. 1850 counts that's not where it's supposed to be will be clearly noticeable and significant while the 4.9 counts from the nebula won't be.
This is of course, relative to the sub-exposure length. The shorter the disturbance as a percentage of sub-exposure length, the less the effect will be on the dim details but it will still effect the stars in a visible way. That tells me that stars are a poor indicator of image quality, especially in longer exposures. This is also what I see in stacking. If I go in and filter out the frames with wonky stars I end up with a good star field but at the expense of S/N in my subject. If I just include everything, the subject is more detailed than it was with the culled stack, even though the stars had minor problems which could easily be solved by sub-stacking the stars and general star removal processes.
The lesson I take from this is that for longer integration times (60 sec or more), culling should be kept to a minimum, if at all. For short exposures culling becomes important, giving you some leverage over the seeing and guiding quality. It's certainly an interesting finding that in my case was unexpected and changes the way I approach the work.
I would be curious about what the more experienced imagers here think.
