It's a challenge, but in a sense it can be extremely rewarding to be able to get good images from a terrible, terrible observation spot! A couple of thoughts:
Obviously you'll have a much easier time imaging with narrowband filters, the narrower the better (3nm might be worthwhile). This has the advantage of making a lot more of your subs usable, as it reduces considerably the issues described below.
If you want to stick to visible spectrum, filters can be nice but they only help so far (the difference, at the end of the day, is not that dramatic). The real challenges are going to be:
- Integration time :
- You're probably looking at 20-30 hours of light to get something good. I've done some project with 40-60 hours, that don't look anything as good as what other people get with less than 10 hours of integration.
- Even with a ton of integration time, the results are not going to look great without some good processing
- You'll realise you're throwing out a large part of all the subs you're shooting (see below)
- Light Pollution Gradients :
- Imaging for a lot of nights to get enough hours of light means that you'll be shooting at the same target from a lot of different angles (high in the sky, closer to the horizon).
- You'll want to take care of the gradients individually rather than putting together all of your subs in one big integration. This can be easier or harder depending on the software you're using to stack.
- I've had situations where integrating each night separately was good, but others where splitting images by where the object was in the sky (e.g. High vs lower down) and stacking multiple nights at the same angles together actually had better results
- Unfortunately, the gradients are not constant in the city, and they might change quite a bit depending on the activity of the city (e.g. on friday/saturday nights vs rest of the week)
- You'll have a lot of bad images among your subs, make sure you check some quality measure to identify "really bad" images that will hamper your stack more than help.
- When you shoot lower on the horizon, the type of street-lights (and building lights) in your city will start to have importance, especially if they're different in different areas in the city. That might mean getting images that are unusable as soon as the object starts dipping below a certain level.
- Processing time and space :
- With big light pollution, you'll end up having to do relatively short integrations (max 1-2 minutes), which means A LOT of images to integrate. This means a lot of processing time, and when you make an error it costs a lot of time
- Any project will destroy your HDD space very quickly
- Splitting your stacking into multiple sessions can help a lot (see comment above about grouping by gradient vs grouping by day)
- As a rule of thumb, if you're stacking at least 20-30 images per session, it's better to make multiple stacks than a single big one. It makes almost no difference in quality stacking 20x sub-stacks of 20 images vs stacking a single stack with 400x images, but it is much much faster to do 20x sub-stacks than a single uber-stack.
