I'm planning to photograph flowing river water with long exposures for a blurry effect. I currently have a 4-stop ND filter which may be good for shooting wide open in sunlight, but is probably not enough for photographing flowing water in sunlight. I have found that both 9-stop and 10-stop ND filters are available where I live (there are probably other options but not all of them are shipped to my country).
According to the sunny 16 rule, ISO 100 and f/16 requires 1/100 s exposure. If stopping down to f/22, 1/50 s exposure is needed. From that, one can see that 9-stop ND filter allows 10 s exposure whereas 10-stop ND filter allows 20 s exposure. If shooting at wider aperture like f/8, 9-stop filter would give 2.5 s exposure whereas 10-stop ND filter would allow 5 s exposure. The wider aperture would allow lower diffraction.
It appears to me that 10 stops would be better in direct sunlight, but is that so? Are there any benefits in a 9-stop ND filter such as ability to use autofocus?
I know that autofocus on this camera works barely in conditions where optimal parameters are ISO 6400, f/2.8 and 1/40 s exposure. Sunlight (ISO 100, f/16 and 1/100 s exposure) is 5200 times brighter than the low-light conditions where I have tested autofocus. Thus, I assume that autofocus should work in direct sunlight even given a 10-stop filter. However, lighting is not direct sunlight all the time. In overcast conditions, f/8 is needed instead of f/16, so light is one fourth of direct sunlight, or 1300 times brighter than the low-light conditions where I have tested autofocus. Thus, I assume autofocus should barely work even in overcast conditions with a 10-stop ND filter. However, the margin isn't huge, so I may run into AF problems.
Also, at this time of the year, my shortest exposure has been ISO-100, 1/500 s, f/2.8 (equivalent to ISO-100, 1/16 s, f/16, i.e. about one sixth of direct sunlight), but when the seasons change the sunlight will become more direct. Anyway, 10-stop ND filter in this light level is not guaranteed to work with AF as the light level is only 800 times the one where I tested autofocus to work.
My worries with 9-stop filter are that if shooting in sunlight, I might not get long enough exposure, and also it might be necessary to stop down to f/22, causing too much diffraction. However, the 9-stop filter could perhaps be combined with a 4-stop filter in demanding (very bright sunny) conditions, giving a 13-stop filter which is surely enough for anything except autofocus. Also, I can take multiple exposures with a 9-stop filter and combined them later.
My worries with 10-stop filter are that if not shooting in direct sunlight, autofocus might not work with a 10-stop filter but might work with a 9-stop filter. Of course, I can always focus without the filter and then put the filter on, but that might be inconvenient. Also, a 10-stop filter has a large gap when compared to 4-stop filter I already have, so if photographing near sunset time, the conditions may change to be such that 10 stops is too much but 4 stops is too little. Additionally, 10+4 or 14 stops is probably too much even in very bright sunny conditions.
So, which filter should I choose to complement a 4-stop ND filter? The 9-stop or the 10-stop one?
The minimum ISO of the crop sensor camera is 100, and the maximum exposure time is 30 seconds, although I can always take multiple exposures and average them during post-processing. I have the ability to use a remote shutter control for the camera, so one exposure doesn't automatically mean one shake of the tripod. If it matters, the lenses are Canon 24mm f/2.8 - f/22 and 50mm f/1.8 - f/22 and perhaps 85mm f/1.8 - f/22. So, there's at least 62x dynamic range in the aperture control; the difference between 4-stop and 9-stop would be 32x and the difference between 4-stop and 10-stop would be 64x.
Related, some general advice about ND filters: What are neutral density filters and how do I use them to create long exposures in daylight? ...where the author of the accepted answer is said to use 1000x or 10-stop filter.