The impact of sub exposure time is very low compared to a) going to a darker place and b) adding more over all exposure time. If you listen to Dr. Glover's video that is the message. The case he makes up is somewhat academic: if you have a given over all time there is some optimal exposure time that is just a little better than another one. My real life situation is this: if the image is poor after one night I put up the scope again and add another night. And/or try again when there is less moon. I was asked the fine detail question recently. I just copy my answer below. Hope it helps to put image quality in a wider context.
Resolution has a lot of limiting factors.
1) the physical limits given by aperture. Light is this famous particle wave dualism. In it's wave nature it is diffracted whenever there is an obstacle in the way like waves on a pond are when there is something in the water. The circular opening of the scope causes diffraction and limits the largest thing you can see. This depends on the wavelength. It is worst for red and best for blue. What you see in the technical data is for green unless otherwise stated.
Dawes and Rayleigh are different ratings, basically two stars are fully divided (i.e. it is black between) or just can be distinguished (i.e. two maxima).
2) Seeing. Air is not vacuum and it has got an optical density that varies with it's physical density which varies by height and temperature. Here the red light is slightly less affected than the blue light. Seeing depends mostly on weather, mainly the jet stream high in the atmosphere but on near ground effects like hot areas of concrete and cold areas of water and wind going up or down a mountain. As far as optical resolution is concerned this is often the limit for ambitious amateurs. Seeing can change within seconds that is why planetary has got a higher resolution when doing lucky imaging.
3) Optical quality of the lenses and mirrors. If a mirror is not as it supposed to be the resolution is limited. This happened to the Hubble Space Telescope and it needed a corrector. A telescope is called "diffraction limited" when the mirror or lens offers more resolution than the diffraction.
4) Chromatic aberration, lateral and longitudinal. The colors to not come to focus in the same place in refractors and so a detail in one color is not resolved as well as in another color. Mirrors do not suffer from it and good triplets also work well.
5) Imaging scale. This is how many arcsecs you have on one pixel. If you use a 300mm focal lenght telescope and 6 micron pixels seeing is not your problem. Drizzle can shift the border a bit further but in general you should have a good sampling. Nyquist and Shannon found you need twice the sampling rate than the highest frequency in the signal. In spatial sampling where you have rows, columns and diagonals 3 times higher is better. So, if your seeing and diffraction limit allows for 2 arcsecs you want 0.66 arcsec/pixel in a monochrome camera, even a lower value in a color camera.
6) The Bayer pattern if you do OSC. As only 2 out of 4 pixels are green and only 1 out of 4 is blue or red a color sensor resolves less than a monochrome sensor. There often is a spatial low pass (blur in other words) on color filters to avoid that a point light source like a star can illuminate just one physical pixel. The light is intentionally scattered to a few pixels. Some cameras have it, some don't.
7) Noise. Noise eats up the fine structure first. Even if your seeing is perfect, your scope can resolve what you want to image, the imaging scale is well chosen and you have no trouble with CA fine detail is just invisible because the signal to noise ratio (SNR) was too low. You need very long over all time to push the noise level down and find the fine detail. Denoise strategies do not work here. Noise reduction is always sacrificing resolution.
8) Tracking. For planetary the exposure times are so short that tracking is not important. For long DSO images it is pretty obvious that the image shows motion blur if the tracking does not follow with the accuracy you want to resolve overt the whole time of one exposure. Auto guiding comes into play here as well as your mounts capabilities. There are two reasons why your stars drift: polar alignment error and periodic error. Recently I posted this unguided experiment:
It happened to have a drift of 0.75 pixels per minute. One minute shots were just ok but not good. (I have near 2 arcsec FWHM in perfect moments, 3 on average. Adding 0.75 is quite a loss of resolution.)
Without guiding you need perfect polar alignment and the periodic error will kill you (unless you go for a $8k+ mount with precision encoders like 10 micron). When guiding the guider takes care of that. In theory. If your mount does not react on a command because of backlash, sticktion or poor balancing or if it overshoots guiding may not succeed or make it even worse. That is why just guiding a poor mount does not lead to good resolution. It is primarily a mechanical system that need to be well engineered, well crafted and well adjusted. All guide program parameter cannot heal what is wrong in the mechanics. If the mount reacts soon and moves smooth and does not overshoot you can guide with low aggression settings because over a period of many seconds only fractions of a pixels need to be corrected. If you have a poor mount that just comes with bumpy or loose bearings huge errors occur within second and you need to use short guide exposures. If that is combined with poor seeing you are chasing the seeing.
9) Flex in the telescope. The telescope is long and heavy. It rotates during the night and gravity pulls all parts down. In my RC6 the rear cell tilted in relation to the tube and the secondary mirror because the tube is weak and the rear cell is a (quote) "brain dead" design. You can have the best guide scope attached bomb save to the tube and fail if the inner parts move. This is one reason to use an OAG. It is behind the scope and can correct the internal flex as well. The second reason to go for OAG is saving weight especially if you are at very long focal lengths and need a long guide scope as well.
10) perhaps, not sure if I want to count it here. You have to do deconvolution in the post processing to correct for the unavoidable error like diffraction. This only works well if you have low noise data, good tracking and a reasonable resolution. In my images it makes about half an arcsecond.