My personal preference on estimations:
1. Estimate in ranges rather than strict values (for example, "8-12h" instead of "10h")
2. Estimate as a team. There is a great, free tool to do it swiftly: Plan IT Poker and wrote more about it here
3. Add confidence level to each estimation (low, mid, high). It might help identify risks, and unknown areas
4. Avoid tasks bigger than 1 sprint, the smaller, the better.
It's also worth mentioning that we shouldn't focus on estimations too much. If possible, I prefer to run a project without any logging and focus on the goal.
Work can be done on different levels and the truth is, the more time you have the more you can achieve if you're focused on the goal.
It's similar to below painting examples. Painting on which you had the most time, usually looks the best. YT or Short