Retopology timelapse of Master Tigress sculpt I did some time ago. The video was sped up 5 times but still has some cuts and pauses (sometimes I just forgot to record stuff :P).
Retopology is kind of a meditative process, and according to windows it took me 20 hours to finish it, though due to some cuts, this number is a little bit bigger.

    To be honest, during the process, I always had a thought that I was doing something wrong, and indeed I was. While editing, I noticed that I could do things differently and don't waste time in some areas. This video is a bit embarrassing to me, but I'm glad that I have made that video. I came to realization about certain things and how I can negate or speed up the process in the future.
I would call this video "How not to do retopology", I doubt I'll ever do a similar video anytime soon, especially considering that the recordings occupied 50gb of my hard drive.

* First of all, following curvature of muscles, imho, is a bit overdone, especially with the limbs. Placing a simple grid structure and differentiating from it would significantly speed up the process if I didn't tried to establish muscle curves first. At the end this excessive topology creates some shading pinches on subdivided mesh and it will be difficult to smooth it out during multirtes sculpting. Besides, if vertex placed in cravises of high-poly mesh it might define shape of it even with simple grid topology and clean face loops. 

* Starting from the face would gave me the idea of polygon density I was going for. I still managed to connect most of the parts without much trouble, but it's still a good practice to start with a topology that allows you to easily create supportive loops and then finish up with details later, even if with the triangles.

     Default retopology tools in blender kidna sucks. It's probably great for simple forms, but with increased complexity comes greater problems that you have to solve with primitive tools. But that's just my problem, apparently I was taking this process it with the wrong approach. 
Using subdivided mesh or external addons definitely would  speed up the process, but I really like retopo vision feature that was added to the Blender year ago, it's really neat, I like it, so I'll be thinking how to make this workflow work.

    At the end, the base topology is ready, some changes will be done at rigging stage, but still I have to model other features. I guess I'll finish this project by the end of February, I'm still not sure how hard it would be to develop tools for rigging.

Final model screenshots: