Hi everyone, I would like to ask you this question by taking the "SUIKINKUTSU feat. Uehara Hiromi
TOKYO SKA PARADISE ORCHESTRA" by Evening as an example:
https://osu.ppy.sh/beatmapsets/327318#mania/2134623

As you know/can see, the map has a lot of Timing Points, or as in chat some call "Variable BPM" normalized with SVs to stabilize at 170 BPM.
I'm not new in the mapping scene, I would like to map a song that has time inconsistencies (probably because it is also based on complex organisms like an orchestra) but, my question is:
How do I choose the appropriate BPM for very short Timing Points between them?
In Evening's map there are some Timing Points with "51.027 BPM" followed by another at "47.51 BPM" so strange to me.

The song that i would like to map is: https://www.youtube.com/watch?v=pEZIYGN5HIo
There is no precise time, it varies from 60 to 64 BPM, I have divided into 9 main parts with similar BPM but I think it needs timing for every single piano note, something like that.

Anyone know how to do it or how Evening did it? I cannot write to him because he has blocked communications with those who are not on his friends list.
By the way a map like this with a lot of timing points, can it be ranked?

Thanks for you time ♥

On variable bpm maps, bpm is rather a very theoretical value.
The goal of placing another timing section is snapping a sound you want to map that would be unsnapped without that timing section. That means that the primary goal with variable bpm timing is getting the offsets right.

A secondary goal and that is why you actually still change the bpm everytime, is lining up the timing sections so that a new timing section always starts on a snap of a previous section.
This has various benefits:
1. Gameplay with the NC mod gives you additional sounds on every white tick. If you have white ticks randomly scattered around, you will get a lot of nonsensical noise that makes the map unplayable with NC.
2. The speed of sliders is consistent with what the song does at present. If you only do the offsets, your sliderspeed will always be the same which you would then have to adjust with SV anyway. I'm not a mania mapper so I'm not sure if this is relevant.
3. Scrolling to a specific snap is more comfortable. If your bpm sections don't line up, scrolling into a new timing sections will put you on an unsnapped time and you will have to scroll farther into the new section and then back to get on snaps at the start of a new section. This can be quite annoying when mapping sections where small offsets adjustments need to be made very frequently.
4. When mapping multi-diffspreads you will often need to map LNs and sliders over multiple timing sections for the lower difficulty. It is feasible to automate the generation of SV sections in 3rd party tools that normalise the SV so that the slider/LN-end snap despite going over multiple sections. If you take 2. into account you should be able to see why an automated solution like that can't work properly for a timing that only determines offsets and does not line them up with bpm too. Basically, it would cause conflicts because you can't automate the SV fix easily if you need to manually adjust the SV too.

If you haven't taken a look yet, check out the Variable bpm timing tutorial by Charles445 where the process of timing variable bpm is explained with a hands-on example that ended up ranked.

Generally speaking, lining up offsets is not a strict must. Especially in low density sections without sliders/LNs like the intro of the map you linked it might very well be overkill.

/edit: After a further look...
I think in mania the combination of bpm+SV actually controls the scrollspeed relative to the user's scrollspeed setting so even if you are mapping a singlediff, you would still need to correct the scrollspeedchanges caused by the bpm changes by adding an according SV.
A script for automatically adding these without lining up timing sections would be more complicated and bloat the beatmap (I think you would need 1 green line for every single snap you use).