Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add difficulty skill for Flashlight mod #14217

Merged
merged 20 commits into from
Sep 21, 2021
Merged

Conversation

MBmasher
Copy link
Member

This is a proposal to implement a Flashlight skill to more accurately measure the difficulty of Flashlight plays. Currently the Flashlight bonus is just a multiplier based on object count onto aim pp and this is inaccurate for many reasons. Read more: Writeup link

Values can be checked at https://flashlight.stanr.info/

@frenzibyte frenzibyte changed the title Add a Flashlight skill Add difficulty skill for Flashlight mod Aug 11, 2021
@MBmasher
Copy link
Member Author

Values look fine to me now

@smoogipoo
Copy link
Contributor

@smoogipoo
Copy link
Contributor

smoogipoo commented Sep 10, 2021

Nevermind, the above look to be issues with data.ppy.sh beatmap dumps:

<html>
<head><title>504 Gateway Time-out</title></head>
<body>
<center><h1>504 Gateway Time-out</h1></center>
<hr><center>nginx</center>
</body>
</html>

@smoogipoo smoogipoo requested a review from a team September 13, 2021 06:46
@smoogipoo
Copy link
Contributor

Top 2000 differences (as of 2020/09): https://docs.google.com/spreadsheets/d/1oHefM278mEs9rXgunspPx2Ty_knzQn25fVd1Ox4-bSU/edit?usp=sharing

Looks reasonable to me.

@MBmasher
Copy link
Member Author

Which plays are these? I have a feeling the biggest differences occur in the lower pp ranges (for example Index's play on paraparamax +hdfl which goes up by about 180pp)

@smoogipoo
Copy link
Contributor

smoogipoo commented Sep 14, 2021

Right, that's from the _top dataset from https://data.ppy.sh/ (top 10000 users).

@smoogipoo
Copy link
Contributor

@MBmasher what's the go on this, holding back until the difference in score 3847791924 is accounted for?

@MBmasher
Copy link
Member Author

MBmasher commented Sep 14, 2021

With this system, I'm not sure if it's possible to account for that without harshly nerfing the values of most other maps. I think it's okay to leave it as it is now, especially considering this is a pretty rare case and even so the difficulty of such plays is still pretty high

edit: to be clear, the map that the score 3847791924 was set on has fast doubles which have big spacing but also with a lot of time between them. I'm certain this is the cause of the value going up by that much

@emu1337
Copy link
Member

emu1337 commented Sep 21, 2021

I have some issues with this PR such as not accounting for FL radius changing with combo, hidden, or big circles overlapping the visible area more than small circles. The strain scaling for each object is also done in an unusual way, but it works. The concerns are minor enough and the values are already balanced so we agreed with @MBmasher to keep the PR as it is. It's definitely an improvement from live, I don't see a reason for not approving.

@@ -9,6 +9,7 @@ public class OsuDifficultyAttributes : DifficultyAttributes
{
public double AimStrain { get; set; }
public double SpeedStrain { get; set; }
public double FlashlightRating { get; set; }
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should we really be adding attributes based specifically on mod names? I don't know this seems like a really bad direction.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What are you proposing? It's a special case so I think it's warranted?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It could be VisualRating or ReadingStrain, something like that. It wouldn't make sense to keep adding ratings for each visual mod, as they all affect the same thing and they usually interact in different ways.

For example, the difference between nomod and HD is not the same as the difference between FL and FLHD.
These interactions should be accounted for when calculating the difficulty rating.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

8 participants