I'm just gonna leave this here
Community
Forum
Performance Points feedback and suggestions (Catch the Beat)
posted
Total Posts
767
remove ppAxiaan wrote:
I didn't see the today's post of Amlink about pp
Could a mod just lock this thread until someone is finally going to tweak the pp system I just don't see the possibility of worthwhile discussion in here.
My image was just a joke about Amlink
I don't support his idea about pp
I don't support his idea about pp
remove pp not the game remove ppSeph wrote:
how about we just remove osu in general hmmmmmMMM
neither do the new thousands of people who will never get to know the true game without this farming meta back when i could proudly tell people i played osu now its a joke and anyone i mention it to finds out about how rank works and finds out about how the community is compared to what he or she has heard before and they drop it immedaitly its very selfish for so many users to care so much about something that only has poisned and ruined what i believe ctb was and could still be what a shame that so many people would rather have a number than a chance to save the game and restore it to the second home so many of us spent our time inAxiaan wrote:
My image was just a joke about Amlink
I don't support his idea about pp
I feel having a rank is a really good thing only if the actual pp was accurate or if there was some other way to determine rank or something
I think everyone agrees that converts are much harder than ctb specifics and therefore should have a higher starrating but why exactly is that the case? Obviosly turning a map into a ctb specific does not change anything and will not get a higher starrating in return. The biggest difference in my opinion is that specifics use much more hyperjumps than converts and this is exactly where the problem is with the current difficulty algorhythm.
1. If you look at fullscreenjumps and halfscreenjumps they play exactly the same but the fullscreenjumps give a higher starrating.
2. Edgejumps are way harder to play than hyperjumps but give lower starrating (most of the time)
In order to solve this there is a very simple solution when calculating the difficulty of a map just change every value you'd give a hyperjump to the value you'd give a hyperjumptrigger (a note that is just far away to trigger a hyperdash / slightly higher distance than an edgejump)
In total this would not boost the converts but rather nerf the overwheighted ctb specifics.
Here are 5 difficulties I made (20 notes each) just to show how much starrating every pattern gives:
https://osu.ppy.sh/beatmapsets/739209#fruits/1559894
If you apply what I said for example all the hyperwalk maps would not be 10*+ but rather 3* because they would look like normal streams to the system (exactly how they are played as well)
1 more change I would make. Just let the system ignore notes in the beginning of the map where you do not need to move to catch them.
This would make all those 20-300* maps where you press nothing 0* and also with all the changes I think Der Wald would be an 8-10* map instead of 19* and maybe rankable then.
Just let me know what you think.
1. If you look at fullscreenjumps and halfscreenjumps they play exactly the same but the fullscreenjumps give a higher starrating.
2. Edgejumps are way harder to play than hyperjumps but give lower starrating (most of the time)
In order to solve this there is a very simple solution when calculating the difficulty of a map just change every value you'd give a hyperjump to the value you'd give a hyperjumptrigger (a note that is just far away to trigger a hyperdash / slightly higher distance than an edgejump)
In total this would not boost the converts but rather nerf the overwheighted ctb specifics.
Here are 5 difficulties I made (20 notes each) just to show how much starrating every pattern gives:
https://osu.ppy.sh/beatmapsets/739209#fruits/1559894
If you apply what I said for example all the hyperwalk maps would not be 10*+ but rather 3* because they would look like normal streams to the system (exactly how they are played as well)
1 more change I would make. Just let the system ignore notes in the beginning of the map where you do not need to move to catch them.
This would make all those 20-300* maps where you press nothing 0* and also with all the changes I think Der Wald would be an 8-10* map instead of 19* and maybe rankable then.
Just let me know what you think.
Pretty much everything you said has been said here before, it's just we have no one to make a new system at all so we're left with what we have.
We don't really need a new system tho just add 5-10 lines to the current one once it's added here: https://github.com/ppy/osu/blob/master/osu.Game.Rulesets.Catch/CatchDifficultyCalculator.cs
????????? lmaoGln_Markus wrote:
We don't really need a new system tho just add 5-10 lines to the current one once it's added here: https://github.com/ppy/osu/blob/master/osu.Game.Rulesets.Catch/CatchDifficultyCalculator.cs
they added the std difficulty calculator here: https://github.com/ppy/osu/blob/master/osu.Game.Rulesets.Osu/OsuDifficulty/OsuDifficultyCalculator.cs
So I'm pretty sure they'll add the one for ctb on the other link I sent before once they're finished.
So I'm pretty sure they'll add the one for ctb on the other link I sent before once they're finished.
pp is a poopsystem 
if you can do it you should talk to peppyGln_Markus wrote:
they added the std difficulty calculator here: https://github.com/ppy/osu/blob/master/osu.Game.Rulesets.Osu/OsuDifficulty/OsuDifficultyCalculator.cs
So I'm pretty sure they'll add the one for ctb on the other link I sent before once they're finished.
Seeing the recent tweaks being proposed for other gamemodes got me interested in seeing if any worthwhile, simple changes could be made for catch. Now any changes to the actual star rating calculation (which are sorely needed) are well beyond my ability but I came up with a few possible tweaks to the performance calculator which I think are worth considering.
I haven't been able to gather any sizeable amount of data on the impact of these changes, if anyone can assist with that it would be hugely appreciated. I do however have a couple of spreadsheets showing some relevant information
This spreadsheet includes a number of example beatmaps with current and proposed PP values
This spreadsheet compares current and proposed multipliers
PP values in these spreadsheets will be around ~1-2pp out due to the examples provided using rounded SR values
Scale accuracy penalty with beatmap total objects
The current accuracy penalty works well enough on shorter beatmaps with less overall objects due to individual droplets having a greater impact on overall accuracy. However with more total objects individual droplets have far less effect as accuracy in catch is just the percentage of objects caught (excluding spinners). Despite there typically being more droplets in a longer map, actual player droplet misses do not increase proportionally with the increase of objects in a map to compensate. As a result the current accuracy penalty is far weaker on higher combo beatmaps.
This proposal aims to scale the penalty with the number of total objects in a map (that contribute to accuracy) so the penalty remains relevant on longer maps. This helps to add more variety between player's performances and relatively buffs SS and high accuracy plays on longer maps.
Note that this still doesn't do anything to consider how difficult the droplets are to catch as calculating that would require changes to the actual catch difficulty calculation.
Example graph of penalty at 3000 total objects
Hidden bonus adjustments
Hidden is a bit of a mixed bag for players, it's incredibly difficult for some and can be considered a reading aid for others. Balancing that aspect of Hidden is always going to be tricky, however it's commonly agreed Hidden is easier at high AR values and much harder at lower. This proposal aims mostly at reducing the Hidden bonus at very high AR values(9.5+), whilst consequently buffing low AR values.
Current hidden bonus is 5% at AR10 and above, increasing an additional 7.5% for each AR below 10
Proposed bonus is 2.5% at AR10, increasing an additional 9% for each AR below 10, and reducing to 1% at AR11
This graph shows how the proposed bonus compares. It's virtually the same bonus at ~AR8.2, increased below it and decreased above it.
Approach Rate bonus scaling above AR10
Approach Rate bonus scaling currently gives a linear bonus from AR 9 to 11. This proposal gives an additional bonus to AR values above 10, scaling from a 1.1 modifier at AR10 to a 1.25 modifier at AR11, up from 1.2 because being Sonic is not easy. Going above AR10 gets increasingly difficult with each increment and this change helps reflect that.
Relevant graph. This change is interesting when looked at with the hidden adjustments. Overall PP values for scores with hidden remain similar but plays without see a fair boost bringing the two much closer together.
Thoughts, questions, and feedback are all very welcome
I haven't been able to gather any sizeable amount of data on the impact of these changes, if anyone can assist with that it would be hugely appreciated. I do however have a couple of spreadsheets showing some relevant information
This spreadsheet includes a number of example beatmaps with current and proposed PP values
This spreadsheet compares current and proposed multipliers
PP values in these spreadsheets will be around ~1-2pp out due to the examples provided using rounded SR values
Scale accuracy penalty with beatmap total objects
The current accuracy penalty works well enough on shorter beatmaps with less overall objects due to individual droplets having a greater impact on overall accuracy. However with more total objects individual droplets have far less effect as accuracy in catch is just the percentage of objects caught (excluding spinners). Despite there typically being more droplets in a longer map, actual player droplet misses do not increase proportionally with the increase of objects in a map to compensate. As a result the current accuracy penalty is far weaker on higher combo beatmaps.
This proposal aims to scale the penalty with the number of total objects in a map (that contribute to accuracy) so the penalty remains relevant on longer maps. This helps to add more variety between player's performances and relatively buffs SS and high accuracy plays on longer maps.
Note that this still doesn't do anything to consider how difficult the droplets are to catch as calculating that would require changes to the actual catch difficulty calculation.
Example graph of penalty at 3000 total objects
Hidden bonus adjustments
Hidden is a bit of a mixed bag for players, it's incredibly difficult for some and can be considered a reading aid for others. Balancing that aspect of Hidden is always going to be tricky, however it's commonly agreed Hidden is easier at high AR values and much harder at lower. This proposal aims mostly at reducing the Hidden bonus at very high AR values(9.5+), whilst consequently buffing low AR values.
Current hidden bonus is 5% at AR10 and above, increasing an additional 7.5% for each AR below 10
Proposed bonus is 2.5% at AR10, increasing an additional 9% for each AR below 10, and reducing to 1% at AR11
This graph shows how the proposed bonus compares. It's virtually the same bonus at ~AR8.2, increased below it and decreased above it.
Approach Rate bonus scaling above AR10
Approach Rate bonus scaling currently gives a linear bonus from AR 9 to 11. This proposal gives an additional bonus to AR values above 10, scaling from a 1.1 modifier at AR10 to a 1.25 modifier at AR11, up from 1.2 because being Sonic is not easy. Going above AR10 gets increasingly difficult with each increment and this change helps reflect that.
Relevant graph. This change is interesting when looked at with the hidden adjustments. Overall PP values for scores with hidden remain similar but plays without see a fair boost bringing the two much closer together.
Thoughts, questions, and feedback are all very welcome