News

01/03/04 BWChart 1.02V : somes changes in the APM chart

Version 1.02V has had a big internal change in how I store and process the APM information. The initial intent was mainly to speed up the display of the APM curve and other resources. But it also had a interesting effect on how the APM comes out, as you can see on the drawing below.

Basically, there are two major changes in how I compute this chart:

1 - the time window that I use to compute the local APM is now centered on the point where I compute the APM
2 - the time window width has been reduced

As a result of 1, the APM chart is now much more accurate regarding time. You can see it on the chart below on lines 1 and 3. Line 1 shows the beginning of the game. In version 1.02U, the curve starts from 0 and is then going up rapidly, which does not reflect reality at all, because when the game starts, players start clicking and hitting keys right away. The 1.02V charts shows the right APM right from the beginning.

As a result of 2, the APM chart is now much more "dynamic", which means it follows the acceleration and deceleration of the keyboard/mouse activity much closer. Line 2 on the chart below shows the difference. At this moment of the game, the player with the orange APM is releasing the pressure very sensibly, but you don't see it on the 1.0U chart because the APM curve is artificially flattened. Version 1.02V on the other hand perfectly shows the big slow down and then the big acceleration that follows.

Of course, a direct consequence of those changes is also that all local maximums will be modified in time and value (as shown on line 3). The APM deviation will also be much more accurate. Make sure to REBUILD YOUR REPLAYS DATABASE in order to see the new values for APM deviation. The average APM itself is not modified.

I think those changes make the APM chart more accurate and more useful. For an average player like myself, it suddenly made obvious how inconsistent my APM was. All my big "slow down" were hidden by the flattening of the curve and now they appear clearly. I went thru some recent replays of mine trying to spot big and long drops in my APM chart. Then I proceeded to watch the replay at that exact time in order to figure out why I was suddenly so inactive. Was I distracted by something? Or was I just simply "watching" the game?

Another interesting thing I noticed is that, although the chart was now showing the big drops in APM, I couldn't see any big drop in the big gosu games. Which means those guys have insanely high APM, but they keep it high throughout the whole game. See the example below, a 16 minutes game between ChRh (302 apm) and Boxer (245 apm). Both player alsmot never go down below 200 apm...

Other changes in version 1.02V

-added units count on map view
-added options to customize the colors of the charts and the text on them
-added background bitmap on charts screen (with display options to center, tile or stretch bitmap)

-added "clean missing" button, to remove from database all replays that have been deleted manually (.rep file is missing)
-replaced radio buttons with a vertical tab to select replay, player or map list (to gain some vertical space)
-reorganized the options page in multiple sub-pages
-fixed bug on map filter. The map name was not saved on leaving bwchart.
-updated the initial aka list for players and maps (thx to distant_voice)
-fixed bug when adding a big list of akas in the "edit akas" dialog

I'll let you see for yourself. Have fun :)

All news...