Android [solved] will smaller collisionbokses save framerate?

[Uberpink]

i have a house instance and its sprite have a simple collision all over, howewer i only need some piksels tall collision at top of the house

Question,, will smaller collision be better for the framerate? in other words does a collision boks cost more for the processing etc the bigger it is?

 

[icuurd12b42]

non precise collision, using the box, is simply rectangle intersection math. so the size matters not... \

 

[Uberpink]

thanks for answer how about precise collision, is that very heavy? wonder if its some kind of possability to measure it heh.. how much it takes compared to non-precice coll,,, maybe theres a fps counter command or something eksisting

 

[FrostyCat]

thanks for answer how about precise collision, is that very heavy? wonder if its some kind of possability to measure it heh.. how much it takes compared to non-precice coll,,, maybe theres a fps counter command or something eksisting

It's absolutely heavy, and the difference in performance is well-known and documented. The gap is so blatant that you can tell from the logic alone how much heavier it is compared to simple rectangular collisions, no need for a profiler.

 

[icuurd12b42]

With precise collision the bounding box is used as a collision precheck. so the performance issues only apply when the bounding boxes of colliding instances overlap where pixel to pixel check is performed...

 

[Uberpink]

With precise collision the bounding box is used as a collision precheck. so the performance issues only apply when the bounding boxes of colliding instances overlap where pixel to pixel check is performed...

aha but that was kind of smart.. so it aint heavy all the time when not colliding

 

[icuurd12b42]

Yeah the blanket "More Efficient" statement is really starting to piss me off... that and the misconception about collision shape in a yoyo approved tut no less...

Read my comment on the video...

 

[kupo15]

With precise collision the bounding box is used as a collision precheck. so the performance issues only apply when the bounding boxes of colliding instances overlap where pixel to pixel check is performed...

Does precise collision break out of the checking once it detects a collision? So if it detects a collision at the first pixel to pixel check, how much slower is it really? The bounding box precheck should help a lot in most cases to producing more first check detection also I would think

 

[icuurd12b42]

Does precise collision break out of the checking once it detects a collision? So if it detects a collision at the first pixel to pixel check, how much slower is it really? The bounding box precheck should help a lot in most cases to producing more first check detection also I would think

I would assume it does break out on the first hit. would not make sense otherwise.

There is the memory management to consider... does it load the 2 images involved into a separate memory spaces to do the comparison, or does it preload all the images in memory at startup and references the memory addresses directly from that... I don't know... I just know that if nothing is touching, there is no difference in cost with rectangle based collision

 

[kupo15]

I would assume it does break out on the first hit. would not make sense otherwise.

That's my thinking. I wish we had control over which direction it did its checks. Like if I'm going to be detecting a landing, it would make sense of the player's collision box to start checking from the lower right and work its way to the upper left.

 

[icuurd12b42]

it's possible it only checks inside the intersecting region, that's the way I implemented my gml version in gm6