Which would use less memory?

[Simflare]

Hey everyone

I had a question, and it's to do with optimization and scaling at runtime.

Say I had a block 128x128 pixels, and it was a ground tile for jumping on. I would like to know which of the two below methods would use the least memory (thinking of a Canabalt style game):

creating a thin detector on the left and top edge (around 5 pixels)
or
creating a thin detector that goes all the way along the left edge , and scaling it across the whole 100 x 100 pixels at run time

I'm aware that it's better to use power of 2 images, but not sure how that affects things when you start scaling objects at runtime.

Thanks
Simflare

[Bipolar_Games]

Hmm, why not just make the object itself a detector? It's a large active object I'm assuming, why not just put it in a Group with all other objects your character can collide with, and test for collision with that Group?

[Simflare]

In some cases (depending on how the game is designed), yes, but this is a hypothetical question, and I want to understand what uses more memory and what is more efficient. It's good to understand to design better games, that's why this is a hypothetical question.

I need someone to tell me which would be faster on the memory, since I'm doing games for Android and optimization is key.

[DaveC]

I'm pretty sure scaling will use less memory, at least it will use less memory for the texture but I'm not sure of the impact on the performance overall (if there are any tradeoffs or a point of diminishing returns)

8x8 is the smallest texture block (so if you had a 2x2 block that doesn't matter, it will still count as 8x8).

So to sum up, if you had a 8x8 block and scaled it up 16 times (to make a 128x128 block) vs a 128x128 active object, the 8x8 will use less memory and should in theory take less of a hit on performance.

of course, if you have a 128x128 block you may as well use that and not have a detector, because otherwise you'll have the block+the detector.

[Simflare]

Awesome, thanks Dave very helpful