In LittleBigPlanet, the best way to utilize the Memorizer is to compress your data. This means taking your data, and shrinking it to the smallest data size you can manage with the same data present.

In this thread, I am counting on you all already understanding compression, general logic, and signal and data storage in LBP, because this is all to circumvent the issues of rounding and only having negative exponents to your disposal.

The first issue, error, cannot really be solved, as it rounds you bits down. E.G. if you had 5006 rounding to 5000 constantly. No way to circumvent the loss of these bits. This seems to occur when multiplying small mantissa against a large negative exponent. A base 10 example could be 5006 * 10^(-8) where it would be lowered floor 5000 * 10^(-8). This may be to me multiplying in an incorrect order, or passing through some sort of arithmetic means that rounds, meaning it could be error of my own making, although, I have checked for this and had not found that to be perticularly (probably) the case, but correct me if I'm wrong.

The second issue comes where you can no longer have and use signals above 100%. Before, when you transfered the exponent values alongside the mantissa, you would multiply them, but the mantissa can now replicate higher negative exponent values. E.G. (.5) and (2^-1). The way that I have best found to circumvent this is in the loss of 2 bits. By forcing the largest and smallest mantissa staying on (.100...01) then the value stays between .5 and 1. This ensures that mantissa cannot replicate an exponent value when multiplied back in. (This is due to log2(mantissa) always being above -1, thus not changing the exponent values).


To summarize: To get the most out of compression in LBPV, you have to abandon the exponent bits that round your mantissa values, because by rounding the mantissa, it is impossible to recover those bits, so it is best to abandon them (unless it's possible to keep those bits? Am I missing something?), and to conserve magnitude in the exponent values, 2 bits must be sacrificed. 6 bit losses due to the issues in LBPV.

Here are my pre-made compression tools

Start a discussion if you want, I'll answer any questions (maybe you can answer some of mine too  )

Actually I don't understand anything. What does this mean? Can you explain it to me?

(lines 4-5: "In this thread, I am counting on you all already understanding compression, general logic, and signal and data storage in LBP, because this is all to circumvent the issues of rounding and only having negative exponents to your disposal.")

But, I'll try to explain it in layman terms for the time being, although if you don't have prior knowledge of LBP's use of compression, then you won't be able to understand this to it's fullest extent.

LBP has a certain amount of accuracy with each signal. With this accuracy, you can only have so many numbers, and this post was meant to target the problem that resides in the vita version -the issue of predetermined loss, and how to preserve as much of it as you can. EG if LBP forces you to lose 5 bits, I hoped to explain how to lose 5, not 6 or more. Sorry if this doesn't make sense, it's the most shallow I can go with still being on the topic at hand.

You mean LBP Vita does have a limit, when it comes to logic parts and their signal strength?

You have played my level BWZ - Oh Russia. I know how to make complex things and I can get them work. But I don't understand your text to 100%. Maybe 50%. I'm sorry.

Ah, I see, let me try to explain. In LBP, everything runs on a signal system. There's on and off, + and -, mantissa (precision), and exponents (range). What I'm doing is explaining how to utilize the majority of the last two. If you are interested in these things, a good segue would be to try and find out about feedback loops, they're a great intro for LBP signals.

It's about limits and beating the game?