Print

[luke]

OK, OK, you guys probably knew about this about 10 years ago already but I thought I might share cos I just found this out today and was well excited about it  .

In the sample editor screen I had done:

file->new->lenghth64->generators->squarewave->4 cycles.

this was all in 8bit

so I had a normal square wave, length 64, with four cycles in 8bit.....THEN, I decided to change to 16bit, so i pressed the little button....it said "do you want to covert sample data?", I said NO WAY!!!

after this there was still the normal square wave shape with just a wierd little bit that had changed...i thought...nah....so I changed it back to 8bit...it said "do you want to convert sample data", I said NO!!!

suddenly, my square wave had become some extra cool wave that I did not know milkytracker was capable of generating...pluss it sounds REALLY REALLY COOL....I think the process can be repeated to further mess with the wave...ITS CRAZY!!!!! Im happy about it....new waves now possible, the way it messes up the waves is really even...its GRRRRREEEAAAT!!!!!

did anyone know about this?

why does this happen?

(sorry for long post, a little too excited perhaps ...hah!)

[luke]

So did you guys know about this already?

got any idea why that happens, its really useful, makes realy good chip sounds.

[raina]

My MilkyTracker crashes when I follow your instructions and click the 16-bit radiobutton.

[luke]

wierd, i get a square wave but with loads of smaller triangle waveforms within the peaks of the square wave...

[raina]

Your observations are the ones that should be noted. Who knows what version I'm running

[Saga Musix]

Well, seems pretty normal to me. If you convert from 16-bit to 8-bit without converting the sample data itself, the tracker will interpret your 16-bit data as 8-bit data, which is obviously wrong and will bring up strange results. I don't know if that's what Milky does since I don't use it, but it would sounds logical to me.

[8ch]

yep, this worked well in ft2. take a 16bit sample, convert it to 8bit (say YES), then back to 16bit (say NO), then back to 8bit (say YES) etc... this will reduce your sample size dramatically. if your notes happen to get out of range (> b7), this method, in combination with transpose (either change the base note of your instrument or traditionally in the pattern), can be used to get them back in range..

[Ev]

I think he's referring to this (I just tried it).

http://img269.imageshack.us/img269/4052/79148762.png

[8ch]

..just discovered that if you do the "conversion trick" over and over again and your sample becomes too tiny it'll just freezes milky (endless hourglass). -on debian with j.a.c.k.- use the resampler instead =P

[luke]

yeah, the picture from Ev is exactly it...