I'm looking for / working on an algorithm that must be very small and simple, with no extra libraries and stuff that looks suspicious to a user using the code. You can see what I mean by reading the below. The user just keeps the dead simple code that he can even understand, and remembers his passcode, and the files are (reversibly) encrypted, and essentially uncrackable.
Here's my method, but is it really solid?:
Say the file (be it a video or text file) is this (read as bytes) > 'we walked through the', to be encrypted. And the passcode is "cloudteethbook". The first byte of the file 'w' looks at the first letter of the passcode 'c', and increments its byte position up by ex. 3 if 'c' is the 3rd of the 256 possible chars. Then the second letter of the file 'e' looks at the second letter of the passcode 'l', and does it again. Once the passcode is used up, it repeats again as it goes deeper through the file.
For a real/worked example (assuming a is 1 and z is 26, while 'space' is 27):
file = 'walked through'
passcode = 'acb'
output = 'xdnlhfawjsrwhk'
Explanation: The W is raised 1 letter up through the alphabet, because of '[a]cb', which is 1 to raise it by, which gives x. And so on. The K in WALKED is also by the 'a', because we restarted the passcode's pass.