What does an Encoding Program actually do?

[TheLostTimberwolf]

When we encode a source to say vcd there are a variety of programs that we use to achive this result. Everyone knows the outside of encoding, ie whoich programs to use and how to use them etc but what happens on the inside?

What does the video encoder lets say TMPGEnc actually do to the incoming source file, so that it can output the desired vcd file?

I got wondering about this when I realised that bad or un-decodable frames where to blame for many A/V sync issues and other problems. I know that most digital video files are compressed in some way, does the encoder mearly decompress it change it and put it out the other end as an mpeg1 file? or is it a bit deeper than this?

just a curious mind lol catch you around and thanks for your time

TLTw

[Greg12]

it analyzes the video looking for things it can destory. it destorys all the light the human eye cannot see, and then it compresses certain images into Intra frames. It than compares the next frame to the I frame and eliminates anything that is the same.


it's all math.

[TheLostTimberwolf]

sounds like a visual equivelant of MP3, cheers for the post

[boi]

mpeg1 first takes the jpeg of each frame. jpeg compression changes the colour space from RGB to YUV (luminance and chrominance) then discards some of this information. the data is then transformed from the time domain to the frequncy domain with the discrete cosine tranform (aka fast fourier transform). the data is also truncated (rounded off) and quantitised.
as well as this, interframe redundancy is exploited ie, imagine a video of a person walking along infront of a blue sky - a lot of the data representing the sky will be the same since it's all blue. so the mpeg compression algorithm subtracts a frame from the previous frame and only stores the difference. this storing of only the difference saves heaps of space.
mpeg2 and 4 use a few more tricks

[scottb721]

What is the purpose of the I, P and B pictures?