virtualdub cropping/compression error?

[popoxDDD]

Right now I am trying to crop some old vids (avi files from hypercam recorded with xvid mpeg-4 codec)
but I keep getting "Cannot start video compression. An unknown error occured(may be corrupt data). Error code -100"


but the strange thing is that...I only get this like half of the time. like the first time I crop a particular vid, it works fine but the second time I get this error message.
or sometimes I get the error message the first time and then it works on a second try...regardless its very annoying.

I am using null transform if that makes a difference....I am doing the exact same things each time so I don't know whats going on...or at least I think I am doing it exactly the same

[Baldrick]

What is the output resolution/video size? Try use a resoultion at least even dividable by 2.


Or try crop with avidemux.

[jagabo]

I agree. That error message is almost always an indication that you have left an illegal frame size for the compression codec. Always leave a frame size that is at least an integer multiple of 4. Multiples of 8 and 16 are even better.

[popoxDDD]

right now its at 480 by 424 (before cropping)
but when I use null transform, there really isnt a way to do it by pixels.. :/ is there?

edit nvm it works now thanks guys!...seems like a silly rule though

[jagabo]

seems like a silly rule though

Not really. Internally, MPEG family codecs usually work in YV12 where the grayscale portion of the picture is at the full resolution, but the color components are both half size (in each dimension). So each frame of your 480x424 video are composed of a 480x424 grayscale image and two 240x212 color images. Hence the requirement that the resolution be at least a multiple of 2 (otherwise there would have to be half pixels). With interlaced video the frame consists of two half images. So a 480x424 interlaced frame is actually two 480x212 "fields". Each of those fields consists of a 480x212 grayscale image and two 240x106 color images. Hence the requirement for frame sizes to be multiples of 4. Finally the picture is broken into 16x16 blocks and further into 8x8 blocks. Hence the codecs are most efficient with multiples of 8 or 16.