Produce a video

From AIRWiki
Jump to: navigation, search

To produce a video about your work is something that should be done, to show the work to firends, at the thesis discussion, to people interested, and to leave some documentation on AIRWiki.

Taking videos

First of all you should take material to be composed in the video. This can be recorded with the tools available in the lab for this (Ask Prof. Bonarini), or with any other tool. Consider the resolution of the material you are about to take: publishing on YouTube does not require a high resolution, but high resolution movies are better than low resolution ones to be used in presentations, like your thesis discussion, and the ones your tutor will probably do (so, please leave him/her a high resolution copy of the final movie!).

BEFORE taking any movie, please WRITE a storyboard with all the scene fully described. This will help you in producing the movie in a short time, nad to think about the movie content.

When you take a movie, please use some stand (available in the lab) to be sure that the camcorder is stable and parallel to ground.

Consider the lighting conditions to be sure that the subject is properly lighted. Sometimes it is hard to take images of a video, and it might be better to save them directly from the source.

Consider that you might also put two or three flows of images on the same image (e.g., to show the behavior odf the robot at the same time of the data from sensors used to drive it.

Assembling videos

Once you have your movie parts, you have to assembly them in a unique, reasonable movie, no longer than 4 minutes. The movie should start with a title page, mentioning the title, the names of the authors, the names of the tutor(s) and AIRLab - Politecnico di Milano. Here is a set of sample pages implemented in GIMP.

Consider the possibility to insert written comments, either in the style you find in the sample page file, or graphically on the video.

Consider the possibility to describe (in spoken English) the content and/or to add music, which should comment or be a background, but not something that is unrelated to the video.

Some tools to assembly materials are available for free both in Windows (Windows Movie Maker), MAC, and Linux (Cinelerra, Kino).

In Linux there is also a really powerful tool, mencoder, available also through command line.

Here are some examples, provided by Bernardo Dal Seno

La codifica avviene in due passate (l'opzione vpass cambia tra le due invocazioni), ma si ottengono buoni risultati già dalla prima a meno che il bitrate è troppo basso. Se lo spazio è importante puoi provare a usare bitrate inferiori (opzione vbitrate). Se vuoi prendere solo una parte del video usa le opzioni 'ss' e 'endpos', se no eliminale.

mencoder -aspect 4:3 -mc 0 -noskip -oac mp3lame -lameopts abr:br=128 -ffourcc DX50 -ovc lavc -lavcopts vpass=1:autoaspect:vbitrate=1000 -ss START_POSITION -endpos SEGMENT_LENGTH -o OUTPUT.AVI INPUT.AVI
mencoder -aspect 4:3 -mc 0 -noskip -oac mp3lame -lameopts abr:br=128 -ffourcc DX50 -ovc lavc -lavcopts vpass=2:autoaspect:vbitrate=1000 -ss START_POSITION -endpos SEGMENT_LENGTH -o OUTPUT.AVI INPUT.AVI


Se vuoi comprimere molto il video, puoi aggiungere un filtro di denoising, che aiuta:

mencoder -vf hqdn3d ...

Puoi anche ridimensionare il video, es.:

mencoder -vf hqdn3d,scale=360:-2 ...

Scala il video a 360 pixel di larghezza (e l'altezza è riscalata in proporizione).

Publishing videos

To publish videos, please send them to Andrea Bonarini who will put them on the AIRLab channel on YouTube.