When doing an interview there need to be at least three people involved. The person being interviewed, you, the interviewer and someone to work the camcorder. It could be argued - I would so argue - that an extra person to handle the sound is a great benefit but this is a policy of perfection.

Interviewing for video is a skill that can be acquired with practice. The key to successful interviewing is research, research and yet more research.

Before you interview a subject you need to know as much as possible about the person you are interviewing.

And you need to read everything that anyone has ever written about the subject.

This is what the Internet is for. Typically it contains everything written in recent times.

The more information you can get, the more research you do, the smoother the interview will go.

Having said that it is vital that you, the interviewer, both asks questions and knows when to shut up.

The problem is that often you will find yourself knowing more about a subject than the person you are interviewing, and the temptation then is to show off your knowledge.

This is fatal.

The viewer is not interested in your views. It is the person being interviewed ? the interviewee ? who should be the center of attention. And before you ask, yes, this is a major problem for me. I cannot keep my big mouth shut.

Be prepared for interviews that go smoothly and those that get a bit ragged. Three examples.

I had an hour interview with Bill Gates in Sydney for Australian television. My questions were well prepared. He would listen to the question, stay quiet for a moment and then give a cogent, grammatical answer. Amazing. As an interview it went like a dream. On the other hand, I got one very wrong. The interview with the late Tony Hancock, perhaps the greatest British comedian of his day, took place in the Sebel Town House in Sydney. It started as a shambles for I had not done enough homework. It evened out after a while and in the end worked reasonably well. As it happens it was the last interview given by Tony Hancock.

Then I did a series of interviews with members of my family. I stayed out of shot and just let them ramble on what they thought about their siblings. It was electrifying stuff. If you were a member of that family.

Write all of your questions down and create supplementary questions in case an answer, a good answer, is not forthcoming. Avoid questions that invite the single word answer 'yes' or 'no'.

If you ask 'Are you in favor of premarital sex?' you will probably get a single word reply, which is not the idea at all.

Phrase your questions so that they lead the person being interviewed into expanding their views. 'Your book suggests that you are against premarital intercourse. What are your views on this?' is much more likely to elicit a full and frank comment than the first question.

To avoid a 'yes' or 'no' answer use the tried and true journalist technique of asking who, what, why, how and when questions. None of these can be answered with a straight 'yes' or 'no'.

Before the interview starts, you, the interviewer, must meet the subject and establish some sort of rapport. There are interviewers, a few, who can go in cold and get a good result. But they are few and far between.

The preliminary chat is, as it were, part of your research. With it you will establish the ability of the person being interviewed to talk, to express themselves, to answer questions. It is possible that this preliminary talk will end in you modifying some of your questions.

In your preliminary chat avoid asking the specific questions you will be asking in the interview.

Instead, indicate general areas of interest. If you ask the specific questions the filmed interview will give an impression that it has been rehearsed.

Before you start your interview have your key questions laid out and ready. You need a certain amount of flexibility but most of the time you will find that your first and logical thoughts or question order is much better than one you compile while winging it.

There are two main way of handling an interview.

The first is where the question is not heard and the questioner not seen. Instead, you get answers that are obviously directed at someone who is out of shot.

A series of answers like this can be edited together from either one person or several, to provide the effect of a continuous interview.

In this sort of interview you ask the question and then you keep your big mouth shut. If some sort of reaction is needed nod or shake your head vigorously or smile encouragement. If you speak you will have to be edited out afterwards. Which is not always easy.

This technique can be seen being used to magnificent effect in the movie 'When Harry Met Sally' which contains a series of such interviews with married couples describing their lives together. Magic.

The other type of interview is where you are both on screen in the manner of a normal conversation. This sort of interview can easily be covered with one camera.

Shoot the interviewee's answers first and then shoot the interviewer from where the interviewee has been sitting, asking exactly the same questions. At the end you do a series of 'noddies' that can be used for cutaways.

The key to making such an interview work is to get the person relaxed. Try to film them in a familiar surrounding so that they do not feel threatened. Keep the camera work and the lighting as unobtrusive as possible.

The first question should be a sound level check and should be totally innocuous.

Start the interview very gently in a chat mode and always move from soft to hard questions imperceptibly. Do not start like gangbusters or the interviewee will clam up or, in the worst case, walk off. It happens.

At the end of the interview I always ask 'Is there some question you would like me to have asked that I have missed out on?'

This allows the subject to expand on a point or deal with an area they feel has been skipped. It is quite remarkable how often you will get an excellent and usable response after that last question.

Start off with a long shot of the person being interviewed facing the interviewer. The interviewer's back appears, which gives a three-dimensional aspect to the shot and gets the scene in context for the viewer. Change the shot sizes in rhythm with the questions. New question, new framing.

Another form of interviewing on video is vox pop ? from vox populi, Latin for the voice of the people ? are quick interviews with people in the street to demonstrate public opinion on a subject.

What you want to end up with is a series of statements that can be cut rapidly together and, in the end, give a clear indication of the current attitude on a subject.

To make the interview more interesting change the shot size as a new question is asked. That is, switch off, zoom in from, say, mid shot to close-up, and then resume filming again.

Use different backgrounds and different eyelines.

Work out how many interviews you want and then shoot to that number with perhaps a 50 per cent safety margin. Do not go on shooting after that point. You could be getting useful footage for another scene rather than wasting your time. In vox pop moderation is the key.

Note carefully that subjects can move backwards and forwards when making a point and may even wave arms around in the air and you need to be prepared for this so they are always in shot. That the camera does not cut off parts of their bodies. Armless interviewees may be harmless interviewees but that is not the point of the excercise.

Gareth Powell has done many interviews for newspapers, magazines and television. He writes about making videos on his site, Digital images, http://www.pixelates.com