This is Mike Figgis's follow-up to Timecode, which split the screen into four and followed the likes of Stellan Starsgard, Saffron Burrows, Salma Hayek and Jeanne Tripplehorn with a series of ideas, but no script, in a film production company in L.A. This time, we're in Venice, with some of the same people and an awful lot of others.

It constitutes a further instalment in his relationship with DV cinematography. It's also not very good, although there are excellently moments, for cameras and actors.

Figgis is one of relatively few established directors to have made a big noise about DV. This is surprising, given the flexibility of the format, but not really, considering that the industry is still entirely centred on 35mm. He is unusual in trying to harness the PD100, and its ilk, throwing it up on the screen in a way that exploits its strengths.

Hotel is an enhancement of Timecode. Some of the shots are beautiful and explore the full potential of DV. This is enabled by the rig that Figgis part-designed, which gives a sure grip, while avoiding the limitations of steadicam, or glidecam. However, in almost every other respect, it is a step backwards.

The conceit is that producer Jonathan (David Schwimmer) and director Trent Stoken (Rhys Ifans) are making a Dogma film of The Duchess of Malfi, starring Saffron Burrows as the Duchess. What we're seeing is a film around that film, part "making of", part Malfi - there are large chunks of the play in the script - and part other odd subplots, which rarely add anything to the inchoate proceedings.

This isn't too bad a starting point and there are moments when Hotel and The Duchess of Malfi are brought together, enriching both. Unfortunately, that doesn't happen often enough. There are plot strains, which have spread, weed-like, over the central structure and should have been chopped off before shooting. There's simply not enough spine in the central idea to support all the elements hung from it.

Whereas Timecode, whilst not scripted, had a relatively intricate storyline, centred around an eventful afternoon, the starting structure of Hotel is much looser. Occasionally, this translates as fluid, but mostly untethered. In the Making Of documentary, we see Burt Reynolds, who appears for 25 seconds, asking Figgis if the actors are supposed to come up with their own identities and backstories and Figgis saying, in effect, "Yes". Schwimmer then chips in with the fact that some of the actors got together the day before shooting started and came up with some inter-relationships. It shows.

At its worst, Hotel is truly bad. There is little that can save the ridiculous, tepid cannibalism scenes in a basement with John Malkovich, the ludicrous caricature that is TV presenter Charlie Boo (Salma Hayek), the lame piano playing and crooning. Some bits float meaninglessly - the flamenco, for example - and Ifans's Trent has all the vitriol, but none of the talent, that would have got him were he is.

Unlike Timecode, the film is a stylistic jumble. The Duchess of Malfi sections are presented in a recessed screen-within-a-screen, whereas I would have thought they might have had more impact had they been woven into the rest of the movie. The night vision sequences - another innovation, extremely effective in principle - are even further recessed for no reason that makes sense.

To call the Malfi film, "Dogma", is a shocking slur. No Dogma film contains muck on the lens, intrusive lens flare, obvious boom and camera shadows, as this does. That Figgis thinks he can get away with this kind of thing is part of the myth of such filmmaking. Patrick Stewart, the cinematographer, stands holding the rig to the camera and explaining that "with this, you are the crew", forgetting the small army of assistants, locations people, grips, gaffers, boom ops, assistant directors, who are in plain view in the documentary.

Nevertheless, moments of Hotel achieve significant things stylistically, photographically and from the performances. It's an achievement to have persuaded so many actors to participate at a fraction of their normal fee. When you think of the reason that Schwimmer might have become involved - the rush of improvisational acting, a world away from Friends - the warning signs are there. He's not bad, but his problem is a lack of direction.

More leadership and forethought might have saved the film and writer/director/co-producer/co-composer Figgis must bear the brunt of blame. He says he remembers feeling at the end of pre-production that "this is such a folly, because there's nothing, and we're about to bring 50, or 60, people together and spend a fortune and I really, in all honesty, have no idea how this is going to turn out." Then, he goes on to say how good it was. His original doubts were sadly accurate.