This is a very old thread, but I just saw it.
I am the technical manager of London Short Film Festival, and the author of the "how to make a DCP" guide the OP referred to.
First off, I'm really sorry to hear that the film did not screen as intended.
And why indeed did the DCP-making guide I wrote suggest adjusting peak audio levels to -10dB in DCP-o-matic?
A little bit of context: 2015 was the first year that LSFF went DCP. Prior to that year, most screenings had been from Digibeta. To say that we had teething problems with the transition to DCP would be understating it. There were 500 films in the festival that year, the task of getting them all to screen was massive, and it was pretty chaotic. We were pretty overwhelmed.
I am very pleased to say that these days it's a completely different story. It's now a very slick, well-oiled machine and at this year's festival the feedback from filmmakers was that they were extremely pleased with how their films looked and sounded in the cinemas. I'm not aware of any film that screened incorrectly, and we didn't get a single complaint from filmmakers. That may not sound like a very high bar to set as success - no-one complained! - but it's a feat not many festivals of this size achieve. It takes very diligent processes and careful work by a lot of people not to allow even one mistake to slip through the net.
LSFF is a funny beast. The mix of films is very wide. At one end of the spectrum there are well-budgeted films which have been through a full post-house grade, mix and online; at the other end there's plenty of zero-budget films which have been mixed on headphones on the director's laptop and there's been no online process at all. So devising delivery guidelines that fit that whole range is very challenging.
LSFF also has a rather DIY ethic, and in 2015 we felt it was in keeping with that ethos to encourage filmmakers to seize the means of production and empower them make their own DCPs with free software like DCP-o-matic, rather than force them to go to post houses who were charging people £1000+ to make a short DCP. In many cases, that sum would be more than the entire budget for the film! The festival at that time was run on a shoestring by a small group of very committed and underpaid people. The festival certainly did not have the resources to make DCPs for all the filmmakers.
The DCP-making guide was written, as one person has commented above, very much as a "dummies guide". It aimed to be unthreatening and to make it as simple as possible for non-technical people to do a decent job of making a DCP while avoiding the most common pitfalls.
Which brings me to the -10dB question. We'd found it was a common problem that the mixes people were putting into DCP-o-matic were too "hot". Perhaps there was a bug in DCP-o-matic at that time, or perhaps it was that mixes peaking on 0dB had "true peak" above 0dB. But the result was that we were seeing a lot of DCPs that had distorted audio. To avoid this problem, we wanted to ensure people attenuated the audio so the peaks were well below 0dB.
But why -10dB? The answer is really unscientific. In DCP-o-matic there's an audio waveform display which has horizontal lines at 0dB, -10dB, -20dB, -30dB. To make it really simple for people, we said no peaks above -10dB because this was easy to explain and easy to see on the waveform display - if everything was below the -10dB line it was "good", if anything went above that line it was "bad".
Obviously, this would lead to very different loudness between films with varying dynamic ranges in their mixes. The intention was that each film would be then adjusted up or down by cinema projection so all the films in a programme played at the same volume. All the cinemas were briefed on this.
In the case of the OP's film, obviously the cinema didn't manage to do this properly. However, just to be clear, I'm not laying the blame with the cinema's projectionists. As I mentioned above, the delivery of the festival that year was pretty chaotic, and some films were delivered to cinemas only days before the screenings. Given how messy it all was, and how little time there was for cinemas to screen-test, it was inevitable that the ball would get dropped somewhere - as it clearly was in this case. I take responsibility for that, and I'm sorry.
Concerning the points made above that filmmakers should get their DCP made professionally: I've changed my position on this since 2015, and we no longer encourage people to make their own DCPs unless they really know what they're doing, and I've taken down the howto guide.
I do stand by the quality of DCP-o-matic. I think it's a brilliant piece of software, and Carl the author is very responsive to bug reports and puts a huge effort into developing it. It's now at the point where it does an excellent job at all the important parts of the process - colour conversion, J2K encoding - and crashes are very rare. However, you do need to know what you're doing to get it right. And the biggest problem is that most people can't play back a DCP once it's made, so you're essentially they are "flying blind".
Our process now at LSFF and the other festivals we (Cinebox) manage tech for is: receive films either as DCP or ProRes or whatever is the best quality version of the film available, and where a DCP needs to be made, we make it ourselves. Regardless of whether filmmakers deliver DCP or ProRes, we do a very thorough QC. We typically reject 30% of materials due to problems found in QC, and we advise filmmakers how to resolve these problems, or fix them ourselves where we can. Essentially, for many lower budget films, we play the role of online edit and mastering. We feel this is an important role in helping people get their films to screen looking and sounding their best, and as a festival LSFF are proud to be able to support independent and low-budget filmmakers in this way.
In short, we've come a long way in the past 3 years!
Sorry I didn't see this post and respond at the time. Of course it's now rather too late! But I hope this explanation makes sense and may be of interest to anyone who happens across this post in future.