Part 2 - Restoration of `Going Down in the World' and Other Items

Conception and aims

Excuse me whilst I lapse into first person for this section - it seems more relevant.

I (Graham Denton) had been one of the team involved in making the films.  Although credited as `Lighting' it was, in reality Alan Fry that arranged the lighting as he wanted it.  I had two roles; the first was to provide power to the lights and the second was to do it without killing anyone.  I was, I
believe, successful in both but perhaps this is why I get `touchy' when I hear comments about using sticky tape to insulate joints!

Much work, far in excess of the physical underground filming, had gone into planning and organising the making of the film. I watched it at the anniversary weekend with very mixed feelings.  I was glad that something, at least, remained of our work but it was sad to see the state of the film of which we had all been so proud. Whist still watching it the idea came to me that it would not be difficult (shows how wrong I can be!) to re-synchronise the sound.  Over the rest of the weekend I remembered the silent copy and thought of the possibility of using it to patch missing video - the idea was growing.

I have always believed in defining the object before starting anything and, in this case, I decided that I wanted to reconstruct, as well as reasonably possible, the original film using whatever remained available but that nothing would be added - no re-recording any of the missing sound, for example. `Reasonably' in this sense meant that I would use my existing hardware and software, allowing for a little but not significant expenditure on whatever would help and that it must be completed in a timescale of, perhaps, a year - carrying out the work in spare time.  Whilst further work to improve the film to a far higher standard (what we may generally call `digital remastering') would be technically possible it was considered that such a task would be beyond the realms of practicality at present - perhaps someone may care to take on such a task in the future.  It may be that `restoration' is rather too grand a title - this is probably better described as a `reassembly'.  I also decided that, if possible, I would add title screens to present the film as an item of club history since this now seems more relevant than the original objective.

The following records the general approach and techniques used in that restoration.  The viewer must be the judge of whether or not the objective has been achieved.

Originals Available

The starting point was the degraded film as shown at the 50th anniversary weekend.  Tony Giles supplied a copy of the digitised version of all the original club film, initially as a single file for computer use.  This has been defined above so it need not be described again here.

A search of the club library, looking for the silent copy, yielded a VHS videotape labelled as `Going Down in the World' though its condition, etc. was not then known.

Although a VHS player was available it had not been used for many years and the risk of damage to the tape was considered too great so the tape was converted, professionally, to a DVD using the services of a local photographer's shop.  Playing the DVD revealed that it had no sound and appeared complete (there was no obvious large sections missing) but that it was, like the original, several minutes short (running for around 15 minutes instead of the original 18.).  Since a copy had now been made the tape was entrusted to the VHS player (that, ironically, worked fine!) and it was confirmed that the tape ran for the same time as the DVD so nothing had been lost in DVD conversion.

The silent copy, as well as appearing near complete, was of generally higher quality without the frame-slip and general deterioration of the original (sound) copy, though it did show signs of age related deterioration; faded colours (but often better than the original), scratches, dust marks, etc.
The fact that the copy was silent had made it unpopular and, perhaps, this resulted in fewer showings and contributed to its relatively better survival condition.

In summary, there were two copies of the video, each having their own individual problems, but only one copy of the sound.  These copies, then, are the material available at the start of the restoration project.

Video

The first stage was to convert the DVD from the professional VHS conversion to an editable file.  A hardware-software combination was purchased for this purpose (and for later use) and this enabled the DVD form of the video to be captured as a file for computer use.

Attention was directed to the silent copy with a view to ascertaining why it was short even though it appeared complete (obvious with hindsight but far from clear initially).  In an attempt to identify missing sequences in both versions a scene-by-scene analysis was carried out.  The length of the
various scenes was timed on both copies and compared in a spreadsheet.  As expected, a number of scenes were significantly shorter in the original, identifying where material had been lost.  The silent copy, however, was different; there were no obvious very short scenes but all scenes were a little shorter than the original copy - at least, where the original remained complete.  After some thought the idea eventually came to mind that this would occur if the silent copy of the film was complete but running fast.

An internet review of cine to VHS conversion followed and it became clear that two basic methods (although with many variations) had been in common use.  The simplest technique amounted to projecting the cine and filming the result with a VHS camera.  This produced a complete copy,
`warts-and-all', of the original film and any associated sound.  It seems likely that the original (sound) version had been converted using this technique or a variation of it.  The second method converted each cine frame in turn to VHS format.  This avoided frame-slip problems and also allowed better alignment of the film during conversion so it usually produced better results BUT the VHS copy would have the same number of frames as the cine that was being transferred.  Since VHS ran at different frame rates to cine it typically resulted in the copy running fast in the ratio of the frame rates.  In the case of `Going Down in the World' (16 frames per second) this would result in the original 18. minutes being reduced to only 14. minutes - very close to the effect observed.

It was decided to slow the silent copy in the ratio 16:20 (based on frame rates) and cut out, electronically, the small areas of damage that existed  - only around 5 sequences and each well under 1 second.  The idea of patching with video from the original was dismissed because the discontinuities introduced would have been more detrimental than the very small effect of the missing video. Having made this decision the actual work involved was small and, soon, the video that was to be used had been produced.

A brief internet review was carried out at this stage to search for software (at an acceptable price) that may be available to `clean' the video, automatically or semi-automatically removing the scratches, dust marks, etc. that had accumulated over time.  Nothing was found and this concept was not pursued further.

Audio

The soundtrack was separated from the original film.  Because of the audio editing software to be used this had to be done by the slightly torturous route of writing the sound from the original to a CD and then reading that back into the audio editor, which was really intended for transfer of vinyl or tape recordings to CD.  The complete audio was passed through a hiss filter to remove the noise typically found on magnetic tape recordings and also a click filter to remove as much as possible of the other noise on the soundtrack.  This degraded the sound in some ways (it produces a slightly muffled effect) but the overall quality was felt to be better than the original.  One problem associated with this only became evident later and that will be discussed in its appropriate chronological place.

The sound was then divided into a number of sub-tracks.  For convenience, terms such as `cut' and `splice' will be used here as if it were a physical process.  Initially, it was cut at the start and end of each section of distortion, deleting the distorted section but without splicing the two parts so created.  Next, the sub-tracks were divided further at all points of discontinuity, identifying those points audibly.  These sections of the soundtrack were identified so that they could be reassembled later in the correct sequence.  The number of sections into which the sound was eventually divided was not, unfortunately, recorded but was probably in the region of 25.

Each of these sections in turn was examined both audibly and visually, using waveform displays in the editing software. A significant number of discontinuities in the form of large clicks were found (probably resulting from splicing breaks in the original cine film) and these were removed by cut and splice techniques.  In all cases the length of sound lost was small, typically less than . second, but in places it resulted in the loss of the occasional word.  This, however, was felt to produce a better overall effect than leaving the discontinuity in the final soundtrack.

A study was then made of what we may call synchronisation points in the film.  Some synchronisation points were clear and needed to be exact; a simple example is the transition of background sound from scraping to water when the streamway first appears in the film.  Others were somewhat vaguer; for example the commentary about climbing between different levels needed to correspond to some point during the climb to the bridge spanning the chamber.  Finally, other points needed only come in the correct (meaning as in the original) sequence and did not fit
specifically to the video.  Additionally, it was, of course, essential that the opening and closing music corresponded precisely to the start and end of the video respectively.

Careful timings were made of both audio and video, using spreadsheets to compare and record these, and some of the shorter sound sequences were then spliced together, patching between them, so that somewhat longer sequences were produced that would be suitable for assembly to
the video later.

We should define `patching' here.  It is clear with a little thought that gaps would often (though not always) exist between the various sections of sound that were to be spliced.  Where this gap did not contain narration but only music or sound effects then a section of nearby sound, of the correct type, was copied and spliced into the gap.  This was done relatively crudely and no attempt was made, for example, to match the beat of the music but the technique proved generally acceptable and only becomes obvious when, whilst viewing the film, a conscious attempt is made to listen for it.

To illustrate, after removing damage and distortion only around 12 seconds of the background music used for the opening title sequence remained.  This is repeated 5 times in the restored version of the film to cover the 1 minute sequence.  Narration was clearly unique at any point in the film so the patching technique was not possible where commentary was missing.

All the above was laborious rather than technically difficult and resulted in 11 sound sequences to be reassembled to the video.

Reassembly

The software used for video editing allowed attachment of sound sequences at various specified points through the video - it did not require that a complete soundtrack was constructed and added in one step.  Accordingly, the various sound sequences were assembled to the video using the
synchronisation data obtained previously and then adjusted more finely to ensure correct synchronisation of these sequences with the video.  At this stage there were, of course, still gaps between the sound sequences.

Where possible, the gaps were patched as they had been when working with just the audio files, a task which was, again, laborious rather than difficult.  Only in very few cases were audio fades used during reassembly of the sound.  Where narration was missing, however, there remained silences and processing of these was approached somewhat differently.

The original concept had been to leave these gaps silent; certainly, nothing was to be re-recorded to fill them.  When playing the film in this form, however, the silences were found to be significant and distracting.  Since (as has been discussed previously) the complete soundtrack included projector noise - a clicking, characteristic of the old cine projectors - it was decided to fill the gaps with this to alleviate the silence, though this apparently simple task proved troublesome.

First, a silent section had to be located but the film always had either background music or sound effects and the only silent section, from which projector noise could be copied, was about 1 second between the title music and the start of the Grieg in the opening shot of the trees and cottage. Further, this section also contained 2 clicks that needed to be removed and although 1 was successfully deleted the other could only be reduced to a small but perceptible `thud'. It was at this stage that a problem due to the earlier filtering was discovered.  This had reduced the level of projector noise, interpreted as unwanted noise by the software, to a point where it was barely perceptible and below the levels that the software was intended to process.  It was too late to change it at this stage (it would have been necessary to start all work on the audio again) so the task
became that of editing this 1 second, very low level signal and then repeating it sufficient times to fill the silences; this was far beyond the intended use of the audio software being used.  After some degree of perseverance a sound was produced for use as `silence'.  This was not good but was felt to be as good as was going to be obtained and certainly better than true silence.  Depending on the reproduction characteristics of the sound system in use when the restored film is viewed, `silences' vary from a low level of projector noise in the background to barley perceptible projector noise but a rhythmic thumping from the poorly removed click.

At this stage, the actual film was complete but, since it was to be presented as a historic artefact of the club, it had been decided to add information screens to the beginning and end.  Production of these screens was the next stage.

Title and Data Screens

The data to be included was the personal decision of the restorer.  The underlying principle was to include that information that would be of potential interest to anyone viewing the film in the future but who had no knowledge of the making of the film of those people that were involved. There would be a compromise between including sufficient information and not producing an effect that distracted from the film, which should remain the principle focus of the restoration.

The picture used as a background to the information screens was selected because it not only provided a visual link between caving and filming but, more significantly, it showed the actual camera that had been used to make the film. Desktop Publishing software was used to create the title and data screens.  This permitted editing of the brightness and contrast of the picture to ensure that text was readable and, of course, allowed
the text to be superimposed and then adjusted as required.  When the image was complete for any screen it was saved as a picture (.jpg format) and the video editing software then used to add these to relevant places in the film. Using video fades with a constant background picture created the illusion of the text, only, fading between screens.

In parallel with the work on the film the information to be included had been researched and checked to a limited but adequate, it was considered, degree to ensure accuracy.  The research on dates is detailed in part 3 and most other data was listed in Pelobates of the time.  Some of this data
was re-printed in the clubs 50th anniversary special edition and this showed a change of name for one cast member.  A check with the editor of both editions confirmed that this was, indeed, a change of name (not an error) and it was decided that the information screen would follow the
precedent set in Pelobates and use the later name. A final detail of the information screens was the addition of the club logo at the start of the film, simply because it seems appropriate.

The Final Media Files

It is worthy of note that, throughout this project the editing of the film had been saved as a `project' and the film itself (as a media file) was not saved and re-loaded at any time other than for testing purposes.  Only at this stage, when complete, was the film saved as a finished media file.  This was essential to avoid the potential progressive deterioration that can occur during such save and reload procedures. The film was saved in high quality .mpeg-4 and .wmv formats.

At this stage it could be argued that the original objective of the restoration had been achieved, at least as well as was reasonably feasible.  It was felt, however, that in view of the additional material available it was worth pursuing the project to ensure that all the clubs filming activities were
preserved.

Unused footage from `Going Down in the World'

The unused footage from `Going Down in the World' had been assembled into a single length, apparently randomly, and included in the digitised files obtained.  Using the same techniques and software as detailed above this was rearranged into sequences relating to the three filming
weekends and assembled into a `film'.  Note that this was not an attempt to make another film from the scrap and neither is it an out-take file - it is simply a record of the rest of the film shot during the making of the clubs film.

Merstham Filming Experiments

Although not of adequate quality to make a film (as detailed above) the footage taken in the quarries at Merstham was part of the digitised file obtained and this, again was preserved by making into a `film' as was the unused footage.  In this case, the scenes seemed to be in a reasonable order (possibly the order they were shot) so this was not re-sequenced and only these few seconds that were not even viewable were edited out.

The Pre-Restoration Version of `Going Down in the World'

A purist may argue that the unrestored version, even if by now a long way from the original, is still the more accurate version and is a true representation of all that remains of `Going Down in the World'.  Accordingly, the title screens were added to the unedited pre-restoration version so that this, also, would be preserved.

Presentation - The DVD

When the restoration was conceived no real thought was given to the final presentation.  At this stage, however, in view of the four films, in the loosest sense, that had been produced it was felt most appropriate to present them in the format that may be considered the nearest equivalent to a 1970s cine film - this is, of course, the DVD.  It was decided to assemble the four items onto a single, menu driven, DVD for use in a DVD player.  The computer files that had been produced would be preserved so that future editing in a computer would remain possible.

`Thumbnails' were extracted from each of the films to use in the menu and a background image was prepared using the camera and helmet photo.  This was used in standard DVD mastering software to produce the final DVD. Using desktop publishing software an inlay card for the jewel case used to store the DVD was produced, having a list of contents on the front cover and a summarised history of club filming and the restoration on the inside.

Only one disc was burnt from the DVD mastering software and this was fully tested and then retained as a master by the restorer.  Further copies (including a copy to the club library) were made by copying from this master or, of course, by subsequent re-copying.