{"id":37400,"date":"2023-09-30T09:00:21","date_gmt":"2023-09-30T07:00:21","guid":{"rendered":"https:\/\/www.radiotimes.com\/?p=1888119"},"modified":"2023-10-01T16:59:35","modified_gmt":"2023-10-01T14:59:35","slug":"could-ai-breathe-new-life-into-vintage-doctor-who","status":"publish","type":"rss_feed","link":"https:\/\/c01.purpledshub.com\/radiotimes\/rss_feed\/could-ai-breathe-new-life-into-vintage-doctor-who\/","title":{"rendered":"Could AI breathe new life into vintage Doctor Who?"},"content":{"rendered":"

AI has been used as a force for good and evil in the show – but what about its real-life use? <\/p>

By Steve O’Brien\n \t\t<\/p>

Published: Saturday, 30 September 2023 at 07:00 AM<\/p>

\n\n

Doctor Who has been exploring the ideas behind artificial intelligence almost from the moment it began. From WOTAN<\/a> to K9 to the Cyberium<\/a> to the TARDIS<\/a> herself, AI has been both a force for good and for bad in the show.<\/p>\n

But while Doctor Who<\/a> has been talking about artificial intelligence for 60 years, it\u2019s only recently that we\u2019ve begun to ask the question: what does its real-world possibilities offer the show itself? There are, of course, ethics issues with many of the opportunities AI provides \u2013 using the tech to generate first-draft scripts, deepfaking long-dead actors, etc \u2013 but one area which may well benefit, without too much controversy, is the classic, i.e. pre-revival series.<\/p>\n

We\u2019ve all seen those AI apps that can sharpen up a foggy old photograph or add a splash of colour to a vintage black and white image, but can that or similar tech be applied to classic television? Part of the queasiness that schedulers and streamers have about TV from the dim and distant is how poor the picture quality is compared to the needle-sharp HD and 4K programmes we\u2019ve become accustomed to.<\/p>\n

Most TV of the 1950s and \u201960s \u2013 in the UK anyway \u2013 was recorded on 405-line cameras, later upgrading through the late \u201960s, \u201970s, \u201980s, \u201990s and much of the \u201900s, to 625-line (or 576i) \u2013 that\u2019s far less visual information than the 1080p of High Definition or 4320p (4k) of Ultra HD, hence the much blearier, fuzzier picture. Older shows made on film, such as Star Trek<\/a> or The Avengers, don\u2019t have this problem, as celluloid can be upscaled much easier.<\/p>\n

Welcome to 60 Days of Doctor Who<\/strong>, RadioTimes.com\u2019s ultimate celebration of all things Who as the show gears up for its 60th anniversary. We\u2019ll be bringing you 60 days of exclusive Doctor Who content, from interviews and newsletter takeovers with the show\u2019s stars to the hottest takes from our team of Doctor Who superfans. Stay tuned!\u00a0<\/p>\n

<\/p><\/div>

\"David<\/div>
BBC Studios<\/figcaption><\/div> <\/div> <\/div> <\/section>

But can AI offer a way of extending the shelf life of these pre-HD programmes? Peter Crocker is a member of the Doctor Who Restoration Team, and is in charge of sprucing up those older episodes for Blu-ray release. He already uses AI \u201cappropriately and sparingly\u201d, he tells us, but cautions that the technology isn\u2019t currently there to upscale, say, an old William Hartnell<\/a> episode into eye-wowing UHD.<\/p>\n

\u201cAt present, the benefits for any vintage TV made on old, non-digital equipment are relatively minor,\u201d he tells RadioTimes.com<\/a>. \u201cAll of the programs for upscaling images have been developed primarily for the movie industry. The vast majority of films recently have been shot or post-produced at 2K resolution and for cinemas, upscaled to 4K.<\/p>\n

\u201cThat process is very good because the source is typically clean and \u2018digital\u2019. Old TV is inherently noisy and smeary, with lots of artefacts. The AI programs just don\u2019t know what to do when pixels are present that might be picture detail but are actually just noise in the image.\u201d<\/p>\n

Part of the problem, Crocker says, is that the jump from a 405-line picture to even standard definition, let alone HD and 4K, is just too big.<\/p>

\u201cGoing from 405-line to SD is a 250% increase in pixel number and SD to HD is 400%, HD to 4K is 400%, and so on,\u201d he says.<\/p>\n

\u201cAnother factor is the range of brightness and colour that can be displayed. One reason modern productions look so much better is the high dynamic range that just isn\u2019t there in recordings from the first few decades of TV. In future, AI might be able to guess what detail might be present hidden in the bright whites or dark shadows that are currently just white or black.<\/p>\n

\u201cWe have to be realistic, though, and for old TV like Doctor Who or Dad\u2019s Army, the aim should be to make the programmes look acceptable on modern huge flat TV screens which would be horribly unforgiving if fed raw, unrestored recordings.\u201d<\/p>\n

Without any kind of a slick-up, there\u2019s a danger that older television will become hard to find, the preserve of cultural scholars and cult TV devotees only. But Crocker believes that the technology will one day be there to digitally recreate vintage TV episodes.<\/p>\n

\"Doctor\n<\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/source><\/picture>\n<\/div>
Doctor Who: The Smugglers (in colour)<\/span><\/figcaption> BBC<\/i><\/span><\/div>\n

\u201cMy Tomorrow\u2019s World prediction is that eventually image and data processing will be fast enough for software to assess a TV programme, identify the elements within the pictures \u2013 people, faces, costumes, sets, props, lighting, etc \u2013 and create a full 3D library of those elements which can be replaced with HD versions generated piecemeal from other matching details and textures,\u201d he explains.<\/p>\n

\u201cHigh quality photos of the sets and actors if they exist could help. The movements will be tracked and reproduced and the elements will be mapped back to produce what is effectively a photo-real, high-resolution computer animation based on the original production.<\/p>\n

\u201cThat technology is there now, in high-budget movies for quite limited use. I think in another 10 years it might be feasible for short, easy clips, but really it\u2019ll be 20-30 years before it would be achievable to an acceptable standard on the sort of budgets \u2013 and material \u2013 I work on.\u201d<\/p>\n

So while we shouldn\u2019t necessarily expect Evil of the Daleks to look as piercingly clear as Eve of the Daleks anytime soon, there are, and will be, AI-assisted ways of smartening up those older stories.<\/p>\n

\u201cIt could certainly give old shows a bit of spit and polish, but it\u2019s not a magic wand.\u201d<\/p>