Hi all! Dave is back again for a guest post on accuracy – please do feel free to comment or discuss as Dave would love to hear your thoughts and ideas on the topic to get a discourse going between games professionals and archaeologists! Enjoy – T
Now then…..I’d thought I’d write a series of posts around “accuracy”….not historical accuracy in games (I’m looking at you Ubisoft) but the accuracy of the 3d model, the scanned data. Have had discussions with Tara in the past regarding this and get the impression you heritage types are pretty hung up about the “100% accurate scans”. From my point of view having dabbled with scan data, I hope to take you through my reasoning over the next few posts and to why I think you’ve lost before you’ve even started…in effect 100% is a flawed notion….99% is more accurate.
But firstly, a passing thought I had the other day….photos…nothing major, just the humble photo album…the personal photos we have of ourselves, our friends and family stored in the cloud, which are on our Facebook walls, the ones our parents embarrass us with when a new partner arrives on the scene….and even the ones stored in museums…
Photos are moments in time which have been captured…never to be repeated again….even with the best Intentions to recreate them. Lighting, time of day, weather, film used, type of camera used, fashion, your mood at the time the photo was taken…so many random variables…seen and unseen captured on film….on digital. Sadly for the purposes of this post (though I’m sure not for her) Tara hasn’t aged one jot, so as a better example, here’s the cast of Star Wars through the ages:
So here’s the thought….isn’t the 3d scanning/photogrammetry of objects the same as a humble photograph? Okay, objects may not age as quickly as us humans but they can get damaged, and discolour etc. Now these are just objects in a controlled museum setting. What about objects and areas in the field…Stonehenge is a good example. Subtle things like the moss and wind erosion. A personal project I’m working on has elements of graffiti both carved and painted onto rock and (sadly) gravestones.
Basically the object you scan on a particular day won’t be the same object you scan 3 years later. On top of that is the scanning technology itself. Anyone who’s used tech in any capacity knows full well that in terms of hardware and software it’s a fast paced world. Scanning technology is BETTER and more ACCURATE than it was 5 years ago. Great, so what happens to all those scanned artefacts that UK museums spent thousands on scanning 5 years ago? Have they lost something? Are they now worthless/less accurate? In terms of photogrammetry, what happens if a photogrammetry trip covers say 4 days in changeable weather? Sunny one day, rainy another, cloudy the last 2 days? And oh, the light quality in the morning will be different to the light at dusk. Yes there are techniques to counteract some of these issues but will it ever be 100% accurate?
So many moving parts, changing variables…can freeze you on the spot just thinking about it. So what do we do? Not much accept embrace it; accept it for what it is. 100% accurate? No but we get as closed to 100% as we can do with the tools we have at our disposal. Aim for the high 95%+ and let the world know it’s 95%+
So here are some thoughts that maybe heritage and digital could maybe look into:
- Treat 3D scans as heritage objects in their own right. Maybe scan then re-scan 10 years later the same object?
- Metadata: Help future heritage generations. Standardise 3d scans and photogrammetry by providing metadata which is attached to each scan. Time of day, date, weather, scanner (or in the case of photogrammetry, camera) used Software used to process the scan. Basically try and record as much as those moving variables so that future generations can process the context of the scan. !!STANDARDISE!!
- Have some form of “accuracy scale” what defines a 95%+ scan? Type of equipment? Software? Define best practices. Let the public be aware of this scale and be honest about it. If you go in a museum and there’s a scanned Roman helmet, maybe in the info it could be a 98% accurate scan. As well as building trust, it will also push the quality bar of the scan/technology further.
- Tell the public that this scan was captured on this date, educate them, let them know that the scan is essentially a photograph and the “real object” may have changed