Tom Goskar

Archaeologist (FSA) specialising in 3D recording and surface enhancement. Working with rock art, inscriptions, and decorated objects. Digital problem-solver and developer of techniques.

Also: traditions and customs, folklore, mandolin player, trad music, audio restoration, synthesisers, e-Estonian 🇪🇪

Based in Cornwall, UK.

#3Dscanning #rockart #enhancement

#archaeology #archaeologist #digitalarchaeology #3Dscanning #3Dcapture #AI #folklore #folkmusic #mandolin

2024-10-18

Heading out to do some fieldwork recording some medieval crosses in high resolution 3D to study their chi-rho inscriptions.

2024-09-20

There is hope still for Sketchfab, I think, according to unrealengine.com/en-US/blog/fa "Sketchfab’s licensable content will move to Fab over the coming months, while Sketchfab.com will remain dedicated to its services that enable creators to upload 3D models and share them publicly or privately via the web." They need to provide more clarity and reassurance if Sketchfab has a long term future. Let us all learn from this.

2024-09-20

@electricarchaeo Oh no indeed. Thomas Flynn (former Sketchfab Cultural Heritage Lead, made redundant earlier this year as a sign of things to come) has compiled some useful info: docs.google.com/document/d/1Gh

2024-09-20

I'm a little late finding out that Sketchfab is going to disappear next year. I shouldn't be surprised, but I am disappointed. So many cultural heritage orgs have their amazing content on there. I have used it since 2014. If your org has content currently on Sketchfab - what's your migration strategy?

2024-07-10

@tkinias I’m doing similar things with IA PDFs. After Tesseract struggled I started using Claude Haiku via the Anthropic API. Works out to be about 0.3GBP per 300 pages to create txt versions with page numbers using the IA zips of individual page scans. I used Claude to write a quick Python script to automate it all. Its accuracy is superb and it’s dealt with the 19th century sources nicely. It might be possible to extend to creating a PDF text later. Happy to share my script.

2024-06-13

@Documentally The camping gear would be rapidly adapted into hang gliding gear! It’s wild out there already.

2024-06-12

Aaaand it’s cancelled. We might be able to get there, but not back again…

2024-06-12

Tomorrow is supposed to be the first (and maybe only) time I’m being taken by helicopter to do a 3D survey on an island. Except that the weather forecast now reckons heavy heavy rain, 50mph gusts, and the site is on the highest point on the island (45m). Not fancying my chances!

Tom Goskar boosted:
2024-05-07

Does anyone know of a paper or dataset that compiles the earliest known dates of agriculture in a given region? Globally, ideally, but I'd settle for a continent or two...

(I know there are several options for the European Neolithic, but beyond that?)

#archaeology #opendata

2024-04-22

@neildixon That’s good to know - thanks!

2024-04-22

@neildixon What spec Mac are you using it with? I’m still on a last-gen Intel iMac but local LLMs might tip me into upgrading before I wear it out :-) GPT4All is very slow (i9, 16GB AMD GPU, 40GB RAM).

2024-04-22

@neildixon are you using GPT4All?

Tom Goskar boosted:
2024-03-27

In January, we launched our new Omeka quarterly newsletter! Sign up today to stay up to date on the latest Omeka news: omeka.org/news/

2024-03-24

Opened ZBrush for the first time in a few years. I still hate the UI. Tempted to sell the license, except apparently it's now owned by Maxon and my license isn't listed in my 'migrated' account, and there was for a while docs on transfers, and a hefty fee, whose URLs are now 404-ing. What a mess.

2024-03-06

Social history…

Graphics on a coach depicting old logos for Facebook, Twitter, Instagram and the now-defunct Google+
2024-03-01

A few people have asked me about how energy intensive LLMs such as ChatGPT & Gemini are to use regularly, and it's a good question (answer: lots). Energy efficiency is being tackled at both silicon and programming levels. Latest 1bit LLMs might enable new optimised chips: arxiv.org/html/2402.17764v1

2024-02-22

Interesting blog post from Shawn Graham on using LLMs in historical research and connecting them to databases: electricarchaeology.ca/2024/02

2024-02-22

@electricarchaeo This is great. Exciting times! I love the idea of a 'grimoire' for prompts/sources/howtos. Nice to see GPT4All mentioned too. I wasn't having a great deal of luck constraining it to my unstructured data (using LocalDocs db).

2024-02-22

I've been fiddling with using ChatGPT 4 as a research assistant and blogged about it here: tom.goskar.com/2024/02/13/unea

I have a feeling that this won't be the only blog post on the matter. The TL;DR of the post is that ChatGPT 4 is a seriously useful tool for researchers if you understand how to use it and are very aware of its limitations (and biases).

2024-02-10

@mia Wishing you a smooth flight! If you visit WA Museum Boola Bardip look out for my Bardi dancers at Stonehenge photo in the galleries 😀

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst