A small update from my ongoing fight with spreadsheets 📊I’ve been building Edit Credits on my own as a personal project, and one part of it has given me far more trouble than I expected: the credits parser.
The idea sounds... Read more
A small update from my ongoing fight with spreadsheets 📊
I’ve been building Edit Credits on my own as a personal project, and one part of it has given me far more trouble than I expected: the credits parser.
The idea sounds fairly simple — upload a spreadsheet and turn it into editable end credits.
In reality, every document seems to have its own logic. Blank rows, merged cells, roles written above names, several names inside one cell, comments in unexpected places… At first, the parser was simply too unpredictable. A new document could produce a completely different result.
After a lot of trial and error, I eventually got it to work reliably. The only problem was that analysing an average credits document could take around an hour. 🫠
That felt slightly ridiculous. In an hour, you could probably do much of the work manually.
So I spent some more time rethinking the process. Instead of analysing and reconstructing every piece of text, the parser now treats cells and repeating spreadsheet structures as the main units of information. It only looks more closely at the genuinely ambiguous parts.
It now processes most larger documents in around 7–10 minutes. One recent spreadsheet with 590 rows took approximately 6–7 minutes. ⚡️
It isn’t instant, but seven minutes is enough to make some tea. An hour is a different story. ☕️
The nicest part is that the first people have now tried Edit Credits with their own documents — files I had never seen and hadn’t used while developing the parser. They did it without any guidance from me and were happy with the result.
That was honestly a relief. I was slightly worried that the system had simply become very good at recognising my own test files. 😅
Edit Credits is available in public beta, so if you work with film or television credits and have a spreadsheet lying around, you’re very welcome to try it:
👉 editcredits.com
And if something breaks, please tell me. Real, slightly messy documents are much more useful than perfect test files.
#PostProduction #FilmProduction #EndCredits #BuildInPublic
I’ve been building Edit Credits on my own as a personal project, and one part of it has given me far more trouble than I expected: the credits parser.
The idea sounds fairly simple — upload a spreadsheet and turn it into editable end credits.
In reality, every document seems to have its own logic. Blank rows, merged cells, roles written above names, several names inside one cell, comments in unexpected places… At first, the parser was simply too unpredictable. A new document could produce a completely different result.
After a lot of trial and error, I eventually got it to work reliably. The only problem was that analysing an average credits document could take around an hour. 🫠
That felt slightly ridiculous. In an hour, you could probably do much of the work manually.
So I spent some more time rethinking the process. Instead of analysing and reconstructing every piece of text, the parser now treats cells and repeating spreadsheet structures as the main units of information. It only looks more closely at the genuinely ambiguous parts.
It now processes most larger documents in around 7–10 minutes. One recent spreadsheet with 590 rows took approximately 6–7 minutes. ⚡️
It isn’t instant, but seven minutes is enough to make some tea. An hour is a different story. ☕️
The nicest part is that the first people have now tried Edit Credits with their own documents — files I had never seen and hadn’t used while developing the parser. They did it without any guidance from me and were happy with the result.
That was honestly a relief. I was slightly worried that the system had simply become very good at recognising my own test files. 😅
Edit Credits is available in public beta, so if you work with film or television credits and have a spreadsheet lying around, you’re very welcome to try it:
👉 editcredits.com
And if something breaks, please tell me. Real, slightly messy documents are much more useful than perfect test files.
#PostProduction #FilmProduction #EndCredits #BuildInPublic
