search

smarsland / pots

1 stars 1 forks source link

bad Beazley numbers #18

Open Armand1 opened 4 years ago

Armand1 commented 4 years ago

In general every Beazley number refers to a single pot. But not always. For some reason, the same Beazley number is sometimes assigned to clearly different pots. (They have different image numbers). One has to look carefully to ensure that it's not just a different view of a different pot, but sometimes it's clear.

In these cases I am going to assign a new subscript to the image. "Vx". For example,

pot_number1012440.0_url.img_3_V1.jpg pot_number1012440.0_url.img_3_V2.jpg

This assumes that the metadata applies equally to all pots with a given Beazley number. I think it does.

I think that this is mainly an issue with poorly-curated fabrics. It seems to be particularly rife in the Cretan vases.

My hope is that this doesn't fuck up analysis downstream, when we try to match vases to their original metadata.

LittleAri commented 4 years ago

Interesting. I'm currently grabbing the ID (re'\d+') from each pot when saving their contours so this might cause some issues if I've already grabbed the contours. Have you seen this with previous datasets yet (with the athenian, boetian, corinthians, east aegean or lucanian)? In these cases, I will save the files as 'cretan_vase_1012440_v1.csv' for example.

Armand1 commented 4 years ago

Good. Yes, it has cropped up sporadically --- but I have ignored it, picking one vase at random. But here we have 5 or 6 vases with the same id. Looking closer at the Cretan vases, it seems to me that most of them are, in fact, Bronze age. I actually doubt that we will use them ultimately. It just expands the study very deep. But still let's see if we can get nice contours.