Roman nerdery on main
Jul. 31st, 2020 12:38 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
I've recently come across Photoreal Roman Emperor Project: 54 Machine-learning assisted portraits. In particular, I liked seeing, in the detailed posts, what sources the artist used as his bases for each portrait.
I will admit that three things I find odd:
- The difference in lightsources in the composite end-portraits. Some are lit softly, some harshly/with higher contrast and some even look washed out/over exposed.
- Hair textures. Okay, wtf? Pretty much everyone's hair is straighter in the end-portraits than in the busts. It's most aparent in say Lucius Verus but it's pretty obvius in most of them. Starting with Augustus!

Does this not look like two different hair textures to other people?! He went from having wavy/wide curl hair to having straight hair. Maybe I'm justused to mediterranean hair mad because OG Augustus has hair v similar to mine and the composite does not.
- Ulpia Severina appears to have been composited entirely from gendershifting her husband Aurelian.

Whici is certainly A Choice but ok! I am more mad that zero effort appears to have been made to recteate the hairstyle shown in the coins. I know recreating numismatic hairstyles is a pain and a half -- having myself tried to recreate Carthaginian female hairstyles based on coins depicting Dido etc -- and it's not like I expct everyone o know who Janet Stephens is, but it took me less than five minutes, starting onthe wikipedia page for Roman hairstyles, to find this video of her recreating a similar, though earlier, hairstyle.
On an only vaguely related note, I wonder when Elagabalus became known as Héliogabale in French. The Ela -> Helio shift likely happened because sun cult but when?
I will admit that three things I find odd:
- The difference in lightsources in the composite end-portraits. Some are lit softly, some harshly/with higher contrast and some even look washed out/over exposed.
- Hair textures. Okay, wtf? Pretty much everyone's hair is straighter in the end-portraits than in the busts. It's most aparent in say Lucius Verus but it's pretty obvius in most of them. Starting with Augustus!

Does this not look like two different hair textures to other people?! He went from having wavy/wide curl hair to having straight hair. Maybe I'm just
- Ulpia Severina appears to have been composited entirely from gendershifting her husband Aurelian.

Whici is certainly A Choice but ok! I am more mad that zero effort appears to have been made to recteate the hairstyle shown in the coins. I know recreating numismatic hairstyles is a pain and a half -- having myself tried to recreate Carthaginian female hairstyles based on coins depicting Dido etc -- and it's not like I expct everyone o know who Janet Stephens is, but it took me less than five minutes, starting onthe wikipedia page for Roman hairstyles, to find this video of her recreating a similar, though earlier, hairstyle.
On an only vaguely related note, I wonder when Elagabalus became known as Héliogabale in French. The Ela -> Helio shift likely happened because sun cult but when?
(no subject)
Date: 2020-07-31 12:05 am (UTC)What a fascinating project! For what it's worth, my immediate reaction to that Augustus is that his hair looks way too English.
-- having myself tried to recreate Carthaginian female hairstyles based on coins depicting Dido etc --
DO YOU HAVE PICTURES?
(no subject)
Date: 2020-07-31 11:55 pm (UTC)I would have to find whichever sketchbook that was in, but that could be arranged ;)
(no subject)
Date: 2020-07-31 12:41 am (UTC)Ulpia Severina being based on her husband's bust is... FINE, I GUESS, but I'm glad she was included nonetheless, because I had no idea there was a (maybe) female Emperor of Rome! I googled her earlier when I first saw this project, and it was definitely the best thing I learned today.
(no subject)
Date: 2020-08-01 12:01 am (UTC)Right?! Like. I don't want to say the sample set for the machine learning was biased but machine learning sample set has notorious bias problems towards northern European descent people.
The reasoning was probably that it was the right time period and they had the same ethnicity so that was close enough. Which, given the lack of data, is a valid choice! Trying to dig around for people she might have been blood related too is another multiple hour project unto itself. Still. Super glad she was included because I also mearned about her from this and she is so fascinating!
(no subject)
Date: 2020-07-31 03:49 am (UTC)(no subject)
Date: 2020-08-01 12:03 am (UTC)