“Mellow & Yellow”
Took the opportunity to walk around the almost empty streets of nyc on Christmas day.
Shoutout to @fr_ncochen and @cillwarraway for the company!
I think my ML model is currently overfitting, which is why it is able to accurately reproduce characters from the training dataset but struggles to come up with anything coherent otherwise. Its also lacking details in the glyphs which i will have to tackle later, perhaps a larger network?
Would love to hear ideas!
(Oh, and also, I realize now that the hex code mapping is inaccurate in the model, prolly due to floating point inaccuracies as im normalizing the entire 100,000+ hex scaler down to [0,1] so thats another thing im going to try fix for the next attempt)
#art #machinelearning #machinelearning #languages #glyphs
2 weeks of my summer in 154mm of film.
My camera’s film advance lever malfunctioned and I had to manually wind and unwind the film everytime it jammed, with no sense of where I was on the roll. This is the result.
Shoutout to @coastalfilmlab for rescanning with overlaps and as Xpan so i could stitch them together!
![I think my ML model is currently overfitting, which is why it is able to accurately reproduce characters from the training dataset but struggles to come up with anything coherent otherwise. Its also lacking details in the glyphs which i will have to tackle later, perhaps a larger network?Would love to hear ideas!(Oh, and also, I realize now that the hex code mapping is inaccurate in the model, prolly due to floating point inaccuracies as im normalizing the entire 100,000+ hex scaler down to [0,1] so thats another thing im going to try fix for the next attempt)#art #machinelearning #machinelearning #languages #glyphs](https://sp1.picnob.com/p/pt_6764612151664467384339_0_5a48eabaceb6b8b6dfc102ec254e3fc5.jpg?o=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&h=c10db3778d8cc120033e934448c92fe3)
