larsWould anyone know where or how to find generative art based on NLP, or people doing such thing?
to expand a little: a designer friend would like to create artwork for magazine articles generatively, basing the resulting artwork on text characteristics.
Jimmy V@lars most of these are not in the 'generative' category - but still might find some inspiration browsing: https://pinboard.in/u:vlandham/t:textanalysis/t:visualization/
@vlandham there's some serious gold in that collection! brilliant stuff, like the punctuation piece for example. Totally couldn't see that from that rock I live under. Thanks a lot!
@lars sure! Hope it might be helpful! Also long ago Yannick and I worked on a text analysis and vis workshop. A compressed slide deck for that is here: http://vallandingham.me/textvis-talk/#1 - with vis at the back. Again, not much generative, but could still serve to get some pointers.
@vlandham Great stuff, of course! I guess the generative is less important than the artistic encoding of text features. Some great gems I didn't know about or have forgotten in there. Might have to wield some code (sentence drawings with arcs?). Thanks Jim!
marco@lars this article explains the basics and shows an example of generated Shakespeare http://karpathy.github.io/2015/05/21/rnn-effectiveness/
@marco nice. Thanks! RNN’s could be a good starting point, maybe visualising differences between original and generated output...
Tomas SA2TMS
pettter ✅@tomas @lars Depending on your definition of "art" and "based on NLP", various markov bots on birdsite (and on here) qualify (like the erowidcoin thing, that Markovs based on drug trip reports and cryptocoin whitepapers).
I _think_ I've also seen folks doing mood analysis of texts for generating melodies that match in meter and mood, but I'm not sure who or where. Shouldn't be too hard, considering the libraries out there. Is it more something like that you're thinking of?
I _think_ I've also seen folks doing mood analysis of texts for generating melodies that match in meter and mood, but I'm not sure who or where. Shouldn't be too hard, considering the libraries out there. Is it more something like that you're thinking of?
Kristin (vis.social Admin)@lars by NLP, do you mean Natural Language Processing? If so, then I've done some, but not as deeply into the NLP as I'd like to. This is a current work-in-progress of mine. The words from the word-bursts are extracted from video transcripts with NLTK. It's still very rough, as I'm working on the viz/art and dataprep in parallel. http://kristinhenry.github.io/randyRainbow.html
@kristinHenry Very nice indeed! Really good to see it progressing as I remember seeing an earlier stage some time ago...
I'll be including your Color Stories of Poetry in my list as this will suit my friend's purpose nicely (feed text to an algorithm to produce an artistic visual representation). Awesome, Thanks!