[{"data":1,"prerenderedAt":239},["ShallowReactive",2],{"project-12838152":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":35,"aspects":36,"strip":42,"post_text":44,"what":45,"said":66,"maker_in_thread":140,"shape":145,"grid":150,"attempts":201,"same_week":202,"maker_shows":33,"maker_hidden":231,"reader_version":232,"card":233},"12838152","Fast Style Transfer","Fast Style Transfer in TensorFlow","title","Show HN: Fast Style Transfer in TensorFlow","https:\u002F\u002Fgithub.com\u002Flengstrom\u002Ffast-style-transfer\u002F","github.com","logane",1477930309,2016,184,20,155,2264,7,"notable","ai","neural style transfer","library",[24,25,26,27],"style transfer","tensorflow","neural network","image stylization","curious",0.173,{"positive":31,"neutral":32,"mixed":33,"critical":34},3,6,2,1,12,{"execution":37,"design":39,"performance":40,"idea":41},{"positive":33,"critical":38},0,{"positive":38,"critical":38},{"positive":38,"critical":38},{"positive":34,"critical":34},[43,38,38,31,38,31,31,34,38,38,38,31,38,31,38,31,31,31,43,43],5,null,{"problem":46,"audience":44,"pricing":44,"motivation":44,"time_to_build":44,"stack":54,"openness":64,"stage":65,"team_size":65},{"value":47,"source":48,"evidence":49},"Applies artistic styles to images quickly using a trained model.","commenter",[50],{"id":51,"by":52,"text":53},"12838644","minimaxir","Terminology note: there is a difference between Style Transfer and Fast Style Transfer. Fast Style Transfer takes a very long time to train a style model (and the output models…",[55,59],{"value":56,"source":48,"evidence":57},"TensorFlow",[58],{"id":51,"by":52,"text":53},{"value":60,"source":61,"evidence":62},"Python","inferred",[63],{"id":51,"by":52,"text":53},"open_source","unknown",{"praise":67,"objections":84,"requests":98,"questions":106,"alternatives":121},[68,80],{"value":69,"source":48,"evidence":70},"Impressive results and fast stylization",[71,75,79],{"id":72,"by":73,"text":74},"12840969","rubyfan","This is ridiculously cool.",{"id":76,"by":77,"text":78},"12839624","visarga","Results look pretty nice. Congrats!",{"id":51,"by":52,"text":53},{"value":81,"source":48,"evidence":82},"Easier setup than Torch\u002FLua implementations",[83],{"id":51,"by":52,"text":53},[85,92],{"value":86,"source":48,"evidence":87},"Checkerboard artifacts in output",[88],{"id":89,"by":90,"text":91},"12838683","dharma1","Nice work! Have you seen this - could help reduce some of those checkerboard artifacts http:\u002F\u002Fdistill.pub\u002F2016\u002Fdeconv-checkerboard\u002F",{"value":93,"source":48,"evidence":94},"Larger images produce repeated patterns",[95],{"id":96,"by":52,"text":97},"12839686","Playing around with the tool a bit, supersampling seems to work pretty well to avoid those artifacts. (Thanks to the speed of Fast Style Transfer, this is feasible.) EDIT:…",[99],{"value":100,"source":48,"evidence":101},"Localised brush-based style application",[102],{"id":103,"by":104,"text":105},"12839964","lhnz","It would be nice if this algorithm could be created in a more localised form, so that you could take a photo and apply the effect with different intensities as if you were using…",[107,114],{"value":108,"source":48,"evidence":109},"Can style be amplified beyond the original?",[110],{"id":111,"by":112,"text":113},"12840390","pizza","So you can transfer style from image A to image B. What I want to know is, can you use style transfer to \"amplify\" image A's style to, say, 500%?",{"value":115,"source":48,"evidence":116},"How does it compare with other fast style transfer methods?",[117],{"id":118,"by":119,"text":120},"12839702","spacehacker","Nice work. I'd be curious how this technique compares with other fast style transfer methods, if any exist.",[122,132],{"value":123,"source":48,"evidence":124},"Neural Doodle",[125,129],{"id":126,"by":127,"text":128},"12846130","ebalit","It has been done with manual segmentation [1]. And the results are mind blowing. There also a lot of work done on segmentation with neural nets [2, 3], so I wouldn't be surprised…",{"id":130,"by":127,"text":131},"12846170","As I mentioned in another thread, there is some work on this: https:\u002F\u002Fgithub.com\u002Falexjc\u002Fneural-doodle",{"value":133,"source":48,"evidence":134},"Google TensorFlow style transfer implementation",[135,139],{"id":136,"by":137,"text":138},"12841634","eitally","Relevant Tensorflow\u002FBrain blog post: https:\u002F\u002Fresearch.googleblog.com\u002F2016\u002F10\u002Fsupercharging-style-transfer.html",{"id":51,"by":52,"text":53},{"comments":34,"answered":38,"key":141},[142],{"id":143,"text":144},"12839157","I hadn't seen that, thanks for the link - it looks useful! Right now the kernel size is not divisible by the step size - I'll try changing the transformation network.",{"commenters":146,"max_depth":147,"maker_replies":34,"first_comment_mins":148,"lifespan_hours":149},16,4,51,23,[151,153,157,159,161,163,165,169,173,176,178,181,182,183,187,189,190,193,195,198],[51,38,43,152,52,38],"project\u002Fpositive\u002Fexecution",[154,38,38,155,156,38],"12843220","commenter\u002Fneutral\u002Fperformance","gedy",[158,38,38,155,52,38],"12843328",[89,34,31,160,90,38],"project\u002Fneutral\u002Fdesign",[143,34,38,162,11,34],"maker\u002Fpositive\u002Fdesign",[96,34,31,164,52,38],"project\u002Fmixed\u002Fperformance",[166,33,31,167,168,38],"12839528","project\u002Fmixed\u002Fidea","hacker_9",[170,33,34,171,172,38],"12839918","project\u002Fcritical\u002Fidea","pavlov",[174,33,38,175,112,38],"12840419","commenter\u002Fneutral\u002Fidea",[126,33,38,177,127,38],"other\u002Fneutral\u002Fidea",[179,31,38,177,180,38],"12843837","santaclaus",[103,147,31,160,104,38],[130,147,38,177,127,38],[184,147,31,185,186,38],"12846241","project\u002Fneutral\u002Fidea","_0ffh",[136,43,38,188,137,38],"other\u002Fneutral\u002Fnone",[111,32,31,185,112,38],[191,32,31,185,192,38],"12841065","gbrits",[118,18,31,194,119,38],"project\u002Fneutral\u002Fperformance",[76,196,43,197,77,38],8,"maker\u002Fpositive\u002Fexecution",[72,199,43,200,73,38],9,"project\u002Fpositive\u002Fidea",[],[203,208,213,218,222,227],{"id":204,"name":205,"points":206,"category":207},"12834327","sudo-touchid",347,"security_privacy",{"id":209,"name":210,"points":211,"category":212},"12843466","My first web app",297,"web_saas",{"id":214,"name":215,"points":216,"category":217},"12803860","Zone Messenger",194,"consumer_social",{"id":219,"name":220,"points":221,"category":207},"12802785","Android 'root' phones via dirtyc0w exploit",143,{"id":223,"name":224,"points":225,"category":226},"12842797","KQ",116,"dev_tools",{"id":228,"name":229,"points":230,"category":212},"12848656","Academic Bio",113,false,"v2",{"id":234,"name":5,"line":6,"ai":231,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":235,"top":18,"kw":236,"maker":11,"domain":10,"tier":19,"attempts":38,"strip":237,"url":9,"mix":238,"praise":69,"objection":86,"sub":21},12838152,0.17,[24,25,26,27],"50030331000303033355",[31,32,33,34],1794036549058]