[{"data":1,"prerenderedAt":348},["ShallowReactive",2],{"project-41997648":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":34,"aspects":35,"strip":46,"post_text":49,"what":50,"said":74,"maker_in_thread":189,"shape":191,"grid":194,"attempts":309,"same_week":310,"maker_shows":41,"maker_hidden":340,"reader_version":341,"card":342},"41997648","AI OmniGen","AI Image Generator with Consistent Visuals","title","Show HN: AI OmniGen – AI Image Generator with Consistent Visuals","https:\u002F\u002Faiomnigen.com","aiomnigen.com","lcorinst",1730308243,2024,169,49,481,3515,14,"notable","ai","image generation","web_app",[21,24,25,26,27],"consistent characters","image editing","diffusion model","text to image","debated",0.198,{"positive":18,"neutral":31,"mixed":32,"critical":33},9,0,10,33,{"execution":36,"idea":39,"docs":40,"performance":42,"design":43,"privacy":44,"pricing":45},{"positive":37,"critical":38},2,8,{"positive":38,"critical":32},{"positive":41,"critical":32},1,{"positive":37,"critical":41},{"positive":32,"critical":32},{"positive":32,"critical":41},{"positive":32,"critical":32},[47,32,47,41,41,48,41,41,48,41,48,47,48,47,47,41,47,48,48,48,47,47,48,32,32,32,32,32,47,48,41,32,32,32,32,32,32,47,48,47],5,3,"AI OmniGen is an advanced AI image generator, offering identity preservation for consistent subject representation and seamless image editing for refined, customized visuals.",{"problem":51,"audience":57,"pricing":57,"motivation":57,"time_to_build":57,"stack":58,"openness":71,"stage":72,"team_size":73},{"value":52,"source":53,"evidence":54},"Generating images with consistent subjects and refined edits","maker",[55],{"id":56},"post",null,[59,67],{"value":60,"source":61,"evidence":62},"Transformer (Phi-3)","commenter",[63],{"id":64,"by":65,"text":66},"42001246","nairoz","> trained from scratch Not exactly. They mention starting from the VAE from Stable Diffusion XL and the Transformer from Phi3. Looks like these LLMs can really be used for anything",{"value":68,"source":61,"evidence":69},"VAE from Stable Diffusion XL",[70],{"id":64,"by":65,"text":66},"open_source","launched","unknown",{"praise":75,"objections":98,"requests":132,"questions":147,"alternatives":162},[76,87],{"value":77,"source":61,"evidence":78},"Unified architecture handles many image tasks with strong instruction following",[79,83],{"id":80,"by":81,"text":82},"41999115","ed","Elegant architecture, trained from scratch, excels at image editing. This looks very interesting! From https:\u002F\u002Farxiv.org\u002Fhtml\u002F2409.11340v1 > Unlike popular diffusion models,…",{"id":84,"by":85,"text":86},"42003203","sswz","Using a single model to unify all image generation tasks, including many computer vision tasks and visual language reasoning, could transform future image generation models.…",{"value":88,"source":61,"evidence":89},"Image editing with reference images works well",[90,94],{"id":91,"by":92,"text":93},"42001259","block_dagger","This looks promising. I love how you can reference uploaded images with markup - this is exactly what the field needs more of. After spending the last two weeks generating…",{"id":95,"by":96,"text":97},"41997671","oatsandsugar","I mean, I struggle even getting Dall-E to iterate on one image without changing everything, so this is pretty cool",[99,118,125],{"value":100,"source":61,"evidence":101},"Generation often hangs or produces no output in the hosted demo",[102,106,110,114],{"id":103,"by":104,"text":105},"41999412","lelandfe","I left all the defaults as is, uploaded a small image, typed in \"cafe,\" and 15 minutes later I am still waiting on this finishing.",{"id":107,"by":108,"text":109},"42002564","cubefox","Same, I left running for half an hour but nothing happened.",{"id":111,"by":112,"text":113},"42004671","Citizen_Lame","Left it running 1 hour nothing happens. Maybe this is a social experiment.",{"id":115,"by":116,"text":117},"42006228","freilanzer","Well, at the moment it seems it's not working at all.",{"value":119,"source":61,"evidence":120},"Character consistency is limited; clothing changes between images",[121],{"id":122,"by":123,"text":124},"42006224","jowday","It’s not really consistent - or anymore consistent than, say, SDXL with IP adapter. Even in their example images the character they’ve input comes out wearing different clothes.",{"value":126,"source":61,"evidence":127},"Image quality is not outstanding",[128],{"id":129,"by":130,"text":131},"42001773","CamperBob2","I was able to clone the repo and run it locally, even on a Windows machine, with only minimal Python dependency grief. Takes about a minute to create or edit an image on a 4090.…",[133,140],{"value":134,"source":61,"evidence":135},"Support for transparent images",[136],{"id":137,"by":138,"text":139},"42000556","fullstackwife","not yet, still can't generate transparent images",{"value":141,"source":61,"evidence":142},"Better text rendering in images",[143],{"id":144,"by":145,"text":146},"42004081","gremlinsinc","Anyone know how it handles Text? That's kind of my deal breaker, I like Ideogram for it's ability to do really cool fonts, etc.",[148,155],{"value":149,"source":61,"evidence":150},"What is the cost per edit",[151],{"id":152,"by":153,"text":154},"42000078","anyi09881","Curious what's the actual cost for each edit? Will this infra always be reliable?",{"value":156,"source":61,"evidence":157},"How the multimodal aspect works",[158],{"id":159,"by":160,"text":161},"42032128","hyuuu","could you elaborate on the multi modal aspect of this model?",[163,170,174,178],{"value":164,"source":61,"evidence":165},"ComfyUI",[166],{"id":167,"by":168,"text":169},"42003004","yieldcrv","Pretty cool, comfy ui and community is too cumbersome for me and still results in too much throwaway content",{"value":171,"source":61,"evidence":172},"DALL-E",[173],{"id":91,"by":92,"text":93},{"value":175,"source":61,"evidence":176},"Ideogram",[177],{"id":144,"by":145,"text":146},{"value":179,"source":61,"evidence":180},"Flux via Replicate or Fal.ai",[181,185],{"id":182,"by":183,"text":184},"42001492","ncoronges","The simplest commercial product for finetuning your own model is probably Adobe firefly, although there’s no API access support yet. But there are cheap and only slightly more…",{"id":186,"by":187,"text":188},"42003435","spaceman_2020","The easiest flux api I’ve seen is with Fal.ai It is expensive though- Flux dev images are like $0.035\u002Fimage",{"comments":32,"answered":32,"key":190},[],{"commenters":192,"max_depth":38,"maker_replies":32,"first_comment_mins":37,"lifespan_hours":193},35,89,[195,197,199,201,203,204,207,208,211,213,216,219,223,227,231,234,238,241,242,244,247,248,251,254,257,260,261,263,268,269,271,274,275,277,281,283,287,290,292,294,296,298,300,302,304,307,308],[80,32,47,196,81,32],"project\u002Fpositive\u002Fexecution",[64,32,32,198,65,32],"commenter\u002Fneutral\u002Fnone",[167,32,47,200,168,32],"project\u002Fpositive\u002Fidea",[103,41,41,202,104,32],"project\u002Fcritical\u002Fexecution",[107,41,41,202,108,32],[205,41,47,196,206,32],"42002878","bob_1200",[111,41,41,202,112,32],[209,41,41,202,210,32],"42005977","grvbck",[212,41,41,202,116,32],"42006225",[214,41,48,215,206,32],"42007112","project\u002Fneutral\u002Fdocs",[217,41,41,218,104,32],"42007202","project\u002Fcritical\u002Fperformance",[220,41,48,221,222,32],"42008059","project\u002Fneutral\u002Fexecution","zamadatix",[224,41,48,225,226,32],"42010526","project\u002Fneutral\u002Fperformance","phromo",[228,41,47,229,230,32],"42008457","project\u002Fpositive\u002Fperformance","YetAnotherNick",[232,37,48,233,206,32],"42002891","project\u002Fneutral\u002Fnone",[235,37,47,236,237,32],"42004295","project\u002Fpositive\u002Fnone","littlestymaar",[239,48,47,200,240,32],"41999161","ilaksh",[115,48,41,202,116,32],[84,243,47,200,85,32],4,[245,47,47,200,246,32],"42000520","wwwtyro",[137,47,41,202,138,32],[249,47,48,221,250,32],"42000981","derefr",[252,47,48,221,253,32],"42001207","Vt71fcAqt7",[255,47,47,200,256,32],"42001389","Multicomp",[258,47,47,200,259,32],"42004078","haccount",[122,47,41,202,123,32],[91,262,47,200,92,32],6,[264,265,32,266,267,32],"42001319",7,"other\u002Fneutral\u002Fnone","101008",[182,265,32,266,183,32],[270,265,32,266,267,32],"42002614",[272,265,32,266,273,32],"42002632","handfuloflight",[186,265,32,266,187,32],[276,265,32,266,259,32],"42004110",[278,38,47,279,280,32],"42000081","project\u002Fpositive\u002Fdocs","KerryJones",[144,31,48,282,145,32],"project\u002Fneutral\u002Fdesign",[284,33,41,285,286,32],"41998823","project\u002Fcritical\u002Fprivacy","empath75",[288,33,32,266,289,32],"42005308","hnbad",[291,33,32,266,130,32],"42006660",[293,33,32,266,108,32],"42007212",[295,33,32,266,130,32],"42017786",[297,33,32,266,108,32],"42019388",[299,33,32,266,130,32],"42020560",[301,33,32,198,108,32],"42026097",[95,303,47,200,96,32],11,[152,305,48,306,153,32],12,"project\u002Fneutral\u002Fpricing",[129,305,47,229,130,32],[159,305,48,233,160,32],[],[311,316,321,326,331,335],{"id":312,"name":313,"points":314,"category":315},"41934746","2048",669,"games",{"id":317,"name":318,"points":319,"category":320},"41930628","RF Hunter",594,"hardware",{"id":322,"name":323,"points":324,"category":325},"41945458","Trench",155,"data",{"id":327,"name":328,"points":329,"category":330},"41935166","Animal Body Fat Match",142,"consumer_social",{"id":332,"name":333,"points":334,"category":330},"41980681","Ding Dong Doorbell",140,{"id":336,"name":337,"points":338,"category":339},"41952779","Mdx",127,"dev_tools",false,"v2",{"id":343,"name":5,"line":6,"ai":340,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":344,"top":18,"kw":345,"maker":11,"domain":10,"tier":19,"attempts":32,"strip":346,"url":9,"mix":347,"praise":77,"objection":100,"sub":21},41997648,0.2,[21,24,25,26,27],"5051131131353551533355300000531000000535",[18,31,32,33],1793991104026]