Samfuran Yadawa Latent
Samfuran ɓoyayyiyar ɓoye suna haifar da hotuna ta hanyar tafiyar da tsarin watsawa a cikin madaidaicin sarari maimakon ɗanyen pixels, rage ƙididdige farashi.
Dubawa
They are the engine behind Stable Diffusion and most modern open-source image generators.
Zurfafa nutsewa
Madaidaicin samfurin watsawa yana koyon jujjuya tsarin surutu: yana farawa daga tsantsar amo kuma a hankali yana musantawa zuwa hoto. Yin wannan kai tsaye akan pixels yana da tsada saboda hoton 512x512 yana da ɗaruruwan dubunnan dabi'u. Rushewar ɓoye, wanda Rombach da abokan aiki suka gabatar a cikin 2022, da farko yana amfani da pretrained variational autoencoder (VAE) don matsa hoto a cikin ƙaramin grid mai ɓoye (sau da yawa 64x64x4, kusan 48x ƙarami). U-Net ɗin yaɗuwa sannan ya koyi yin magana a cikin wannan ƙaramin sarari na ɓoye, wanda rubutu ke jagoranta ta hanyar jan hankali. A ƙarshe mai gyara VAE yana sake gina pixels masu cikakken ƙuduri. Wannan matsi na fahimta yana kiyaye bayanan ma'ana mai ma'ana yayin watsar da dalla-dalla da ba za a iya fahimta ba, yana samar da ingantaccen tsararru mai yuwuwa akan GPUs masu amfani.
Fahimtar Fasaha
Dabarar maɓalli ita ce raba matsi na fahimta daga ƙirar ƙira. VAE tana ɗaukar dalla-dalla babban mitar pixel sau ɗaya, kuma U-Net kawai ke ƙirƙira ƙananan rarrabawar ɓarna. Ana yin alluran sanyaya rubutu ta hanyar jan hankali, inda fasalulluka na U-Net ke halartar abubuwan saka alama daga mai rikodin rubutu kamar CLIP. Saboda latents sun fi ƙanƙanta sau 48 fiye da pixels, kowane matakin ƙin yarda yana da rahusa sosai a cikin ƙwaƙwalwar ajiya da FLOPs.
Dabarun Tasiri
Gudu da sikelin
Kayayyakin AI na iya sarrafa aiki da bincike, ganowa, da ayyuka masu alama a sikelin.
Gina zaɓuɓɓuka
Ƙungiyoyin ƙirƙira za su iya samar da ra'ayoyi cikin sauri tare da ƙarancin bita da hannu.
Ƙungiya da aikin aiki
Ayyuka na iya amfani da siginar hoto da bidiyo waɗanda a baya suke da wahalar aiwatarwa.
Makomar Samfuran Yaɗawa Latent
Rushewar ɓarna yana faɗaɗa sama da hotuna zuwa bidiyo (Stable Video Diffusion), kadarorin 3D, da spectrograms mai jiwuwa, duk suna amfani da girke-girke iri ɗaya damfara-sannan-denoise. Bincike yana turawa zuwa ƙananan matakai na samfur ta hanyar distillation da ƙirar daidaito, mafi kyawun VAEs waɗanda ke adana rubutu mai kyau da fuskoki, da kuma tsarin gyare-gyare masu gudana kamar waɗanda ke cikin Stable Diffusion 3 waɗanda ke daidaita yanayin tsara don sauri, sakamako mai fa'ida.
Aiwatar da Gaskiyar Duniya
Stable Diffusion yana haifar da zane-zane da ƙirar ra'ayi daga faɗakarwar rubutu akan GPU ɗin mabukaci guda ɗaya
Adobe da Canva suna ba da ƙarfin rubutu-zuwa-hoto da fasalulluka-cika abubuwan da aka gina akan kasusuwan ɓoyayyen ɓoyayyiyar baya.
Studios na wasan suna samar da taswirorin rubutu, sprites, da fasahar ra'ayi na muhalli don haɓaka riga-kafi
Ƙungiyoyin tallace-tallace da tallace-tallace suna ƙirƙira abubuwan izgili da samfuran talla ba tare da hoton hoto ba.
Hatsari & Tsare-tsare
Haƙƙoƙin hoto da yarda na iya zama haxarin doka idan ba a fayyace ba.
Ayyukan samfuri na iya bambanta a ko'ina cikin haske, ƙididdiga, da mahalli.
Ƙarya tabbataccen ƙila ba za a iya lura da shi ba sai dai idan an kula da ƙofofin amincewa.
Taswirar Hanya
Ƙayyade ma'auni na karɓa don daidaito, tunowa, da farashi na kuskure.
Gwada tare da bayanan da suka dace da ainihin yanayin samarwa.
Ƙara bita na ɗan adam don ƙarancin amincewa ko tsinkaya mai tasiri.
Bi diddigin ƙirar ƙira kuma sake ingantawa bayan canje-canjen kamara ko saitin bayanai.
Ci gaba da Bincike
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Latent Diffusion Models quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Jagora na gaba
Samfuran Yadawa Bidiyo
Tambayoyin da ake yawan yi
What is Latent Diffusion Models?
Samfuran ɓoyayyiyar ɓoye suna haifar da hotuna ta hanyar tafiyar da tsarin watsawa a cikin madaidaicin sarari maimakon ɗanyen pixels, rage ƙididdige farashi. Su ne injin da ke bayan Stable Diffusion da galibin masu samar da hoto na zamani.
Menene babban ƙirƙira na ƙirar yaɗuwar ɓoye idan aka kwatanta da yaduwar sararin samaniya?
Rushewar ɓoyayyiyar ɓoyayyiya tana matsa hotuna zuwa ƙaramin sarari mai ɓoye ta amfani da VAE, don haka ƙima mai tsada yana faruwa akan ƙima kaɗan fiye da cikakkun pixels.
Wane bangare ne ke danne hoto a cikin sararin samaniya kuma ya sake gina shi daga baya?
VAE da aka riga aka horar da ita tana ɓoye hoton zuwa cikin ƙaramin grid mai ɓoyewa kuma mai yin rikodin sa yana sake gina pixels masu cikakken ƙuduri a ƙarshe.
Ta yaya aka saba allurar rubutu a cikin U-Net mai karyatawa a cikin ɓoyayyen ɓoye?
Ana ciyar da abubuwan shigar da alamar rubutu daga mai rikodin rubutu zuwa yadudduka mai jan hankali, barin fasalulluka na sarari su halarci saƙon.
Me yasa yin aiki a sararin samaniya yana rage farashin lissafi?
Jira - madaidaicin dalili shine grid na latent ya fi ƙanƙanta (sau da yawa ~ 48x) fiye da hoton pixel, don haka kowane mataki na ƙin yarda yana buƙatar ƙarancin FLOPs da ƙwaƙwalwar ajiya.
Wanne samfurin buɗaɗɗen tushe da aka yi amfani da shi sosai ya shahara da yaɗuwar ɓoye?
Stable Diffusion, wanda aka saki a cikin 2022, ya kawo ɓarna a ɓoye ga kayan masarufi da karɓar taro.