Yadda Ake Amfani da AI Video Prompt: Ashana da Taba Sigari Suna Yin Fada Ashana Ta Yi Nasara
A zamanin yau, fasahar Artificial Intelligence (AI) tana sauya yadda ake ƙirƙirar hotuna da bidiyoyi. Abin da a baya yake buƙatar kyamara, 'yan wasa, wuri na musamman da kayan aiki masu yawa, yanzu ana iya fara ƙirƙirarsa da hoto guda da kuma prompt mai kyau.
Wannan darasin ya bayyana yadda ake amfani da wani hoto mai ban mamaki na ashana da aka haɗa da ƙananan sanduna tare da taba sigari, sannan a mayar da shi bidiyo na kusan seconds 10 inda ashannar da taba sigarin suke tafiya, su fuskanci juna, su yi faɗa, sannan ashannar ta yi nasara.
Manufar wannan bidiyon ita ce ƙirƙirar gajeren labari mai ban sha'awa da zai ja hankalin masu kallo. Ba wai ana ƙarfafa shan taba ba ne; ana amfani da taba a matsayin wani abu na cikin labarin kirkirarren AI.
Menene AI Video Prompt?
AI Video Prompt umarni ne ko bayanin da ake bai wa AI domin ya fahimci irin bidiyon da ake son ƙirƙira. Prompt mai kyau yana bayyana abubuwa kamar wanda yake cikin bidiyon, abin da yake yi, yadda camera za ta motsa, yanayin haske, muhalli da salon bidiyon.
Misali, maimakon ka rubuta kawai:
“Make the cigarette and matchstick fight.”
Za ka iya rubuta prompt mai tsari wanda ya bayyana abin da zai faru daga farkon bidiyon har zuwa ƙarshensa.
Wannan yana da muhimmanci musamman idan kana son bidiyon ya kasance da flow, wato motsi ya kasance cikin tsari daga farko zuwa ƙarshe.
Masu ƙirƙirar AI video suna ba da shawarar a bayyana visual elements da motion elements a sarari, musamman subject action da camera movement.
Me Ya Sa Wannan Prompt Ya Fi Yin Flow?
Babban matsalar da mutane suke fuskanta wajen ƙirƙirar AI video ita ce AI ta iya rikicewa idan aka ba ta abubuwa da yawa a lokaci guda.
Misali, idan ka ce:
“Ashana da taba suna tafiya, suna gudu, suna tsalle, suna yaƙi, camera tana zagaye da su, sai su tashi sama, sai ruwa ya sauka...”
AI na iya kasa fahimtar wanne abu ne ya kamata ya faru da farko.
Saboda haka, wannan prompt ɗin an tsara shi kamar ƙaramin fim:
Farko → tafiya → haɗuwa → fara faɗa → kai hari → kaucewa → nasara → ƙarshen bidiyo.
Wannan tsarin yana taimaka wa AI ta fahimci sequence ɗin abin da ake so.
Yadda Labarin Bidiyon Zai Fara
A farkon seconds biyu, ashannar da taba sigarin suna rayuwa kamar ƙananan halittu.
Suna tafiya a kan ƙasa a hankali suna zuwa juna.
Camera tana nuna su duka domin mai kallo ya fahimci cewa za a samu wani abu.
A wannan lokacin bai kamata a cika bidiyon da abubuwa da yawa ba. Ana buƙatar motsi mai sauƙi da fahimta.
Lokacin Da Suka Fuskanci Juna
Bayan sun kusanci juna, sai su tsaya suna kallon juna.
Wannan ƙaramin lokaci yana da muhimmanci saboda yana ba bidiyon labari. Mai kallo zai fara tunanin cewa faɗa zai fara.
Camera na iya yin slow push-in, wato ta kusanto su a hankali domin ƙara suspense.
Google DeepMind ya bayyana cewa camera framing da camera movement suna daga cikin muhimman abubuwan da za a iya saka cikin Veo prompt domin samun irin kallon da ake so.
Farkon Faɗan
Bayan sun tsaya, taba sigarin ta fara kai hari.
Tana motsa jikinta zuwa wajen ashannar kamar tana ƙoƙarin ture ta.
Ashannar ta kauce da sauri.
A nan ne bidiyon zai fara samun action.
Amma saboda bidiyon seconds 10 ne, bai kamata a cika shi da dogon faɗa ba. Ana buƙatar a mayar da hankali kan wasu ƴan motsi masu muhimmanci.
Wannan ya dace da shawarar masu ƙirƙirar AI video cewa gajeren clip ya fi samun sakamako mai kyau idan an mai da hankali kan babban action ɗaya maimakon haɗa abubuwa da yawa marasa alaƙa.
Yadda Ashana Za Ta Yi Nasara
Bayan taba sigarin ta kai hari, ashannar ta yi saurin kaucewa.
Sai ta zagaya gefe ɗaya.
Daga nan sai ta yi wani motsi mai ƙarfi ta bugi taba sigarin.
Taba sigarin ta rasa daidaito sannan ta faɗi ƙasa.
A ƙarshe ashannar ta tsaya tana nuna cewa ita ce ta yi nasara.
Wannan ƙarshen yana ba bidiyon “payoff”, wato abin da mai kallo yake jira tun daga farkon labarin.
Muhimmancin Timestamp A Prompt
Idan kana son samun flow mai kyau, zaka iya raba seconds 10 ɗin zuwa sassa.
Misali:
0–2 seconds: Ashana da taba suna tafiya zuwa juna.
2–4 seconds: Sun tsaya suna fuskantar juna.
4–7 seconds: Taba sigarin ta fara kai hari, ashana ta kauce.
7–9 seconds: Ashana ta kai hari ta kayar da taba sigarin.
9–10 seconds: Ashana ta tsaya cikin nasara yayin da taba sigarin take ƙasa.
Wannan tsarin yana taimaka wa AI ta fahimci lokacin da kowane action ya kamata ya faru. Wasu jagororin AI video prompting suna nuna cewa timestamp prompts na iya taimakawa wajen samun ƙarin iko akan timing da sequence na motsi.
Prompt Na Bidiyon
Ga prompt ɗin da za ka iya gwadawa:
“Create a 10-second continuous cinematic video using the uploaded image as the exact visual reference. 0–2 seconds: the matchstick figure and the cigarette suddenly come alive and slowly walk toward each other. 2–4 seconds: they stop face-to-face and prepare to fight. 4–7 seconds: the cigarette attacks first, swinging toward the matchstick, but the matchstick quickly dodges and moves around it. 7–9 seconds: the matchstick jumps forward and strikes the cigarette, knocking it down. 9–10 seconds: the matchstick stands proudly beside the fallen cigarette as smoke rises. Keep the same characters, same environment and consistent appearance throughout. Smooth continuous motion, realistic physics, cinematic camera movement, dramatic lighting, detailed smoke, ultra-realistic 4K quality.”
Yadda Za Ka Saka Hoton
Idan kana amfani da image-to-video, ka fara da hoton da kake so AI ta yi amfani da shi a matsayin reference.
Wannan yana da amfani saboda hoton yana ba AI bayani game da yanayin scene, abubuwan da ke ciki da appearance ɗin abubuwan.
A wannan misalin, hoton yana nuna ashannar da sandunan da aka tsara kamar mutum, tare da taba sigari.
Sai prompt ya fi mayar da hankali kan motsi da camera, maimakon sake bayyana komai daga farko.
Wannan tsarin ya yi daidai da jagororin image-to-video prompting waɗanda ke nuna cewa idan image ya riga ya bayyana subject da scene, prompt zai iya fi mayar da hankali kan motion da camera movement.
Google Flow
Idan kana son gwada wannan irin aikin, zaka iya amfani da Google Flow, wanda Google ya ƙirƙira a matsayin AI filmmaking tool.
Flow yana amfani da samfuran Google kamar Veo wajen ƙirƙirar cinematic clips da scenes. A halin yanzu, Google Flow yana da nau'ikan Veo daban-daban da kuma wasu hanyoyin ƙirƙirar bidiyo daga text ko image/reference, gwargwadon model da account ɗinka.
Shafin Google Flow:
Google Flow – AI Creative Studio
Google Flow Help:
Google Flow Help – Get Started
Muhimmin Abu Game da Flow
Kafin ka fara, ka duba model ɗin da aka zaɓa a Flow saboda tsawon bidiyo da wasu capabilities na iya bambanta tsakanin models da subscription tiers. Google Help ya bayyana cewa wasu Veo 3.1 modes suna tallafawa 4, 6, 8 ko 10-second videos, yayin da wasu features suke da takamaiman tsawo ko requirements.
Saboda haka idan ba ka ga zaɓin seconds 10 ba, kada ka ɗauka cewa prompt ɗin ne matsala. Zai iya kasancewa model ɗin da ka zaɓa ne bai tallafa wa wannan tsawon bidiyon ba.
Tips Domin Samun Bidiyo Mai Inganci
Abu na farko shi ne ka yi amfani da clear action. Ka bayyana abin da characters za su yi.
Na biyu, ka bayyana camera movement. Misali:
“slow tracking shot”
ko
“slow push-in”
ko
“low-angle cinematic shot.”
Na uku, ka kula da lighting. Misali:
“soft cinematic lighting”
“dramatic natural lighting”
ko
“moody cinematic atmosphere.”
Na huɗu, ka bayyana style kamar:
“ultra-realistic cinematic video”
“miniature realistic film”
“4K cinematic quality.”
Na biyar, kada ka cika prompt da umarni masu karo da juna. Jagororin AI video prompting suna nuna cewa prompts masu tsari da bayyanannun umarni suna taimakawa wajen samun sakamako mai daidaito.
Idan Bidiyon Bai Yi Daidai Ba
Idan ka generate bidiyo kuma bai yi yadda kake so ba, kada ka canza komai a prompt lokaci guda.
Ka fara da gyara abu ɗaya.
Misali, idan ashannar ba ta yi nasara ba, ka ƙara bayyana ƙarshen:
“the matchstick delivers one clear final strike and the cigarette falls to the ground.”
Idan camera ta yi motsi da yawa, ka rage camera instruction.
Idan characters sun canza shape, ka ƙara bayani game da consistency da kuma amfani da hoton reference.
Wannan hanyar yin ƙananan gyare-gyare ɗaya bayan ɗaya ana ba da shawarar ta wajen jagororin prompt na AI video domin fahimtar abin da ke aiki da abin da baya aiki.
Kammalawa
Wannan ra'ayin na ashana da taba sigari yana nuna yadda za a iya ɗaukar wani hoto mai sauƙi a mayar da shi ƙaramin labarin cinematic ta amfani da AI.
Sirrin samun kyakkyawan sakamako ba wai kawai amfani da kalmomin “4K” ko “cinematic” ba ne. Abu mafi muhimmanci shi ne ka bayyana abin da zai faru, wanda zai yi shi, yadda zai motsa, yadda camera za ta nuna shi, da kuma yadda labarin zai ƙare.
Idan bidiyon seconds 10 ne, ka raba shi cikin flow mai sauƙi: tafiya, haɗuwa, fara faɗa, kai hari, nasara da ƙarshen scene.
Google kuma yana ci gaba da ƙara capabilities a Flow da Veo, ciki har da ingantaccen realism, prompt adherence, editing da audio a wasu features.
Don haka idan kana son ƙirƙirar irin waɗannan gajerun bidiyoyi masu ban mamaki, ka fara da hoto mai kyau, sannan ka yi amfani da prompt mai tsari da bayyananniyar sequence.
Ƙarin hanyoyin koyon AI Video Prompting:
Google DeepMind – Veo Prompt Guide
0 Comments