Dala Amavidiyo e-AI nge-Bang Story: Kusukela Embonweni Wendaba Kuya Ku-MP4 Eqediwe

Yini i-Bang Story?

I-Bang Story uhlelo lokusebenza lwekhompyutha oluwumthombo ovulekile oluguqula umqondo wendaba ube ukukhiqizwa kwevidiyo okuphelele. Uhlinzeka ngomqondo, bese uhlelo lukusiza ukudala umbhalo ohlelekile wesigcawu nesigcawu, iziyalezo ezibukwayo, izithombe noma amavidiyo e-AI, ukulandisa, amagama-ncazo avumelanisiwe, izinguquko, umculo, nama-overlay.

Iphrojekthi yokugcina ihlanganiswa kumhleli womugqa wesikhathi bese yenziwa ibe yi-MP4 nge-FFmpeg. Amaphrojekthi nemidiya ekhiqiziwe kugcinwa kukhompyutha yakho, kanti izicelo ezidinga i-AI zithunyelwa ngqo kubahlinzeki obalungiselela bona usebenzisa okhiye bakho be-API.

I-Bang Story imahhala ngaphansi kwelayisensi ye-MIT, kodwa izinsizakalo zangaphandle ze-AI zingakhokhisa ukukhiqizwa kwezithombe, ukukhiqizwa kwamavidiyo, amamodeli olimi, noma ukulandisa.

Izici Eziyinhloko

• Ukugeleza komqondo kuya kumbhalo: Khiqiza indaba ehlelekile nombhalo wokulandisa wezigcawu ngazinye.
• Izilungiselelo zephrojekthi eziguquguqukayo: Khetha ubude obusukela kumasekhondi angu-30 kuye emizuzwini engu-10, ulimi, ifomethi engu-16:9 noma engu-9:16, umlandi, kanye nesitayela sokubuka esingu-2D, 3D, noma esingokoqobo.
• Ukukhiqizwa kwe-storyboard nge-AI: Khiqiza iziyalezo ezibukwayo, amashidi abalingiswa, iziqondiso zekhamera, izinguquko, nemiyalelo yokunyakaza.
• Ukuhlanganiswa kwe-Higgsfield: Khetha kumamodeli ezithombe angu-14 namamodeli evidiyo angu-34, ngokusetshenziswa kwezikweletu okulinganisiwe okuboniswa ngaphambi kokukhiqiza.
• Ukulandisa okuqhubekayo: Khiqiza ukuqoshwa okukodwa okuqhubekayo nge-Gemini TTS noma i-ElevenLabs bese ukuhlukanisa ngokuzenzakalelayo ngokwesigcawu.
• Ukuhlela umugqa wesikhathi: Nciphisa, sika, susa, futhi uhambise iziqeshana, ukulandisa, amagama-ncazo, umculo, nama-overlay.
• Amagama-ncazo anembile: Sebenzisa i-Whisper transcription yasendaweni ukuze uthole amagama-ncazo esitayela se-karaoke avumelaniswe nokulandisa okukhiqiziwe.
• Ukuthekelisa kwasendaweni: Yenza amafayela e-MP4 anokulungiswa okungu-720p, 1080p, noma 4K ngama-fps angu-30 noma angu-60.

Izidingo

Uhlelo oluhlanganisiwe nokusekelwa kwamagama-ncazo anembile kuhloselwe i-Windows 10 noma i-Windows 11 engu-64-bit. Ikhodi yomthombo ingasebenza naku-macOS, nakuba i-macOS ingakahlolwa ngokuphelele futhi isifaki nokusethwa kwamagama-ncazo e-Whisper okwamanje kutholakala ku-Windows kuphela.

Uzodinga:

• Ukhiye we-Higgsfield API wezithombe ze-AI, amashidi abalingiswa, namavidiyo.
• Umhlinzeki oyedwa wokukhiqiza izindaba: i-Google Gemini, i-OpenRouter, i-Groq, noma i-endpoint yangokwezifiso ehambisana ne-OpenAI.
• Umhlinzeki wokulandisa: i-Google Gemini ye-Gemini TTS, noma ngokuzikhethela i-ElevenLabs.
• I-Node.js 22 noma entsha uma usebenzisa uhlelo oluvela kukhodi yomthombo.
• Uxhumano lwe-inthanethi lokufakwa kokuqala nezicelo ze-AI zamafu.

Faka i-Bang Story

Inketho 1: Sebenzisa i-START Script

Indlela elula yokusetha idinga ukukhishwa kwe-LTS kwe-Node.js. Ngemva kokulanda noma uku-clone-a indawo yokugcina, qalisa iskripthi sesistimu yakho yokusebenza.

• Windows: Chofoza kabili START - WIN.bat.
• macOS: Chofoza kabili START - MAC.command.

Uma iskripthi se-macOS singenayo imvume, vula i-Terminal enkombeni yephrojekthi bese usebenzisa:

chmod +x "START - MAC.command"

Uma i-macOS isaqhubeka nokuyivimba, chofoza ifayela kwesokudla bese ukhetha Open. Ekusebenzeni kwayo kokuqala, iskripthi se-START sifaka okuncikile kwe-npm, silanda i-Electron, sakhe uhlelo, bese siluvula. Lokhu kungathatha imizuzu eminingana.

Ukuqalisa okulandelayo kusebenzisa kabusha ukwakhiwa okukhiqiziwe. Uma uguqula amafayela angaphansi kwe-src, susa inkomba ye-out noma wakhe kabusha ngaphambi kokusebenzisa i-START futhi.

Inketho 2: Sebenzisa Kusuka Kukhodi Yomthombo

Abathuthukisi bangafaka okuncikile bese baqalisa uhlelo ngokulayisha kabusha okusheshayo:

npm install
npm run dev

Imiyalo ewusizo yokuthuthukisa ihlanganisa:

npm run typecheck
npm run build
npm run dist

npm run typecheck yenza ukuhlolwa okuqinile kwe-TypeScript, npm run build ibhala ukwakhiwa kohlelo ku-out, kanti npm run dist idala isifaki se-Windows ku-dist.

Yakha Isifaki Se-Windows

npm run dist

Isifaki esiphumayo siqanjwe ngokuthi dist/BangStory-Setup-<version>.exe. Ukuze udale inkomba ye-Windows ephathekayo esikhundleni sesifaki, sebenzisa:

npx electron-builder --win --dir

Lungiselela Okhiye Bakho Be-API

Vula Settings › AI services and keys ngemva kokuqala uhlelo. Isigaba ngasinye sinokukhetha umhlinzeki, futhi kusetshenziswa umhlinzeki okhethiwe kuphela.

1. Faka ukhiye we-Higgsfield owuthole ngenkinobho ethi Copy API key. Ifomethi elindelekile ithi KEY_ID:KEY_SECRET.
2. Khetha i-Google Gemini, i-OpenRouter, i-Groq, noma i-endpoint yangokwezifiso ehambisanayo ukuze kukhiqizwe izindaba.
3. Khetha i-Gemini TTS noma i-ElevenLabs ukuze kulandiswe, bese unikeza imininingwane ehambisanayo.
4. Gcina izilungiselelo, bese ukhetha imodeli ohlwini lwamamodeli oluseshekayo olulayishwe kumhlinzeki wakho.

Ama-endpoint angokwezifiso angaxhuma kunoma iyiphi iseva ehambisana ne-OpenAI API. Izibonelo ezishiwo yiphrojekthi zihlanganisa i-Ollama ku-http://localhost:11434/v1, i-LM Studio, i-vLLM, i-DeepSeek, kanye ne-xAI Grok ku-https://api.x.ai/v1.

Ku-Windows, okhiye babethelwa nge-Electron safeStorage kusetshenziswa i-Windows DPAPI. Bathunyelwa kuphela ezinsizakalweni ezihlobene nemisebenzi yabo.

Dala Ividiyo Yakho Yokuqala

1. Faka Umqondo Wendaba

Esikrinini sasekhaya, faka umqondo omfushane. Isibonelo:

Bekungaba njani ukuphila kwansuku zonke e-Batavia ngeminyaka yawo-1700?

Chofoza u-Start ukuze udale iphrojekthi. Umqondo ocacile onesihloko, indawo, isikhathi, kanye nethoni ehlosiwe unikeza umkhiqizi wendaba isiqondiso esiwusizo.

2. Khetha Izilungiselelo Zephrojekthi

Setha ubude obuhlosiwe, ulimi, isilinganiso sobubanzi nobude, umxoxi, kanye nesitayela sokubukeka. Sebenzisa u-16:9 kumavidiyo avamile e-YouTube noma u-9:16 kuma-Shorts, ama-Reels, kanye ne-TikTok.

Ungakhetha futhi amamodeli ezithombe namavidiyo e-Higgsfield ephrojekthini. Hlola isilinganiso samakhredithi esibonisiwe ngaphambi kokukhiqiza imidiya ukuze uqonde izindleko ezingaba khona zesevisi yangaphandle.

3. Buyekeza Umbhalo

I-Bang Story ikhiqiza ukwakheka kwendaba nombhalo wezwi eliqoshiwe wesigcawu ngasinye. Ungaphathi lesi sigaba njengesinyathelo sokuvuma ngokuchofoza kanye. Hlola amaqiniso, ukuhamba kwesikhathi, ukuphindaphinda, ukuphimisela, nokuthi isigcawu ngasinye siyawufaka yini umnikelo emqondweni oyinhloko.

Ungahlela, wengeze, ususe, futhi uhlele kabusha izigcawu. Ukulandisa kungafaka amathegi enkulumo asekelwayo afana no-<short pause>, <chuckles>, noma <whispers> ukuze kuqondiswe indlela yokukhuluma.

4. Khiqiza I-storyboard

Ngemva kokuqedela umbhalo, qhubekela ku-storyboard. I-Bang Story ilungisa iziyalezo zokubukeka, ulwazi lwabalingisi, amashidi abalingiswa, ukunyakaza kwekhamera, izinguquko, kanye neziqondiso zokunyakaza zesigcawu ngasinye.

Buyekeza iziyalezo ezikhiqiziwe ngaphambi kokusebenzisa amakhredithi. Izincazelo zabalingiswa kufanele zihlale zifana kuzo zonke izigcawu, futhi iziyalezo zokunyakaza kufanele zihambisane nesigcawu kunokuba zengeze ukunyakaza njengomhlobiso nje.

Khiqiza izithombe namavidiyo adingekayo nge-Higgsfield. Amamodeli azenzakalelayo abhalwe phansi yiphrojekthi yi-GPT Image 2.5 Sunburst yezithombe kanye ne-Kling 3.0 Standard yamavidiyo, kanti amanye amamodeli asekelwayo angakhethwa ngephrojekthi ngayinye.

5. Khiqiza Ukulandisa

Khetha i-Gemini TTS noma i-ElevenLabs ukuze udale ukulandisa. I-Bang Story irekhoda ukulandisa njengokuthatha okukodwa okuqhubekayo bese ikuhlukanisa kube izigcawu. Le ndlela isiza ukugcina izwi nephimbo kungaguquki kuyo yonke iphrojekthi.

I-ElevenLabs inikeza isikhathi esinembile segama ngalinye. I-Gemini TTS ayinikezi isikhathi sezinga lamagama, ngakho isikhathi samagama-ncazo silinganiselwa kuphela ngaphandle kokuthi usebenzise ukuvumelanisa kwe-Whisper kwasendaweni.

6. Hlela I-timeline

Sebenzisa umhleli ukuhlela nokucwenga ukukhiqizwa. Hudula unqenqema lwesiqeshana ukuze ushintshe ubude baso, cindezela u-S ukuze usike endaweni ekhethiwe, bese ucindezela u-Delete ukuze ususe okokusebenza okukhethiwe. Ukulandisa, amagama-ncazo, umculo, kanye nezendlalelo kungasuswa noma kuhanjiswe ngokuzimela.

Izinguquko zilungiswa endaweni yokuhlangana kwesiqeshana ngasinye futhi zingabukwa ngokushesha. Ukunyakaza kwekhamera kungasetshenziswa kuzo zombili iziqeshana zezithombe ezimile namavidiyo.

Engeza umculo ongemuva lapho kufanele khona. I-Bang Story ingehlisa ngokuzenzakalela ivolumu yomculo ngesikhathi sokudlala ukulandisa. Iphinde ibe nomkhiqizi weziyalezo zomculo we-Suno noma enye isevisi yomculo ye-AI, kunokuba ithi ikhiqiza umculo uqobo.

Amalogo, ama-watermark, kanye nezendlalelo zombhalo kungabekwa ngqo ekubukeni kuqala. Gcina izendlalelo ngaphakathi kwemingcele ephephile, ikakhulukazi lapho ulungiselela amavidiyo aqondile ezixhumaneni ezivala izingxenye zesikrini.

7. Yenza Isitayela Futhi Uvumelanise Amagama-ncazo

Umhleli unikeza amagama-ncazo esitayela se-karaoke anamafonti angu-15 nezilawuli zombala, usayizi, kanye namabhokisi angemuva.

Kumaphrojekthi e-Gemini TTS, landa imodeli ye-Whisper Small ecishe ibe ngu-488 MB ku-Settings › Accurate captions. I-Whisper isebenza kwasendaweni futhi ingavumelanisa amagama-ncazo nenkulumo yangempela ngomehluko ojwayelekile ongaba imizuzwana engu-0.1–0.2. Iziqeshana ezikhona zingavumelaniswa futhi kuthebhu ethi Caption yomhleli.

I-ElevenLabs isivele inikeza isikhathi segama ngalinye, ngakho ayidingi i-Whisper ngale njongo.

8. Khipha Ividiyo

Khetha ukulungiswa kokuphumayo okungu-720p, 1080p, noma 4K kanye nezinga lamafreyimu elingu-30 noma 60 fps. I-Bang Story inikezela nge-MP4 kwasendaweni isebenzisa i-FFmpeg futhi ihlose ukufana nokubuka kuqala komhleli.

Amagama-ncazo angashiswa angene kuvidiyo, futhi angagcinwa njengefayela elihlukile le-.srt. Umsindo ulungiswa ube ngu-−14 LUFS, izinga iphrojekthi elikhomba njengendinganiso ye-YouTube.

Amaphrojekthi Agcinwa Kuphi

Kuma-Windows, i-Bang Story igcina isizindalwazi sayo se-SQLite nemidiya ngaphansi kwe:

%APPDATA%\Bang Story

Ifayela le-studio.db liqukethe idatha yephrojekthi, kanti ifolda ye-projects iqukethe izithombe, amavidiyo, nomsindo. Amaphrojekthi agcinwa ngokuzenzakalela, futhi ungaphoqa ukugcinwa ngokucindezela u-Ctrl+S. Idatha evela egameni langaphambilini lohlelo elithi Studio Cerita ithuthwa ngokuzenzakalela.

Amathiphu Athuthukile

Lawula Izindleko Ngaphambi Kokukhiqiza

Qedela imibhalo neziyalezo zokubukeka ngaphambi kokudala izithombe noma amavidiyo. Izilinganiso zamakhredithi e-Higgsfield zivela ku-endpoint yayo ye-estimate, engawasebenzisi amakhredithi, ngakho sebenzisa lezi zilinganiso lapho uqhathanisa amamodeli.

Thuthukisa Ukuhamba Kwesikhathi Esigabeni Sombhalo

Qondanisa inombolo nobude bezigcawu nobude obukhethiwe. Ividiyo emfushane emile izuza ekulandiseni okufingqiwe nasekushintsheni okusheshayo kwezithombe, kanti ividiyo ende echazayo ingasekela izigcawu ezihamba kancane nemininingwane eyengeziwe yomongo.

Gcina Ukuhambisana Kwezinhlamvu

Hlola imininingwane yabalingisi abakhiqiziwe namashidi ezinhlamvu ngaphambi kokukhiqiza yonke imidiya yezigcawu. Ukulungisa incazelo yomlingiswa kusenesikhathi kusebenza kahle kunokukhiqiza kabusha iziqeshana ezimbalwa ezingahambisani.

Sebenzisa Amamodeli Endawo Ukuhlela Indaba

Uma usebenzisa iseva yasendaweni ehambisana ne-OpenAI, yihlele njengendawo yangokwezifiso yokukhiqiza izindaba. I-Ollama, i-LM Studio, ne-vLLM kubhalwe njengezibonelo ezihambisanayo. I-Higgsfield isadingeka ekusebenzeni kohlelo lokusebenza lokukhiqiza izithombe, amashidi ezinhlamvu, namavidiyo.

Qinisekisa Izinguquko Zokubuka Kuqala Nezokuthekelisa

Lapho uguqula ikhodi yomthombo, khumbula ukuthi imithetho eyabiwe ngaphansi kwe-src/shared isetshenziswa ekuziphatheni kokubuka kuqala nokuthekelisa. Abanikeli kufanele baqhube zombili lezi ziqinisekiso ngaphambi kokuthumela izinguquko:

npm run typecheck
npm run build

Indawo yokugcina iphinde icele abanikeli ukuthi bahlole izinguquko ezibonakalayo noma zokuthekelisa besebenzisa ithuluzi layo lokuthuthukisa, futhi babhale izinguquko ezifanele ku-changelog.

Qonda Okusebenza Endaweni

Ukugcinwa kwephrojekthi, ukunikezwa kwe-FFmpeg, nokuqashelwa kwamagama-ncazo kwe-Whisper kusebenza kukhompyutha yakho. Izicelo zendaba, imidiya, nokulandisa zithunyelwa kubahlinzeki be-AI obakhethayo. Lo mehluko uwusizo lapho uhlola ubumfihlo, izidingo zokuxhumeka, nezindleko zezinsizakalo.

Isiphetho

IBang Story ihlanganisa ukubhala okusizwa yi-AI nokukhiqizwa kwemidiya nokuhamba komsebenzi komugqa wesikhathi okuvamile. Inqubo ethembeke kakhulu iwukuhlela abahlinzeki kuqala, ukulungisa umbhalo ngaphambi kokukhiqiza imidiya ekhokhelwayo, ukuqinisekisa ibhodi lezindaba ukuze kugcinwe ukuhambisana kokubonwayo, ukuvumelanisa ukulandisa namagama-ncazo, bese ubuka kuqala wonke umugqa wesikhathi ngaphambi kokuthekelisa.

Ukuze uthole olunye uhambo oluqondisayo, bukela isifundo sevidiyo esisemthethweni se-Bang Story. Siyatholakala futhi ngaphansi kokuthi Settings › About ngaphakathi kohlelo lokusebenza.