Yapay Zeka ile Logline Nasıl Yazılır: Üret, Ele, Yeniden Yaz
Yapay zeka ile logline nasıl yazılır: 20 seçenek üretin, sıkı eleyin, üçünü tutun ve son cümleyi elle yazın. Birebir prompt'larla ve yapay zeka videosunun gerçekten çekebileceği şeyler için bir filtreyle.
Bu derste öğrenecekleriniz
- Logline'ın neden sonraki her kararın pusulası olduğu
- Kısa film için küçültülmüş basit bir logline formülü
- Yapay zekayla 20 logline üretme prompt'u ve kısıtların neden önemli olduğu
- Yapay zeka logline'larını üçe indirmek için üç soru
- Yapay zekayı bir eleştirmen olarak kullanmak, sonra son versiyonu kendiniz yazmak
- Yapay zekaya özgü bir filtre: video modelleri onu gerçekten çekebilir mi?
Logline, filminizin neyi anlattığını söyleyen tek cümledir ve yapay zeka ile film yapımında ardından gelen her plana o karar verir. Bu derste bir tohumdan, aklınızdan çıkaramadığınız birkaç görüntüden yola çıkarsınız ve yapay zekayı bir yazım ortağı olarak kullanıp türler, tonlar ve bakış açıları arasında 20 logline üretirsiniz. Ardından onları üç sert soruyla elersiniz, klişelerin üstünü çizersiniz, üçünü tutarsınız ve yapay zekayı seyircinin ne bekleyeceğini soran bir eleştirmene dönüştürürsünüz.
Son logline elle yazılır, bir arkadaşınız üzerinde test edilir ve bir filtreden daha geçirilir: video modelleri onu gerçekten çekebilir mi? Kurs projesi bunun nedenini gösterir: The Photo'nun ilk versiyonlarından birinde Thomas bir zarftan bir anahtar çıkarıyordu; modellerin sürekli başaramadığı küçük bir jestti, bu yüzden fikir yeniden yazıldı.
Tam transkript
İngilizce anlatımın transkripti.
0:04Every film starts as a vague feeling. Siblings. An old photograph. A car in the rain. You can feel there's a film in there, but you can't say what it is yet. And that's the real problem. Not writing. Deciding what your film is about. That decision fits in one sentence, and it's called a logline. In this lecture, you'll use AI to explore dozens of options in minutes, and then choose like a writer.
0:34So why one sentence? Because a logline is your compass. It tells you who the film is about, what they want, what stands in the way, and what's at stake. Every later decision checks against it. The look, the cast, every single shot. If you can't say your film in one sentence, your shots won't know what they're for.
0:58Here's a simple formula. When something happens, a specific character must do something, or else they lose something. For a short film, think smaller. One character, one moment, one change. Here's the logline of our course project. Parked in the rain, on a quiet street, a man hands his sister the one photograph that proves they were once close, hoping it will say what he never could.
1:26Now, where does AI come in? Start with a seed, the few images you can't stop thinking about. Then ask the AI for 20 loglines, each from a different angle. A drama, a thriller, a comedy, a ghost story, the same scene told from the sister's point of view. You're not looking for the answer yet. You're mapping the territory, faster than you ever could alone.
1:52Here's the kind of prompt I use. Here is my seed, a brother and a sister, an old photograph, a parked car at night, rain. Give me 20 loglines for a three -minute short film. Vary the genre, the tone, and the point of view. One sentence each, under 30 words. Notice the constraints. A length, a format, a range. Constraints are what turn AI output from noise into options.
2:22Then you filter hard. Three questions. Can it be shot with very few characters and locations? Does someone change by the end? And is it yours? Does it touch something true for you? Be ruthless with the generic ones. The first answers of a language model are the average of everything it has read. That average has a name. Cliche. Keep three loglines. Cross out the rest.
2:52Let's try it on three of the AI's answers. A grieving brother discovers a magical photograph that lets him travel back in time. That's a cliche, and it needs effects and locations you could never keep consistent. Cross it out. A sister learns her brother sold their mother's house. Strong conflict, but it's told, not shown. Keep it as a maybe. A man hands his sister an old photo in a parked car. Small, visual, and something changes. That one goes to the top.
3:26Next, turn the AI into a critic. Paste your three loglines and ask, What's the cliche here? What would the audience expect, and how could we surprise them? Which one has the strongest ending? A good sparring partner doesn't write for you. It asks the questions you are avoiding.
3:46And then you write the final version yourself. By hand. Read it out loud. Look for the one specific detail that makes it yours. Two children on a beach, seen from behind, is worth more than ten adjectives. Never paste the machine's sentence as your final logline. Your voice is the only thing in this process that no model can generate for you.
4:10Then test it on someone. Read your logline once to a friend. Then ask two questions. What is the film about, and how do you think it ends? If they can't picture a single image, add a concrete detail. If they can guess the ending, you've just found your next problem to solve.
4:29One more filter, and this one is specific to AI filmmaking. Can the models actually shoot it? Every character needs a reference sheet. Every location has to stay consistent. Every object a character handles is a risk. In an early version of our project, Thomas pulled a key out of an envelope. The video models kept failing at that small gesture. So we rewrote the idea. The photograph is already in his hand. Write for what the camera can deliver.
5:00Keep a constraints list right under your logline. The number of characters, locations and key props, and anything risky to shoot. For the photo, two characters, one car, one photograph. No crowd, no action, one line of dialogue. That list will follow you through the entire course.
5:22Here's your exercise. Write down your seed. Generate 20 loglines with AI. Filter them down to three. Let the AI attack them, and write your final logline by hand. You'll find the logline worksheet in the resources. In the next lecture, we'll turn that one sentence into a structure. The beats of a short film that works in three minutes.
Kursun tamamı
YouTube oynatma listesiBölüm 1 · Yapay Zeka ile Film Yapımına Hoş Geldiniz
Bölüm 2 · Yapay Zeka Filmi için Yazmak
Bölüm 3 · Filminizi Tasarlayın
Bölüm 4 · Çekim Listesi ve Storyboard
Bölüm 5 · Filmi Üretin
- 5.1Her Plan için Doğru Video Modelini Seçmek5:18
- 5.2Yönetmen Gibi Prompt Yazmak: Kamera, Aksiyon, Performans5:55
- 5.3Referanstan Videoya: Karakterleri ve Setleri Modelde Tutmak5:13
- 5.4Kötü Planları Düzeltmek: Onar, Yeniden Kadrajla, Yeniden Üret4:53
- 5.5Atölye, 1. Bölüm: Adım Adım Senaryo ve Storyboard6:21
- 5.6Atölye, 2. Bölüm: Adım Adım Yapım5:02