Apps and sites we build and maintain alongside this one.
Mock tests, study material, books and an AI-powered study app.
Automated video production and channel management tools.
Track, issue, and manage books with a simple mobile app.
Score and review handwritten answer sheets on the go.
A study companion app built for students.
Read the latest blogs and news updates.
Create presentations quickly for online classes.

Creating a cinematic short film used to require expensive cameras, a massive crew, and weeks of tedious editing. Today, generative AI tools have completely democratized filmmaking, allowing anyone with a phone or laptop to generate realistic, high-quality video clips directly from text prompts. If you are a creator looking to learn how to make an AI film (ai film kaise banaye), Google Flow is one of the most powerful tools available today. By combining Google's advanced video models like Veo 3.1 with an intuitive scene-building timeline, you can control your narrative from a single workspace.
In this guide, we will walk you through the end-to-end workflow of creating a 45-second cinematic google flow short film. From writing a structured shot list to maintaining character consistency and assembling your final edit with sound in CapCut, you will learn the exact steps to turn your creative ideas into a polished movie. Let's dive into the step-by-step process.

To make a successful AI film, you cannot rely on random generations. You need a structured plan. Every second counts, especially when you are working with limited credits. We recommend starting with a 5 to 8 shot sequence that tells a simple, high-impact story. For this tutorial, we will use a story called 'The Last Metro' about a young Indian man named Kabir who finds a mysterious glowing envelope at midnight.
Here is a sample shot-list table designed for a 45-second short film:
| Shot # | Visual Description | Camera Movement | Duration | Model Recommendation |
|---|---|---|---|---|
| 1 | Kabir (25, Indian man, blue jacket) running down metro stairs. | Tracking shot (front) | 8 seconds | Veo 3.1 Fast |
| 2 | Kabir panting on the empty platform, looking around. | Slow pan left | 6 seconds | Veo 3.1 Lite |
| 3 | Close-up of Kabir's face as he notices something on a bench. | Zoom in | 6 seconds | Veo 3.1 Lite |
| 4 | A glowing blue envelope lying on a metallic metro bench. | Dolly in | 6 seconds | Veo 3.1 Fast |
| 5 | Kabir picking up the envelope, looking at it with curiosity. | Orbit up | 8 seconds | Veo 3.1 Fast |
| 6 | Metro doors open, warm golden light spills onto Kabir. | Zoom out | 10 seconds | Gemini Omni Flash |
Before you start generating, it is crucial to understand how Google Flow manages its credits so you do not run out mid-project. Google Flow operates on a credit-based system:
Note: Free users may experience generation blocks during peak hours (roughly 2 PM to 5 PM UTC) and cannot access 1080p upscaling without a paid plan.
Go to labs.google/fx/tools/flow and create a new project. Rename it to 'The Last Metro'. Change your settings to a 16:9 aspect ratio for a cinematic film (or 9:16 for vertical reels). For your first shot, select the Veo 3.1 Fast model. Enter a highly detailed prompt that defines the subject, setting, lighting, and camera movement.
Copy and paste this prompt for Shot 1:
Cinematic shot, a 25-year-old Indian man with short black hair wearing a dark blue jacket running down the stairs of a modern Delhi metro station, neon lights, night, motion blur, camera tracking him from the front, highly realistic, 8k resolution.
Adjust the camera motion settings in the bottom menu. Select 'Dolly In' or 'Tracking' to match your script. Generate the clip and review it. If it matches your vision, save it to your project.
One of the biggest challenges in AI filmmaking is keeping the character's face and the environment identical across shots. Google Flow provides three powerful features to solve this:
Use this prompt for Shot 3 (Kabir's close-up), using the Shot 1 frame as an ingredient:
Cinematic close-up of a 25-year-old Indian man with short black hair, looking down at a bench with a surprised expression, dramatic neon lighting, night, shallow depth of field.
Once you have generated all 6 clips, open the Scenebuilder tab inside Google Flow. This is your built-in storyboard and timeline editor.

The man slowly opens the envelope, revealing a soft blue glow on his face.
While Scenebuilder is excellent for organizing your timeline, you will need a dedicated editor like CapCut, Premiere Pro, or Remotion to add professional audio and color grading.
Yes, Google Flow offers a free tier with 50 daily credits, which is enough to experiment with Veo 3.1 Lite and Fast models. However, unused credits do not roll over, and free users may be blocked from generating during peak hours (2 PM to 5 PM UTC). Upscaling to 1080p and using the premium Veo 3.1 Quality model (which costs 100 credits per generation) require a paid Google AI subscription.
You can maintain face consistency by saving a high-quality frame of your character from an earlier shot using the 'Save Frame' button. Then, add that saved image as an 'Ingredient' (character reference) in the prompt settings of your next shot. You can also use the 'Jump To' feature in Scenebuilder to move the character to a new location while preserving their look.
Veo 3.1 generates native audio contextually paired with your video. If the audio generator produces low-quality or glitched sound, Google Flow will fail the generation to protect your experience and automatically refund your credits. If this happens, slightly modify your prompt or try generating again.
'Extend' is used to continue the action of a single shot (adding 4 to 8 more seconds of continuous movement based on a new prompt). 'Jump To' is used to transition a character or object from one shot to a completely different setting or angle while preserving their visual details, acting as a smart narrative cut.


