Create Your First Video

Creating a video in TutorFlow starts with a topic and ends with a rendered MP4. The process has two phases: generation, which is automatic, and editing, which is optional but recommended before you render.

Step 1: Choose a format

Before entering your topic, select an aspect ratio:

Format	When to use
Landscape 16:9	Course content, lectures, onboarding, training clips
Portrait 9:16	YouTube Shorts, Instagram Reels, TikTok

Landscape videos support up to 10 minutes of content. Portrait videos are capped at 3 minutes and use a hook-first script structure suited to short-form platforms.

The format is set at creation time and cannot be changed after the video is generated.

Step 2: Describe the topic

The topic field drives everything: the script structure, the narration tone, the keyword selection, and the Pexels footage search terms. A more specific topic produces a more usable first draft.

Instead of: "Machine learning"

Write: "A beginner-friendly introduction to how neural networks learn, for university students with no math background."

You can also provide optional details (up to 2,000 characters) to give TutorFlow more context about the audience, the scope, or any specific points to include.

Step 3: Set the video length

Use the duration slider to set a target length. TutorFlow uses this to determine how many scenes to generate and how much detail to cover per scene.

Landscape shortcuts: 30s, 1m, 2m, 3m, 5m, 7m, 10m

Portrait shortcuts: 15s, 30s, 45s, 1m, 2m, 3m

For a first video, a 1-2 minute length is a practical starting point. It is easier to extend a short video than to condense an overly long one.

Step 4: Choose a voice

Select the narration voice preference: female or male. TutorFlow maps this to an optimal AI voice for the topic language and generates TTS narration for every scene.

Language detection is automatic. If your topic is written in Korean, the narration will be in Korean. If it is in English, the narration will be in English.

Step 5: Generate the video

Click Generate. TutorFlow runs three tasks:

Script generation. The LLM produces a scene-by-scene narration, assigns keywords and text templates to each scene, and suggests Pexels search terms for footage.
TTS generation. Each scene's narration is converted to speech and stored. This runs in parallel across all scenes.
Footage retrieval. Pexels is searched for a matching stock clip per scene. Clips are uploaded and stored. This also runs in parallel.

You are redirected to the scene editor immediately after script generation. Scenes populate as TTS and footage arrive, typically within 30-60 seconds for a 10-scene video.

Scenes with a warning indicator have a missing audio or video asset and can be retried individually from the editor.

Step 6: Review and edit

The scene editor opens automatically after generation. Before rendering, check:

Script. Does the narration match the intended tone and detail level? Edit individual scene scripts directly in the editor.
Footage. Does each clip match the scene content? Search for alternatives or upload your own.
Templates. Does the text overlay style suit the scene? Switch between keyword, title, quote, and none.
Timing. Use the NLE timeline to align audio, subtitles, and video offsets.
Transitions. Set fade, slide, wipe, flip, clock wipe, or iris transitions between scenes from the transition picker.

Use the Remotion Player to preview the full video before rendering.

Step 7: Render

When the video is ready, click Start Render. TutorFlow triggers a cloud render that:

Combines all scene footage, audio, and text overlays
Applies transitions between scenes
Produces a single MP4 file

Render time is typically 1-3 minutes for a 1-2 minute video. A progress indicator is shown during rendering. You will be notified when the render is complete and the MP4 is ready.

If a render fails, you can retry without being charged again.