Summary
This guide walks through four ways to make AI videos from scratch, all inside the OpenArt platform. It starts with the simplest approach, a single text prompt that produces a multi-shot scene, using Seedance 2.5 at 1080p and 20 seconds with audio generated in the same pass. A shot-planner tool breaks the scene into individual shots with camera directions, and a single instruction about which side of the frame the subject stays on keeps the cuts consistent.
The second method gives the creator more control. Five location images are generated first with GPT Image 2.6, then attached to a 30-second Seedance generation as a start frame plus references, producing one continuous drone flight through a cave, forest, gorge and valley with smooth transitions.
The third method covers character consistency. The creator builds reference sheets for characters, props and locations, using three panels per character: a headless full-body front view, a full-body back view and a face close-up on a flat gray background. A sibling character is made by borrowing only skin tone, hair and eye color from the first sheet, so the two look related without being copies.
The video also notes that current models can generate roughly double the scene length they could at the start of the year, and mentions OpenArt Black Friday pricing.
📺 Source: Youri van Hofwegen · Published November 13, 2026
🏷️ Format: Tutorial Demo
