Urgent.News

What's breaking now, across hundreds of outlets.

Tech

Same Character, 25 Clips: How ULTRA Was Made

Originally published on the Character Lock blog . ULTRA is a 3-minute AI brand film (3:13) with one bear dad holding across more than 25 clips. It was made with a method called Character Lock: define the character once, lock it on a sheet, direct every shot against that sheet, and check every frame before delivery. Here is how it was actually made, step by step. The ask ULTRA sent a script and a…

ULTRA is a 3-minute AI brand film, consisting of 3 minutes and 13 seconds, featuring a bear dad who appears in over 25 clips. The film, created using the Character Lock method, involves defining a character once, locking it on a sheet, directing every shot against that sheet, and meticulously checking each frame before delivery. Initially, ULTRA received a script and a reference ad, which depicted a father constantly anxious and a friend who later switched to a nicotine-free pouch.

The challenge lay in maintaining the characters' consistency throughout the 3-minute story, regardless of the setting or changes in props. To address this, the bear dad was chosen as the main character and his companion, a golden Labrador, was selected to complement him. By keeping the characters constant throughout, the audience could identify with the stressed dad as the bear's tension remained unchanged throughout the film.

The film's creation involved seven steps. First, the reference ad was dissected second by second, analyzing story beats, shot lengths, product placement, and the timing of the benefit. Next, sheets were created for both the bear dad and his friend, featuring an ear notch and a ban on mascot-suit looks or color casts. The product, the ULTRA tin, was locked in place through sheets that ensured its appearance remained consistent across all shots.

The locations included a kitchen, a garage, and a driveway, each equipped with a light source that motivated the scene. Eighteen still frames were directed, specifying which sheets to attach, the lens, height, and lighting conditions. The animation process took place in Google Flow using Veo 3.1, where each frame was animated in short takes, ensuring the product remained the same in every shot.

Finally, the film was completed by adding captions, sound beds, cleaned dialogue, and a music swell integrated into the reveal, achieving an overall LUFS master of minus 14. The process revealed that maintaining a constant character over 25 clips and maintaining six distinct characters in one film required the same skill - a lock, a sheet, and a frame-by-frame check. This system proved ULTRA's success as a full-length film, demonstrating the effectiveness of the Character Lock method.

Rewritten by urgent.news; may contain errors — check the original.

Read the original at dev.to →

Report a problem with this story

More in Tech

Change a screenshot to the required aspect ratio in the tab before you commit the hero

The docs call for a 16:9 hero, or the OG template is 1200×630. The file in the PR is a 4:3 laptop screenshot. The layout letterboxes it, or a later object-fit: cover chops off the toolbar you meant to…

  • Hero image must be 16:9 aspect ratio, 1200×630 pixels
  • Layout team treats aspect ratio as initial check, not CSS compensation
  • Use in-tab ratio tool to adjust image before committing

A Japanese NLI checker for the space between generation and trust (sealed 0.961, zero-shot baselines 0.79–0.81)

TL;DR — We released ja-nli-v1, a small Japanese natural language inference model: given a premise and a hypothesis, it returns probabilities for entailment / neutral / contradiction.

  • ja-nli-v1 Japanese NLI model achieves 0.9612 accuracy on sealed test
  • Outperforms zero-shot baselines with 0.79–0.81 accuracy
  • Model trained on synthetic data from public-domain government documents

How I Automated 10M+ Network KPI Records a Day with Python and Apache Airflow

Originally published on pmpdilshan.github.io . Every hour, a mobile network generates performance counters for every cell: traffic, drops, handovers, throughput, congestion.

  • Mobitel generates over 10 million network KPI records daily
  • Automated pipeline using Python and Apache Airflow reduces effort by 40%
  • Automation frees up time for optimization work and KPI centralization

More from Saturday 21 November →