This FREE AI Video Does Something New!

The video introduces a groundbreaking feature in AI video generation—character referencing—implemented for the first time. The speaker explains the different workflows in AI video: text-to-video, image-to-video, and video-to-video, highlighting their strengths and weaknesses. Text-to-video is noted as the most inconsistent, often producing variable results, especially with character consistency. Surprisingly, a new video model called Viu has implemented character referencing, allowing users to upload a character reference image and generate videos featuring that character consistently. Examples are shown to illustrate how Viu works, including tips on achieving the best results.

Various character reference tests are performed using Viu, starting with a noir-themed character video. While the character's appearance slightly differs from the reference image, the result is still impressive. Another experiment uses a warrior image to create a cinematic setting, resulting in a consistent but slightly altered character. Further trials using real and illustrated references show mixed results, highlighting the limitations when working with non-photographic characters. Notable results include a footballer rendered as a pirate, showing how Viu keeps character traits intact while adding creative elements, such as a dagger magically appearing.

The speaker experiments with personal character referencing by uploading images of themselves, resulting in videos depicting various scenarios, like driving along the Amalfi Coast and investigating crime scenes. The consistency of AI-generated 'self' across different contexts is praised, despite minor discrepancies in facial features and overall rendering. The speaker humorously contrasts the adventurous AI version of themselves with their real-life experience, underscoring the potential and fun of character referencing. Upscaling the videos with tools like Topaz is recommended to enhance quality, showing subtle but noticeable improvements in detail.

Viu's upscaling capabilities are tested and compared with other tools like Korea’s creative video upscaler. Examples highlight how upscaling can enhance video quality by adding detail and refining animations, though some visual inconsistencies remain. Despite some morphing issues and limitations, the creative upscaling feature sets Viu apart, enhancing the generated content significantly. The speaker showcases various clips, demonstrating how upscaling improves scenes with added texture and clarity, making the output more visually appealing and cohesive compared to the original low-res versions.

The speaker concludes that while Viu is not a direct competitor to leading video generation tools like Luma or Runway, its unique character referencing feature makes it a valuable secondary tool. Viu’s pricing model, which includes 80 free credits per month, is considered fair, though the credit consumption for generation and upscaling is noted. At $7.99 per month, Viu offers an accessible entry point for those interested in exploring AI-generated videos with character consistency. The speaker encourages viewers to try Viu and provide feedback, highlighting its potential for future development with creative upscaling still in progress.