romantic thriller short film, generated by Google Veo 3. This video is an example of detailed, diverse, realistic character models; continuity with characters and environments between cuts; music; voice acting; subtitles; and
product placement. Google Veo can be purchased at multiple subscription tiers and through Google "AI credits". The software itself can be run by two different consoles,
Google Gemini and Google Flow. Gemini being geared towards shorter, quicker, and faster projects, using the Gemini AI chat model, with Google Flow, which is essentially a
movie editor allowing users to create longer projects with continuity, using the same characters and actors. Users can create a maximum of eight seconds per clip. Additionally, video content can be created using Whisk in the
Google Labs platform. Google Veo has a simple interface and dashboard. However, those who have little to no experience in
transcribing or
filmmaking may face issues when writing prompts, with the software misunderstanding what the user intended by their prompt. So prompts, which are the forefront of the software, need to be not only clear but also specific. When it comes to human models, Veo is able to generate several ethnicities and body types. The software is also capable of generating
stand up comedy routines,
music videos, animals, cartoons, and animation. Prompts need places, people, and things in each scene, in addition knowledge of film and camera lingo such as
panning,
zooming, and terms for
camera angles. Veo, however, has strict guidelines and blockades to their software. Before a clip is generated, the algorithm computer software reviews it, and if it is • inappropriate • too graphically sexual • illegal • showcasing graphic abuse, assault, or fighting (unless the prompt specifies that it is a fictitious martial arts scene etc.), gross behaviors •
antisemitic •
racist •
homophobic • depicting current
regimes,
rioting,
blood, gore, or
warfare, (unless in some cases the prompt specifies that it is fictitious
period drama) the clip will not be generated. In addition, Google Veo cannot and will not generate character actors that look identical to celebrities or real-life individuals. Users have primarily complained that, regardless of how descriptive and detailed their prompts are, Google Veo often misunderstands the input, resulting in completely different outputs. Common issues include the emulation of incorrect subtitles and captions, the generation of complex scenes that are incomplete due to the maximum length, the production of garbled and nonsensical speech, and character models that appear deformed in both appearance and movement. Users have also reported that their prompts and generated content are falsely flagged as violating guidelines, along with a variety of other issues and complaints. However,
trial and error may have to be used with Veo for optimal results. ==Reactions==