Text-to-video
Updated: October 17, 2024 at 5:09 PM EDT
*this is an image*
Automating short-form video creation
Content creation is a lengthy process. Faceless content in particular requires idea genesis, research, writing copy, selecting images and video, writing a script, and reading over the script.
Luckily, as of 2024, we have tools for each of these components. To accomplish automation we used GPT-4 (video structure, script), Bard (prompt reinforcement), ElevenLabs (speech), bootleg Midjourney API (visuals), and FFmpeg (stitch).
UI/UX for an obvious core transaction
The UI design was inspired the minimalistic home screen for search engines. Because our core transaction is video generation, we wanted the prompt box to be featured in focus and for the user to interact with it pretty immediately.
prompt input with vertical carousel selection:
If it wasn't obvious what to do with the text box or how to best formulate a prompt, several options were provided as shown above.
Note
In the future, I might revisit this. I genuinely believe that in the medium to long term - the majority of content will be created by AI or with AI assistance.
As an artist and creative, I have mixed feelings - maybe it should be seen as a tool like everyone says... or maybe we could all benefit from being luddites every once in a while.