An A.I, behavioural psychology and blockchain enthusiast here. Disclaimer: zero coding skills
You can call me fud. Long time listener of Ben and Singularity but first time caller
Some background about myself. I’m an independent film maker & a creative visualizer working in the advertising industry for a few years now. My specialization is in optimizing visual storytelling for advertisements through previsualization(previz) that caters to specific demographics based on the brand needs. Previz content includes storyboards, stillomatics and cinematics.
Some example of previz films can be found in the link below
Most advertising agencies are constantly pitching to clients with prototype concepts of stories/scripts to further push their brand to the markets. And the demand for previsualization also has consistently increased as we further delve into the digital world. The problem I m currently facing is the lack of human resources to execute the overload of scripts, eventually ending up in turning down jobs.
For the last few years, i ve been looking at machine learning alogrithms in Natural Language, Image Processing and Procedural Generation of content to automate the different steps of the previz process. I ve tried documenting them to the best of my knowledge with an example of how the algorithm should function based on the input script.
I will be opening up the details of the project in the project blog. Feel free to be chime in, join the conversation and be critical
Happy to join the Singularity fam Exciting times!!!
Nice work Fud! Do you intend that the artworks are generated by AI, or augmented by AI? Do you see opportunities for embedding the outputs (e.g. storyboards) in an Extended Reality environment? Are the inspirations for a client artwork based on search and embedding of their brand and associated images that represent brand values?
Quick disclaimer: ASB’s animated storyboards isnt my work. ASB is a world renowned studio for 2D & 3D previz products… delivering one of the best previz outputs. Most outputs (30 seconder films) seen in their website typically takes 3-4 weeks. Also their high quality comes with a high price and requires significant number of skilled human personnel to execute.
Most of my deliverables are 2D due to time constraints from the client side. I will be sharing my work alongside a solution example document. Regarding my deliverables, the typical time to execute a 30 second previz film is about 6-7 days for a project. Sometimes there are 2 films to deliver within a weeks time where two teams are executing the job in parallel. The clients primarily use these films to test the script’s/story’s effectiveness (focus group research) in delivering the message to the target demographics. Since the timelines are super compressed, the clients are generally willing to compromise on the quality…as long as the message is engaging effectively with the audience.
Regarding artworks generated or augmented by AI… eventually yes i’d prefer direct AI generation of visual from text detection. Since i lack knowledge of GANs, i approached the documentation addressing unique attributes/variables required to generate images through an augmented approach. Generally, the previz/research scripts are the first drafts of the marketing campaign idea. Since the scripts are subject to change based on the focus groups reactions, there are generally a lot of iterations till the script is 100% effective (brand-wise atleast). The iterations on artwork is the most time consuming since things need to be drawn/painted from scratch. This is where having control over the unique attributes/parameters will help narrow down the artwork that needs to be revised based on the feedback.
Regarding extended reality outputs, absolutely yes…eventually. It would be great to simulate a 3d animation of the script on the clients desk/table to show how the script/film would look while pitching. This falls under the procedural generation aspect. This requires integration with a game engine like unreal or houdini. But currently, the previz outputs are are just displayed on a screen so people from every corner of the room in the focus group has good visibility.
Regarding art inspiration representing brand values, absolutely yes. The color palettes, costume styles, types of actors representing their brand etc. Its key that every visual matches their brand guidelines. Hence i ve tried covering most of the parameters that represent a brand uniquely from their competitors…through the augmented approach.
