Video is an excellent medium for engaging and attracting audiences, but the traditional process of creating a video can be time-consuming and expensive. Companies like Synthesia, Colossyan and D-ID create AI-powered tools that allow you to produce videos of a speaking presenter using only text input, lowering cost and effort. Video might become possible for a customized product introduction or as part of an onboarding package for new employees!
Closer look at D-ID
We looked at de features that the different products provide and decided to take D-ID for a spin, they offer multiple ways of creating customized videos featuring talking avatars and have incorporated natural language and image generation tools as part of their process.
Using the pre-made presenters and voices you can create a video in minutes. This method is really easy and gets the best results. We recommend you try it yourself. D-ID’s integration with GPT-3 enables users to extend their input text using AI.
One feature that could be added to the GPT integration is the ability to ask the AI for re-write suggestions or variations on the text in a different tone of voice. This function could expand the possibilities for video creation even further.
D-ID also provides the option to create presenters using a photo or an AI generated portrait of a presenter using Stable Diffusion. You can also provide your own audio recording instead of a text input. We tested both features.
Android Presenter
We created a futuristic looking android presenter for Digital Explorer using the AI image generator. We selected a voice provided by the platform that we thought was appropriate and entered our text. Below the video of our android in a speaking role.
While the result was not perfect, it was impressive considering the minimal effort required. D-ID automatically adds this presenter to the library for future use.
Digital Copy
We also experimented with creating an anchorman using a photo of the author. A digital copy of yourself, ready to present in future videos, would be a nice asset. This method required a bit more effort since we wanted to record the audio, which was then turned into text by the platform and used to animate the face in the photo. The result was somewhat in the uncanny valley, but with better photos and input from D-ID support, the this could be improved.
Finally, for some use cases, users can take a 3D figure or drawing of a face and animate it for videos. Viewers are often more forgiving in these cases since they don’t expect complete realism. Motion capture and 3D production are rapidly advancing fields, so we can expect even more developments in the future.
Overall, virtual video presenters offer an innovative solution for creating videos quickly and cost-effectively. A virtual presenter could be a valuable tool for your line of work. Could you use a video presenter? What would be good cases?
