While everyone else is busy making ChatGPT and Bing’s chatbot generate multiple-personalities, I’ve been experimenting with converting my text content into video format with really awesome looking characters combining Midjourney’s AI image generation with D-ID’s ability to do lipsync AI video creation.
Create Your Unique “Talking Head” Avatar With Midjourney AI Images
The first step is to apply a little bit of creativity to the AI image platform, and input a prompt that will get you a front profile image of a character for D-ID to animate. You will not get away with just a flat line for the mouth, their software needs to detect where the eyes, nose and mouth are, it will then do some “head movement” in addition to lip syncing. This means that horns, hair or other parts of the image will remain flat, and non-animated.
I use the “aspect ratio” command of 3:2 to get the width wider than the height, to better fit the general “landscape” layout of video. There’s dozens of different art styles to choose, or themes to fit your particular brand. SEO Arcade is nerdy, arcady and sci-fi so let’s go for a cyberpunk girl.
Now Let’s Animate Our Midjourney Image
Once you’ve downloaded your portrait style image from Midjourney, it’s time to visit D-ID, and upload your new image.
- Medium Shot, Frontal facing
- Neutral facial expression, closed mouth
- Minimal head size within the image is 200x200x
- Good and solid lighting
- Size -up to 10mb
- No face occulusions (stuff in front of the face)
Add Your Voice: Recording or Text To Speech
There’s a fair selection available of basic AI voices, with the American ones even giving you the option to be angry, cheerful, sad, friendly or helpful which changes some tone and inflection. In this case, I went with a New Zealand accent, for funsies.
Improve The Voice By Providing It Yourself (Or use better AI)
Make quick summary or recap videos
This is a new, unique tool that could let you re-purpose your written text into video format, or make a typical tweet jump out from the rest.
How Much Do These AI Services Cost?
Midjourney gives general users a few free credits, but you need at least a basic monthly subscription to have general Commercial license to the images you generate. Free accounts are Non-Commercial Creative Commons. When you have at least a Basic account, “You own all Assets You create with the Services, to the extent possible under current law”. Basic is $10 a month, Pr is $30 a month.
D-ID gives you 10 minutes a month for $5.99, so it’s better for short snippets than long videos. It’s $299 a month for 65 minutes, which is $4 a minute so you would want to really have a plan for larger scale use to get a proper ROI. I haven’t seen any other tools or real competitors to the “talking-head” animation yet, if anyone has a suggestion tweet it at me!
BTW Play.ht is $19 a month for 20,000 words for that improved AI reading voice.