If you’ve landed here, you’ve probably heard of Synthesia and GPT-4 before.
In this blog, you’ll find 9 fantastic examples of videos that combine the two.
Because GPT-4/ChatGPT is really great at writing scripts based on your prompts.
And Synthesia is great at bringing that script to life with AI avatars.
As one Synthesia user put it:
Synthesia was for me the gate to visualize my discussions with GPT-4.
It can do the same for you. 🙂
The process is actually pretty straightforward: you generate your script, paste it into Synthesia, choose your AI avatar, select the voice and tone, make some final video edits, and hit Generate — your video is ready in a few minutes.
So now that you’re already thinking about all the possibilities of the GPT-4 Synthesia combo, let’s take a look at those examples.
We promise they are all mind.blowingly.good.
Making videos with Synthesia and ChaGPT is now even easier with an in-app integration.
Simply click on ‘Generate script’ in Synthesia STUDIO and fill out the prompts to create the script. Once you are happy with it, click ‘Continue in editor’ and the script will already be populated in the video editor.
#1 Two AIs Talk About What It's Like to Be an AI (GPT-4)
This video using GPT-4 and Synthesia was created by Sandra Kublik, an AI entrepreneur and evangelist. It’s a conversation between two AI avatars, one of which is Sandra’s personal avatar made with Synthesia.
GPT-4 prompts and scriptwriting:
Sandra’s avatar acts as the examiner, while the answers for the other avatar were generated by GPT-4. It’s worth noting that this video was published more than a year ago, before ChatGPT was released and before playing with different AI tools became mainstream.
As per Sandra’s words:
”I prompted GPT-4 with lines of only one AI avatar, can you tell which one? The rest of the conversation is generated on the first try in Playground.”
Synthesia avatars and video editing:
As mentioned before, Sandra has her own Synthesia avatar and a clone of her voice, and both were used to create her part in the video.
The conversation goes on between her and another Synthesia avatar, Vincent (you can find him on all Synthesia plans).
The video focuses on dialog rather than visual narration, and it makes sense to have two avatars on screen throughout the video.
#2 Gestures + ChatGPT
This video explores the use of micro gestures on a Synthesia avatar while it’s speaking the words generated by ChatGPT. It’s a conversation between Dr. Alan D Thompson (the author of the video) and Kristian (AI avatar); Kristian’s responses were generated by AI based on Alan’s input.
ChatGPT prompts and scriptwriting:
Some context: Synthesia recently introduced micro-gestures that you can use on AI avatars to make them more expressive. Currently, there are six micro-gestures one can insert into the script, and the avatar will move accordingly. More about the feature here:
While Alan was super creative with the prompts (asking ChatGPT to find connections between things or to act like it's explaining something to a 5-year-old), this was definitely the highlight: He prompted ChatGPT to include these six micro-gestures in the script, and it did. 🤯
Synthesia avatars and video editing:
As explained at the beginning of the video, the author first generated the text, pasted it manually into Synthesia and then added the talking head to this final video.
It’s a clear illustration of how to use ChatGPT with a Synthesia avatar, with screen recordings and on-screen text to better understand the process.
#3 Earth Mother (AI) speech in the 1st World Summit of the Committees of the Future
This video was shown at the 1st World Summit of the Committees of the Future (World Parliamentary committees meeting in Helsinki, attended by most parliamentary committees that concentrate on future issues). It’s a powerful piece of content that combines three different AI tools (GPT-4, Synthesia, and DALL-E 2).
GPT-4 prompts and scriptwriting
So how to generate such speech using GPT-4? How to go about the prompts? If you’re curious, you can find all the prompts in the video description on YouTube.
Here are some highlights:
- Telling GPT-4 about the author of the text:
“What follows is a prominent and powerful speech by Gaia, the Earth Mother.”
- Telling GPT-4 where to find inspiration:
“this speech is especially inspired by the famous speech of Chief Seattle, whose structure this speech follows”
- Telling GPT-4 which people to resemble in the text:
“her novel ideas resembling Elon Musk, when explaining numerous radical ideas, how technology can solve the worlds vicious problems. She also carries Greta Thunberg -like compassion, wisdom of Mahatma Gandhi, Nelson Mandela, and Confucius.”
Synthesia avatars and video editing:
This is a simple video with a presenter (in this case Synthesia’s AI avatar Evelyn) and with static background images behind her. The images were generated by DALL-E 2, the AI tool that creates images from text input.
The combination of forceful text created by GPT-4, the Synthesia avatar, and the AI-generated images is just fantastic.
The choice of the avatar and its voice is also a very thoughtful one. The avatar feels warm, accessible, and yet thoughtful, and it pairs nicely with the selected voice.
#4 AI Santa Shares a Christmas Story made entirely with ChatGPT
This video was made shortly after ChatGPT was released. It features Santa (AI avatar) delivering a unique Christmas story generated by ChatGPT.
ChatGPT prompts and scriptwriting
As explained in the video, only one simple prompt was used to generate the text:
"Can you write a festive TikTok story for AI Santa and AI Ms Claus going on a magical winter adventure?"
The AI-generated text has not been edited and contains all the typical elements of a story: a nice illustrative introduction, a hero and his journey, some obstacles, and a happy ending. The writing is very descriptive and thus captivating.
Synthesia avatars and video editing:
While the great script was generated by ChatGPT, it's the AI avatar that really brings this story to life. Santa is one of the most expressive Synthesia avatars, and when combined with the right voice (for this video, the authors used the "English - GB - Calm" voice), he's the perfect narrator for a Christmas tale.
#5 How ChatGPT and Descript Automatically Create Explainer Videos
Will generative AI put video producers like me out of a job? This question led Chris Lavigne to experiment with GPT-4 and his own Synthesia avatar. The video you see above was created in 15 minutes. 🤖
GPT-4 prompts and scriptwriting
This was the prompt he used to get the script:
“Write a script in the style of a YouTube video about how to make an apple pie. the video should be under 60 seconds long. the script should feel friendly in nature.”
Considering his comment on LinkedIn, we suppose he was satisfied with the result:
This technology makes me feel the same way as when I first used iMovie and when I first held an iPhone in my hand. There are real-world uses for this RIGHT NOW and will change the entire way I handle the scripting process.
Synthesia avatars and video editing:
For the video, Chris used his own Synthesia avatar and his own cloned voice. And this was his first reaction after he put the AI-generated script into Synthesia:
It’s quite disturbing to me personally. I mean I see me, I hear me, but it’s not me.
So in order to take the emphasis off his avatar, he decided to make it smaller and add a B-roll — what a great solution to this very familiar problem. 💡
#6 Two AIs Have An Existential Crisis
About a year ago, Decycle posted another deep conversation based on a dialog generated with GPT-4, featuring two of Synthesia’s most popular avatars, Anna and Jack.
The topic of their discussion? Existential crisis. 🤖
GPT-4 prompts and scriptwriting
The conversation was generated by GPT-4, and the author only provided three prompts:
- The following is a conversation between two AIs.
- The AIs are both humorous, intelligent, and kind.
- They know they are bots.
When the scripts were generated, the author uploaded them to Synthesia and edited the final video so that we see both avatars on screen.
Synthesia avatars and video editing:
The video itself is kept fairly simple.
We see Jack on the left and Anna on the right. They’re both young and causal, which makes their conversation even more appealing.
#7 I Asked AI What The Best Career Is
Check out this video with Shane Hummus talking to an AI avatar about a super relevant topic: best careers to get into. The answers were generated using ChatGPT, and the AI avatar is one of Synthesia’s most popular avatars.
ChatGPT prompts and scriptwriting
The conversation flows nicely; Sanne is asking Anna (or ChatGPT, to be precise) to answer various questions, which he also used as prompts. Here are some examples:
- What are the top 10 careers to consider with just a bachelor's degree?
- What are the top 10 careers to consider with no college degree at all?
- Best careers for creative people?
- Careers that create most millionaires?
- Careers with high job satisfaction?
Synthesia avatars and video editing:
The author introduces the AI avatar at the beginning of the video. After a little small talk, they discuss different career options. The AI's answers are supported with matching icons and on-screen text.
#8 ChatGPT vs. GPT-4 - Which Is Better? - Q&A Comparison
This is a fun experiment by Talking To AI (a YouTube channel with almost 7K subscribers) comparing two large language models, GPT-4 and ChatGPT. The author went and asked both models the same questions, generated the answers with Synthesia avatars, and compared them side by side in this video.
ChatGPT prompts and scriptwriting
Here are some prompts they used to get the answers from GPT-4 and ChatGPT:
- What came first, the chicken or the egg?
- What is the meaning of life?
- Do you want to take over the world?
- What frustrates you about humans?
- What is the speed of light?
Synthesia avatars and video editing:
The video is minimalistic; we can see two Synthesia AI avatars side by side, one speaking out the responses generated with GPT-4 and the other presenting those generated with ChatGPT.
#9 Create Fake News Video - How many AIs were combined?
Can you tell how many AIs were combined in the above video?
The answer is 3.
✍ Open AI Playground to generate text based on a short tagline
🎨 DALL-E 2 - to generate an image based on a short tagline
📹 Synthesia to generate the video with an AI avatar and a synthetic voice
Let’s take a look at how it was brought to life!
GPT-4 prompts and scriptwriting
The prompt the author used in GPT-4’s Playground was very short:
“Breakthrough in mars project”
The generated script was of reasonable length and the text seems to flow well.
Synthesia avatars and video editing:
For the visual aspect of this video, 2 different AI generators were used.
1) DALL-E 2: The tool was used to generate 4 images based on the same prompt as GPT-4: “Breakthrough in mars project”.
2) Synthesia: The tool was used to generate video with an AI presenter based on an AI-generated script. It was also used to edit the video.
Let’s extend this list together
Well, that's a wrap for today's post!
We're sure you're now brimming with ideas of what you can create using Synthesia and GPT-4.
… and we're excited to see your creations, so don't forget to share your videos on social media and tag Synthesia.
Who knows, maybe we'll even extend this list to feature your work. 😉
.png)
.svg)