D-ID Speaking Portrait: Create Digital Talking Characters from Watercolor Portraits
D-ID Speaking Portrait technology enables photorealistic avatars, using just text or audio as input. As an experiment and soon-to-be long-term project, I tested D-ID’s Speaking Portrait technology on Chinese watercolor master artist, Xiang Li and her latest empress collection painting.
Why we needed D-ID and the Speaking Portrait Technology
Xiang Li, a master artist from the Forbidden City of Beijing, China has dedicated her life to creating portrait watercolor paintings based on very important historical fiction and figures in China. Such as the Dreams of the Red Chamber collection featuring 413 characters, and her most recent project since 2012 focusing on Chinese empresses from various dynasties including the Tang, Song, Yuan, Ming, and Qing.
Li’s expertise is incomparable, and her research effort is unparalleled. But marketing her work and engaging with an audience do require strategy and effort. Li’s intention is not only to produce incredible artwork through her extraordinary skills and patience, but she also wants to reach a younger generation who appreciates and can help preserve ancient history in China. Together with Li, we have been eager to find a technical solution that helps bring the characters in her portrait paintings to life.
Prior to discovering D-ID, we wanted to work with developers, 3-D designers, voice actors, and historians who need to be fluent (or at least familiar) with the Chinese language and history. The cost is beyond our budget, and it would have taken years to test and iterate. With D-ID, we were able to create digital talking characters in minutes.
What we are able to create in minutes: Digital Talking Characters using D-ID’s Speaking Portrait Technology
Li painted half a dozen Ming Dynasty empresses on a silk panel. To help tell the story of each one, Li provided text descriptions in Chinese, which we were able to import into D-ID, select from an available list of Mandarin Chinese speakers, apply speaking style and emotion, and export these videos in seconds.
Social media reactions from people because of D-ID’s Speaking Portraits
Immediately we noticed an uptick in Xiang Li Art’s social media engagement on Instagram and Facebook. In addition, Li also posted these videos on her WeChat account for her followers in Mainland China. People were stunned by these videos and the comments have tripled.
Prior to using D-ID, Li described her paintings using text alone. These historic figures are often hard to put into text that is relatable to people today. From a content creation perspective, it has been challenging for both Li (the artist) and our content creation team. Not to mention most Chinese phases used to describe these empresses are hard to read and understand (think about it as writing like Shakespeare!), which would have made audio recording a sizeable task.
3 simple steps to create a talking character from Li’s portrait painting
You can use the portrait as-is, or crop it closer to the character’s face before uploading to D-ID. As seen in our examples above, you can include or exclude the character’s background, body, and other elements in the painting. D-ID is able to identify facial features including the head, eyes, and lips, as well as their emotions, therefore making speaking more natural and realistic.
- Once you are ready, click on “+ADD” to upload a photo
2. After the image is loaded, it’s time to give your new character a script, which can be entered on the right-hand side of the screen. I’ve included our exact setup in Mandarin Chinese. No worries, D-ID supports over 100 languages.
We are especially excited about the multiple language options. This way able to let the empresses tell their stories in 100+ languages!
3. When you are ready to create the talking character video, just hit the purple button “Generate Video”, and you are done!
Note: what if you run into any technical issues and are unable to generate the talking character video? We provided a helpful solution to triage one of the comment errors here: D-ID: Face not detected error FIXED!
Adding your own voice to the talking character
You may have noticed an option inside D-ID to “Upload Voice Audio”. It can be your own voice or someone else’s (as long as you have the rights and permission to use it).
To generate a natural voiceover for your talking character, we recommend ElevenLabs among a few other choices. Check out Best AI Voice Generators: Top 3 options (2023)
Accessing the video library
All the previously created videos can be found under “Video Library”. You can download any one of them, and share the videos directly on social media such as Facebook, LinkedIn, etc.
Pricing for D-ID
D-ID Speaking Portrait is part of D-ID’s subscription model. If you are new to D-ID, you can use their trial for free for 14 days. That’s what we did, and we were to create multiple quality clips from the trial alone.
After that, their “Lite” subscription is $5.99/month but as you can see, it increases rapidly from there. Even for only 10 minutes per month, we are able to 10-20 quality talking character videos.
Conclusion: D-ID’s Speaking Portrait
In short, we LOVE it. This is such a promising tool for creators and artists. It helps us reach the new ground by coming up with solutions that wouldn’t otherwise be possible.
D-ID also helps save a ton of time when creating something as sophisticated as a digital talking character.
We were also impressed by the different Chinese accents included in D-ID’s studio. Li and I had a good laugh at the realistic accents that made her empresses seem more relatable. After all, not all of them speak Mandarin Chinese, but also other dialects in China. That was brilliant and a pleasant surprise. Cherry on top I should say!
What are your thoughts? Please let us know in the comments below how you use (or envision using) D-ID to create your work and tell your stories. We are eager to hear!
You might also like…
- D-ID: Canva integration, chat.D-ID and Creative Reality Studio
- D-ID: Face not detected error FIXED!
- Best AI Voice Generators: Top 3 options (2023)