Subscribe
Sign in
D-ID's Creative Reality™ Studio — Use generative AI to create future-facing videos
The Creative Reality™️ Studio is the world’s first platform to combine GPT-3, Stable Diffusion and D-ID’s unique face animation technology. Our generative AI will turn your vision into a talking avatar within seconds.
1,044
Replies
Sagee Ben-Zedeff
I love the integration with Stable Diffusion and GPT-3. I think it solves the two main blockers for creating "talking avatars" - creating the actual avatar AND generating the content itself at a highly professional manner. Kudos to the team!
Share
Thanks @altryne for your feedback!
Yes, we have a public open API - available for our enterprise customers.
Generating a video takes ~50% of the video length, i.e. 1-min video will take ~30 seconds to render.
Please feel free to reach out in case you have additional questions.
Thanks for the feedback, @altryne. Like Yaniv said, rendering takes roughly half the time of the video playtime, so in the case of short videos, literally seconds.
@ron_friedman @yaniv_levi2
That sounds awesome, especially the speed!
I just finished a hack-a-thon project this weekend that does this via wav2lip and TTS via uberduck.
It uses chatGPT to respond with videos voiced by UberDuck, so you can talk to say Alan Rickman or Peppa pig, and chatGPT will respond with a video. We were only able to do lips with wav2lip for the hack-a-thon but I would love to just see how better this would be with your tech (and of course share on twitter with the proper attribution)
Any chance I can get a few credits to play around with this with my bot?
More info (with wacky examples) here:
https://twitter.com/altryne/stat...
I already used 4 credits to do some tests... really impressive... it would be nice to have some more free credits for the community here on PH! :)
@alexheikel Thanks Alex, we're excited to see what the future holds. What's your favorite tool?
@gilperry123 @donna_grif Thank you Donna for your support! very much appreciated 😌
@benrosenzweig Thanks Benji, we're really excited about all the positive feedback
Thanks, @benrosenzweig. The use cases are really endless. All you need is an idea and the AI will do the rest for you
hi @gerasim_sergey we appreciate you drawing this to our attention We’re really sorry if one of the team has overstepped the mark in eagerness for our product. We have strict orders never to ask for votes just to ask for feedback.
Whoa, I have copied some text from my website and pasted it to ChatGPT to generate a video script, and paste the script into D-ID, and BOOOM! My marketing video was ready in 2 minutes! check this if you want to see the result https://twitter.com/mehmetkose/s...
Thank you @mehmetkose for sharing. Great to see how easy it became creating videos :)
Feel free to send us any future feedback!
This is exactly the process we're going for - simple, fast and efficient. Thank you @mehmetkose
Awesome product @gil_perry !
I am building AI live stream tool and would like to use D-ID for the AI host.
Any chance I can get access to API to start building without calling (the closest available date is next week)? So I can start building right now..
@gil_perry @archiolidius Thanks Alexander ! We saw your inquiry and someone from our team will contact you shortly :)
@tochepa Hey Patrice! thanks for letting us know, its seems to work fine on our end - are you still having problems with it?
Thank you @on - definitely a super important milestone for D-ID. Thanks for your support!
I've worked on 3D digital assistant technology with a face + voice and one of our biggest challenges was the "uncanny valley" effect (This is apart from other related issues like how do you get the lip movements correctly with the audio, expressions, etc.). Even working with the best 3D facial animation company in the world (who has done work on AAA games and movies), we could not get the quality to what would be okay to a normal consumer. And then there's challenges for the roundtrip time taken for the STT, knowledge discovery, TTS, for the rigging system to animate the 3D facial structure, and all of it to come together.
Just looking at your demos I can say that this is really really good. The approach of going full generative gives you creative control over the uncanniness of your avatars and your face animation being on 2D images has much less constraints than a fully rigged 3D modeling solution has, making it feasible to achieve higher quality.
Good luck!
@puneet_kohli thanks for your feedback. Like you say, sometimes 3D animation can be incredible, but it can also feel like going a long way round, to do something that's relatively straightforward in 2D. We'd love to see the kind of thing you are going to do with it
@mattski2000 I no longer work with that team / company that was working on the project but I believe they killed the project soon after the generative face animation world started to pick up and the general reaction to the avatar was that "this is really weird / creepy".
@puneet_kohli thanks Puneet. Yes, when we started, the uncanny valley was a big challenge to overcome. To this day, our machine learning team is still making updates so that we can continue to get better and better.
@gil_perry I can see you folks are already making strides there with respect to the lip/jaw animation. The upper part of the face (eyes/eyebrows/etc) still need some work for added emotion. It's an interesting subject for sure which I did spent a decent amount of time exploring and learnt about in the process, happy to chat if you folks are up for it.
(The product I'm launching soon is from my current line of work which is totally unrelated)
This can solve so many video issues for companies. I love it and can't wait to use if for my next project.
Spot on, @roypovar. Dramatically reduces the hassle of video production and infinitely customizable to every audience