Show HN: Infinity – Realistic AI characters that can speak

It's much easier to use standard CSS packages, and these come with more standard styles. Our team doesn't have much experience building websites, so we just went with the standard styles. We used TailwindCSS.

No, not related. We just took some of Loopy's demo images + audios since they came out 2 days ago and people were aware of them. We want to do an explicit side-by-side at some point, but in the meantime people can make their own comparisons, i.e. compare how the two models perform on the same inputs.

Loopy is a Unet-based diffusion model, ours is a diffusion transformer. This is our own custom foundation model we've trained.

Breathtaking!

First, your (Lina's) intro is perfect in honestly and briefly explaining your work in progress.

Second, the example I tried had a perfect interpretation of the text meaning/sentiment and translated that to vocal and facial emphasis.

It's possible I hit on a pre-trained sentence. With the default manly-man I used the phrase, "Now is the time for all good men to come to the aid of their country."

Third, this is a fantastic niche opportunity - a billion+ memes a year - where each variant could require coming back to you.

Do you have plans to be able to start with an existing one and make variants of it? Is the model such that your service could store the model state for users to work from if they e.g., needed to localize the same phrase or render the same expressivity on different facial phenotypes?

I can also imagine your building different models for niches: faces speaking, faces aging (forward and back); outside of humans: cartoon transformers, cartoon pratfalls.

Finally, I can see both B2C and B2B, and growth/exit strategies for both.

+1 for the lips matching the dubbed speech, but I'm not sure about cloning the actor's voice. I really like dubbing actor's unique voices and how they become the voice of some characters in their language.

Our transformer model was trained to generate videos that are up to 8s in length. However, we can make videos that are longer by using it an an autoregressive manner, and taking the last N frames of output i to seed output (i+1). It is important to use more than just 1 frame. Otherwise ,the direction of movement can suddenly change, which looks very uncanny. Admittedly, the autoregressive approach tends to accumulate errors with each generation.

It is also possible to fine-tine the model so that single generations (one forward pass of the model) are longer than 8s, and we plan to do this. In practice, it just means our batch sizes have to be smaller when training.

Right now, we've limited the public tool to only allow videos up to 30s in length, if that is what you were asking.