Hey folks! I've been playing with EchoMimic quite a bit and have been impressed by its quality on certain types of videos. I work at Sieve and we recently stood up an API interface for EchoMimic and other top models from companies like Hedra and Infinity AI.
We found that applying CodeFormer to the output tends to make particular videos (real people) look better though the primary issue with EchoMimic is that it is really sensitive to which source image you pick (whether the mouth is open, etc). This is where Hedra and Infinity are more robust.
Hope folks find this interface useful for experimentation + production :)
Hey folks! I've been playing with EchoMimic quite a bit and have been impressed by its quality on certain types of videos. I work at Sieve and we recently stood up an API interface for EchoMimic and other top models from companies like Hedra and Infinity AI.
https://www.sievedata.com/functions/sieve/portrait-avatar
We found that applying CodeFormer to the output tends to make particular videos (real people) look better though the primary issue with EchoMimic is that it is really sensitive to which source image you pick (whether the mouth is open, etc). This is where Hedra and Infinity are more robust.
Hope folks find this interface useful for experimentation + production :)