I guess what we’re all really looking for in Metahumans is performance, and for characters to appear inhabited. Trying a few tests with MH in 5.6 I’ve been super impressed with the improvements, but of course that just leaves you wanting more
One thing that struck me was that while the new emotion settings in MH performance are great, they’re currently a little basic despite percentage weighting. What turned out to work fairly well was running the performance multiple times with different settings, then blending them together in layers in sequencer and adjusting weights there. Although that’s fairly easy to do, with a bit of refinement it feels as though it would make a great tool in its own right. Some of my asks are probably already available in Zendyn so maybe I should stop Scrooging and buy the plugin, but ‘Micro Saccadic Movement’ and ‘Blink Driven Pupil Constriction’ would be high up my wish-list (and more control over constriction / dilation in general).And voice. The test in the link uses Eleven Labs, which is great but to me seems pitched a bit high for the character. Voice is intrinsic to character, and as Metahumans become more realistic I think they’re going to have to tackle this head-on because jumping between apps breaks the flow.
Still - wow. Getting this far with a TTS model is pretty amazing, so massive Kudos to the MH team. Just - can we have some more? Here’s a quick test with layering performances at 4k with a bunch of After Effects. (Warning - may contain poetry). https://vimeo.com/1103413920/2f4ade7cf0