I don't use this for external use, but in a corporate firm for promotional posts and videos. I don't use the models, just the image instruct, the lip sync, and the video.

For the image instruct, I find it pretty good. It can do a similar action to using ChatGPT, where I can upload an image. Example: A very basic scene from Blender with a mascot robot, a human for sizing representation, and a blank background. Asked the tool to: "Add a sports car next to the robot. The robot's arms are by its side (it was in T-pose). Make the man next to the robot more realistic and human. The man's arms are in his pockets (was in A-pose), and he is smiling. Replace the background with a corporate entrance to a skyscraper. Business people are coming out of the doors." Wish i could do the ChatGPT style and have a conversation iterating the image.

The image came out looking very good and pretty realistic. Man had hands in pocket and was smiling. It replaced the 3D model of the man with a realistic human and added the car and background. Then transferred that to the image generation part of the tool. The tool was just as good as the current crop of image generation out there and just as good at composing the scene as ChatGPT. Not as good as text, though.

I would describe the video generation capabilities of generative AI as outstanding. It works better for people and robots than the results I get with SORA for image-to-video conversion. Gemini can't do that yet and it trades blows with Klings. Don't have all the extra controls though. Still, the quality is very good, and I'm satisfied to use it as my main image to video generation. Considering I've got tier 3, I will get great bang for my buck for the amount of videos.

Lip-sync works, but since I use a robot, it’s a bit hit and miss. If it had the feature of being able to box, highlight the area of the mouth like the tool called Dzine, then it would be excellent. Need a bit of work on head movement though.

All in all, this is a keeper for me, and I will continue to use it for my video work.