It's still very much in it's infancy.
As far as local vidgen goes, here's couple of SVD img2vid examples, takes an image and makes a video from it, results are more or less random. The first one is the plain output, after that I've added AI frame interpolation:
svd2.mp4 (Size: 903.2 KB / Downloads: 1,220)
Please log in or register to view svd2.mp4
23fps_svd2.mp4 (Size: 1.9 MB / Downloads: 1,192)
Please log in or register to view 23fps_svd2.mp4
22fps_svd1.mp4 (Size: 1.89 MB / Downloads: 1,228)
Please log in or register to view 22fps_svd1.mp4
And a couple of rapid growth clips made in A1111 using animatediff with prompt travel. :
00002-2608477988.mp4 (Size: 1.21 MB / Downloads: 1,250)
Please log in or register to view 00002-2608477988.mp4
15fps_00029-186200598.mp4 (Size: 1.41 MB / Downloads: 1,236)
Please log in or register to view 15fps_00029-186200598.mp4
And a few made in comfyUI. ComfyUI gets more consistent results because it also allows negative prompt travel plus a few other tweaks that help. But I'm still in the very early stages of playing around with it. An upscaling step would be nice but I haven't got around to setting up a workflow for it.
15fps_cui1.mp4 (Size: 1.09 MB / Downloads: 1,222)
Please log in or register to view 15fps_cui1.mp4
15fps_cui2.mp4 (Size: 1.11 MB / Downloads: 1,203)
Please log in or register to view 15fps_cui2.mp4
15fps_vid_00001.mp4 (Size: 1.3 MB / Downloads: 1,219)
Please log in or register to view 15fps_vid_00001.mp4
I'm barely scratching the surface of what it's capable of. Hand weirdness is still a big problem. Here's a couple of pics attached from which a workflow can be extracted, for those already familiar with A1111 or comfyUI and animatediff: