Runway ML offers img2vid services, text/image to image and video-to-video. It works similarly than Luma Labs Dream machine, but you need to input starting frame. Ending frame not supported. This plugin supports uploading the source file to your DropBox account, so you need not to worry about extra hazzle. Api keys can be obtained here: https://dev.runwayml.com/
ACT2 is now supported. For that, you need reference image of how the video should look like and reference video of what movements the the video should follow. Reference video can’t be bigger than 32mb (use trimming to scale down or shorten the video if necessary)
For creating images, you need to have at least one reference image as well as tag for the character. The way it works is that you reference the character in the prompt. See the example picture below. Tag there is ‘orc’ and in prompt it is referenced as ‘@orc’
