How Google Flow environment is treating me?

Well, I tried Google Flow and made these two ads for my children:

What can I say? My impressions are generally positive.

The working environment is much more predictable. You don't have to constantly fight it to make it follow the prompt. In addition, they have what they call an "Agent" — a built-in AI assistant that helps create an overall storyboard, break it down into individual shots, and suggest ideas along the way.

Another huge advantage is character consistency. You start by uploading a photo, and Flow creates a character from it. After that, the platform largely handles appearance consistency on its own, without the constant identity drift and corrections that I have to deal with in Digen AI.

The main drawback is the price. The $10/month plan gives you 50 credits per day. That's roughly enough for three 10-second clips at 720p quality. There is an option to upscale to 1080p. As far as I can tell, it's not true native 1080p but rather an interpolated version. Still, it looks quite decent.

What I haven't been able to achieve yet is seamless continuity between scenes. In other words, there is a noticeable jump between the end of one scene and the beginning of the next, with the character appearing in a slightly different position or pose. I have to hide it with fade-outs, fade-ins, or other editing tricks.

Ironically, despite all of Digen AI's unpredictability, I can specify the first and last frame, and it will reliably generate everything in between. So, if I can't fully solve this issue in Flow, I may end up using Digen Ai to create those transitional "patches."

Of course, it's entirely possible that I simply don't know the proper workflow yet. I'll be watching more tutorials and read their documentations. There’s probably way to handle it.