Actually, Damoang's AI Dang feels like... AI (developed with) Dang
so an ordinary person like me just lurks and moves on.
Of course, while lurking I have picked up a few good pieces of information and look at them when bored.
Anyway, as I wrote before, rather than for coding I have been trying various things with AI subscription services,
and this time I heard they make music videos with AI, so I opened my wallet once.
It is called Neural frame.
Grand, is it not?
I wanted to make a music video for a song I made with suno, so I searched around
and settled on two final candidates: Higgsfield and Neural frame.
(Of course, those two are all I looked into.)
What I wanted was my song + lip sync,
and while I could tell Higgsfield looks varied and good,
there was not much information about song + lip sync.
By contrast, Neural frame's concept is simply "we make music videos with AI",
so I settled on that.
Registering for a one-month mid-tier plan costs $39.
That is expensive... I know video eats a lot of computing resources, but it is expensive..ㅠㅠ
When you register they give you 2,400 acorns (not a joke, the credit unit really is Acorn....),
which is roughly enough to make two videos about 3 minutes long with the older Seedance or older Kling models.
But of course, editing or changing things along the way drains acorns bit by bit,
so to make two videos you would have to spend frugally.
There is a corner that makes shorts,
so first I took the chorus and made a short.
I generated a character from images I had prepared in advance (+credit consumption)
and made it using preset backgrounds and so on. (If you want to see it, click link #1 please)
It costs 20 credits per second, so making about 20 seconds drains credits fast.
Personally the result was not bad.
The character acts prettier than I had imagined, which is a bit much,
but anyway the lip sync was passable.
So this time I steeled myself and decided to make a music video,
and worked hard at it for half a day today.
(Of course, their GPUs did the hard work...)
Ah, but as expected, a video for a whole song is not easy.
The description makes it sound like a single "click" (the service is even called Autopilot)
but to edit it to your liking along the way you have to "click" a dozen more times,
and past a certain number of times a "click fee" is charged.
The result was not that satisfying either.
I get the feeling that badgering Gemini to make videos one by one and stitching them together was somewhat better.
(Of course that takes more time...ah, and there is a watermark too...link #2 is where it is.)
Of course you can choose the generation model, and perhaps it is because it is not the latest Seedance or Kling.
Those are pricey, so even making one 3-minute video is a stretch.
Anyway, since I spent credits (money = acorns) making it, I will register again next time,
but I think I will use Neural frame only when making shorts.
A 20~30 second lip sync short is convenient after all.
For longer videos I should make good use of Gemini Omni Flash, which renews monthly.
(Google... how about putting in a lip sync feature...?)
That is all for my $39 review.
▶ Original source: https://youtube.com/shorts/1XOui9oJN10?feature=share
▶ Original source: https://youtu.be/vGSvTMS1Igw
