I have many original stories that I’ve written over the years.
But I am new to AI videos. I’ve been using chatbots, capcut, powerdirector, suno, grok plus and sondo. I’m open to hearing ideas about other apps I could try.
Here are some examples:
Here’s what i would really like help with. I want to record the voices myself then have ai tools lipsync them for me. The random generated voices are cramping my style.
Sometimes, third arms or 5th legs show up etc or objects disappear. Do you just keep regenerating till it’s perfect?
I have just started AI also. My background is Video film, CGA animation, story development and script writing. (No, I have no screen credits.) The visuals look really good. The lip sync looks great also.
I don’t know what is possible with your generation software. I would change camera movement if possible. (Just a suggestion). Pull the camera back for an establishing shot of the taco stand. Then push in once for a static shot of both characters. Or you could break it up into multiple shots… Establishing Shot, then medium close ups (Heads) of each character as they talk. Maybe do a close up of the jar. But you will have to then edit it together.
You could record your voice and then give to Minimax H3 video model (Seedance 2.0 really does not take audio input well), and refer to that audio in the input prompt, and paired also with in text form then you can have the character speak in your voice. Another route, that I usually use, is you generate the video normally and then use Elevenlabs voice changer, to change that voice in the video to yours. I prefer myself this route. But of course it requires that you clone your voice.
The better apps are the paid ones. I use OpenArt.ai on a paid subscription. Higgsfield is one of the best ones, but it’s pricey. The third arms or 5th legs are all in the quality of the prompts you write. You need to include ‘negative prompts’ and tell whichever AI is doing the generating NOT to do things, as well as what to do. I use chatbots like Grok to write the prompts. I have a ‘AI Task Force’ of five different AI helpers: Gemini, ChatGPT, Claude, Grok, and Google. They each take on roles that movie studio production companies use. Google is Research, Gemini is my Director of Photography, ChatGPT Plus is my director (I pay for that one, but the rest are free tiers), Claude is my script supervisor, and Grok is my Special Effects Expert. I use friends and family for voice-overs. The ‘Twilight Converstaion at the Taco Stand’ looks pretty good. The Style reminds me of the 1982 dark fantasy film The Dark Crystal, Best Regards, Mike DeRosa
Take a look at ElevenLabs - they have voice changer that takes your recording and substitutes another voice. Most AI consolidators offer lip sync, though there are some gotcha’s - the mouth needs to be visible otherwise it loses the plot. AI halucination is a well known phenomenon - one way to mitigate it is to generate key frames as stills. You can use those as start/end or as references. I have found this can help a lot.
