Building MotionVideo AI to Test If Creators Want More Control Over AI Video

I recently launched a new AI video tool called MotionVideo AI.

The idea is simple:

Instead of generating a video only from a text prompt, users can upload:

A reference image
A motion reference video
Then generate a new AI video with controlled movement

The goal is to make AI video generation more predictable, especially for use cases like character animation, avatars, mascots, AI influencer clips, and short-form social content.

Why I built it

AI video generation is improving very fast, but I noticed one recurring problem:

The output can look beautiful, but the motion is often hard to control.

A text prompt can describe the style, scene, and character, but it is still difficult to describe exact body movement, gestures, timing, and rhythm.

For example, if a user wants a character to wave, walk, dance, or follow a specific movement, prompt-only generation can feel unpredictable.

That made me interested in motion control as a product direction.

The core workflow

The current version focuses on one simple workflow:

Upload a reference image

This defines the character, avatar, mascot, or subject.

Upload a motion reference video

This defines the movement.

Generate the output

The final video follows the motion from the reference video while using the uploaded image as the visual subject.

The product is still early, but I wanted the first version to be very focused rather than trying to build a full AI video platform from day one.

Who I think this is for

I am currently thinking about a few potential user groups:

AI influencer creators

They may already have a character image and want to create more short-form content with motion.

Social media creators

They want fast visual content for TikTok, Reels, Shorts, or other platforms.

Brand and marketing teams

They may want to animate a mascot, avatar, or product character for simple social or promotional videos.

Character and avatar creators

They want to turn static characters into moving videos without using traditional animation tools.

The challenge

The biggest challenge so far is not only technical.

It is product clarity.

When people hear “AI video generator,” they may expect a normal text-to-video tool.

But MotionVideo AI is more specific:

Reference image + motion video = motion controlled AI video

So the homepage needs to explain this very quickly.

I am currently experimenting with showing a three-part visual workflow above the fold:

Motion Reference Video
Reference Image
AI Motion Output

This seems much easier to understand than only showing the final generated video.

Current monetization idea

The pricing model is credit-based.

Users can generate videos with credits, and later I may add subscription plans for heavier users.

I think credits make more sense for AI video because generation costs can vary and users may not need the same volume every month.

What I want to validate

Right now, I want to validate a few things:

Do users understand the motion control workflow quickly?
Are creators willing to upload both a reference image and a motion video?
Which use case has the strongest demand?

AI influencers, avatars, mascots, dancing characters, marketing videos, or something else?

Would users pay for higher-quality output, faster generation, or no watermark?
Is “motion control AI video generator” a strong enough SEO entry point for a broader AI video platform?

What I am working on next

The next improvements I am considering:

Better homepage examples

Especially examples that clearly show input video + input image + output video.

Motion templates

For users who do not have a motion reference video ready.

Better onboarding

To explain what kind of reference images and videos work best.

More use-case pages

Such as AI avatar motion control, character animation from video, and mascot video generator.