Yao Yao@YaoHuynh
Spoil Alert 🚨🚨🚨
People often think these videos of "Kiss-cam meme" are simple.
In reality, each one goes through three major steps, combining AI automation and predictive human control — a process that delivers around 95% success rate in final generation.
Some say this project looks easy, that it could be done in a few hours. Hmm...
That’s true — but only for people who already have deep experience in image and video generation. For anyone new, it could take days just to get one usable result.
Even experienced creators often spend hours tweaking prompts, images, and layers — with less than 50% success.
👉👉👉That’s why our workflow is special: It’s structured, repeatable, and scalable — difficult to replace even by future AI improvements.
Yes, AI will get smarter. But the complexity of our process ensures precise, high-quality generation — something end-to-end automation still can’t achieve. Today, even the best AI tools often produce broken or nonsensical videos without structured control. Our system bridges that gap — ensuring consistent, personalized outputs at scale.
Ok, chitchat too much. How it work?
👇👇👇👇👇
📷 Step 1: Chatbot → Generate the Prompt Form (You can try the chatbot on the link below the comment - we will share all chatbot from our project free in the future if funded)
- The chatbot helps generate the creative “form” for the next stage.
- Users only need to upload their personal image, and the chatbot prepares a detailed visual instruction at the output.
- That output prompt defines everything for the next step output — characters, background, composition, and style.
🔥🔥🔥At this step, we can easily customize by editing chatbot instructions:
– Change the people (not Charles Hoskinson anymore)
– Change the art style (not Simpsons, maybe Pixar or anime)
– Change the setting (different stadium, concert, or event)
So, change one thing, we will have one new variation/ new template
📷 Step 2: Image Assembly → Generate the Scene
- The system takes the prompt from Step 1 and combines four images:
1. Background (the Kiss Cam LED screen)
2. Cartoon-style Charles Hoskinson
3. Real face of Charles Hoskinson
4. The user’s uploaded image
All of these feed into the image-generation AI (currently using SORA) to create the still scene — the initiation frame for the video.
🔥🔥🔥At this stage, we can produce endless variations:
– Change the background → new location
– Change the person → new pairing
– Change the AI model → new visual style
One single master setup can become hundreds of different templates.
📷 Step 3: Video Generation → Bring It to Life
- The output image from Step 2 becomes input for image-to-video AI.
- We use a fixed or mega prompt to animate the scene — adding subtle movements, camera pans, and expressions to create a living, personalized moment (Depending on the tweak)
- You can try this general prompt:
```
```
- For our “Kiss Cam with Charles Hoskinson,” the fixed prompt animates the camera and character motions as the sample you can see, and in the simpson style.
- But other different tweak can use other different customized prompts to achieve new moods — romantic, funny, cinematic, dramatic, or event-based.
⸻ 📷 The Result: Complex, But Reliable
🌻🌻🌻This single template took around 60 hours of research, testing, and adjustment to perfect.
But now, that one master template can be reused and remixed by artisans (our creative editors) into hundreds of unique variations — new faces, styles, backgrounds, even sounds in the future.
🛻🛻🛻We’ve also built a dedicated system that lets end-users generate these videos completely for free — no need to buy or connect any AI accounts.
In our workflow, tools like GPT, Sora, or Veo normally require paid or professional accounts. But our system already integrates these capabilities — for example, we use our own DALL·E access to generate images directly.
👉👉This means users can create their personalized videos in just a few minutes, without any setup, subscriptions, or technical barriers.
💕💕💕💕💕
So now you can see how much effort and time we’ve truly put into this. If our proposal gets funded, it will not only motivate our team but also let us help the community grow even faster. We’ve already done the hardest part — building the system, proving the process, and making it work.
The next step is the easy one: using the funds to incentivize the community to generate their own personalized videos — spreading Cardano’s spirit, culture, and stories everywhere.
🚨🚨🚨So please, please, please, take a moment to read our proposals carefully. You don’t have to trust my words — just look at our work and decide for yourself.
👍👍👍But we’re confident: if you vote for us, you won’t regret it.
