AI slop is video that looks fake: a face that changes between shots, plastic skin, a warped label, words that shouldn't be there, a camera that does something random. It isn't the model's fault. It comes from guessing: whatever you didn't decide, the model decides for you, differently every time.
Why you shouldn't skip the prep+
1. Follow the process. The model is not the point
A new model comes out every week, and the gap between the top ones is small. The gap between a planned video and a guessed one is huge. Every module here uses the same steps: script, screenplay, pictures, clips, edit. Get those right and any good model works.
2. Garbage in, garbage out
The model can only use what you give it. A vague script, a blurry product photo or a reference that already looks computer-made gives you slop, on any model, at any price.
3. The prompt is one input. The pictures matter more
Your character, product and place pictures carry the face, the label and the light into every frame. That's why about 80% of the time on any video goes into the images; the clip itself is the last step. When a clip looks wrong, fix the picture first, not the prompt.
4. Use templates for your prompts
Every lesson gives you a template. Copy it into Claude, add your details, and it writes a prompt that doesn't forget the camera, the light, the exact count of things, or what to avoid.
Why this course feels slow
On purpose. This course is the fundamentals: you make every picture and every clip by hand, so you know what each step does. In the next course, Editing video with Claude, Claude and OpenDirector do these steps for you, and making a video feels like magic. But you can only steer that magic if you know what's underneath. So use the modules ahead to practise: first frame to video, reference to video, every step yourself.
Quick answers
What is AI slop?
AI video that looks fake: faces that change, plastic skin, warped labels, stray text, random camera moves. It comes from leaving decisions to the model.
Will a better model fix slop?
Rarely. The top models are close to each other. A clear process and good input pictures matter far more.
Why do the pictures matter more than the prompt?
The model copies faces, products and places from your pictures into every frame. A wrong picture gives a wrong video, whatever the prompt says.
Keep going
