AI graph generation I believe most of them will be, at present, a lot of online cue words or intelligent body, as long as a simple text description can come out. Then the graph-generated video ofdescriptorsHow to write differently from others. The same picture, others vivid and natural movie level effect, their own generation of video motion amplitude single.
Today we're going to learn how to use AI to help us write descriptors for videos of pictures to achieve director-levellens handling techniques.
The tools are used as follows: Beanbag, or Dream AI
Practical Demonstration
step by step
Step 1: Generate a picture of our demo
Here we generate a demo image directly using Beanbag, using Beanbag Image Generation Super Creative 2.0.
Enter, the prompt word for the screen we want to create:
For example: to generate eight Ghibli style animation, colorful and bright, two girls talking in front of the desk, the two are very happy, the ratio [4:3] (of course, the style we can customize, scene customization, ratio customization)
This gives us 8 renderings, choose the one you feel most satisfied with and download it in HD without watermark.
The finished picture is as follows
Step 2: Pic to Video
When you get the finished picture, I'm sure most of you do the same thing, just throw it to i.e. Dream or Korin and write a few simple motor words to go to the raw video. Roughly, they will look at the picture and talk, e.g. 2 girls talking to each other, reaching out and touching their heads, smiling, etc.
But today, in a different way, let the AI go and generate the director's subplot for us.
Continuing with Beanbag, open the Beanbag AI dialog and upload our finished image. (Note: open Deep Thinking)
Input: this picture as the first shot, to generate a multi-camera, shooting the girl is very happy video, the requirements of four coherent sub-scene, focusing on the expression of the two people, sub-scene script structure: the scene, the point of view of the operation of the camera, the content of the screen, the camera switching the fixed hard cut with a concise word can not be more than 400 words (this demand we can modify according to their own creativity)
Hold on a second. The descriptors for all four subplots are output.
Subtitle 1scenery: Medium Viewperspectives: Fixed lensScreen content: A classroom scene with a brunette girl in a sailor suit and a brown-haired girl in a pink dress sitting opposite each other, the brunette girl with bright eyes and a grin, the brown-haired girl responding with the same grin, books and coloring pencils neatly arranged on the table, and the trees swaying outside the window.Camera switching: Hard CutSubtitle 2scenery: Close-upperspectives: Fixed lensScreen content: Focusing on the brunette's face, she smiles showing her baby fangs, her eyes curving into a crescent moon, and her cheeks flushing a light red, vividly conveying joy.Camera switching: Hard CutSubtitle 3scenery: Close-upperspectives: Fixed lensScreen content: Turn and cut close-up of the brown-haired girl, her brow is stretched, the corners of her mouth are raised wide, her eyes are full of smiles, and her facial expression is subtle showing happiness.Camera switching: Hard CutSplit-screen 4scenery: Medium Viewperspectives: Fixed lensScreen contentThe two return to the center view, the brunette smiles and claps her hands, the brunette leans forward slightly and smiles with dimples, the interaction is natural, and the atmosphere in the classroom is relaxed and pleasant.Camera switching: Hard Cut
With the descriptors in place, we're next in going to generate the video using either Instant Dreams or Korin.
Here is the instant dream demo used:
Open Instant Dream AI: https://jimeng.jianying.com/
Click on the video to generate
Click to enter a description of the prompt words below
Click on the model to select Video 3.0
Enter the descriptors for the four subplots we split into above:
start generating
Learn to use these descriptors AI writing application and you too can write accurate video generation prompts, so go ahead and give it a try.
Test points: AI write the description of the cue word is direct director level sub-shot transport lens, suitable for a variety of characters, scenery, animals, etc., so we due to do the beauty play or short drama class is very good to use, of course, out of their own manual and constantly tuned to optimize.
Practical difficulty: ★✰✰✰✰
Tools and methods have the last is also a test of perseverance and creativity, the lack of one can not~