I'D LIKE TO KNOW HOW TO IMITATE REAL PEOPLE LIKE THISsculptureProcessShort VideoHOW? IT'S ONLY THREE STEPS TO COMPLETE, AND THE TOOLS USED ARE WELL-KNOWN AI SOFTWARE, AND NEW HANDS CAN FOLLOW. HERE'S A DETAILED BREAKDOWN OF THE VIDEO PRODUCTION PROCESS:
I. AI TOOLS USED
I.e., dreams:
https://jimeng.jianying.com/ai-tool/home
MOTION AI:https://www.1ai.net/44135.html
CLIP (PCEND)
Detailed production steps
Step 1: Generate a "watercoloring process spectrometry"
The core is the generation of five coherent lenses from "Blank to Text" to ensure consistency of character and perspective, as follows:
1. PREPARE REFERENCE DIAGRAMS: FIRST, FIND A BLANK PAPER (E.G. A BLANK A4 BY CAMERA) AS AN INITIAL REFERENCE DIAGRAM。
A little tip: If you want the next background to be more neat, it is suggested to use a white picture directly; if you want to add a scenery, you can also take a "blank paper with a good background" (e.g., on a desk, on a gallery) to avoid the need for additional adjustments to the subsequent background。
2.Generate a mirror in a dream:
Open a dream, click on [photogram generation], upload blank paper as a reference map, model selection of picture 4.0, scale 9:16, enter a hint: start with this map, generate five spectroscopy images under the same lens, from blank drawings, threadworks, initial graphics, fine-screeching to the written five processes, and maintain a person's image and lens perspective, with the theme of drawing a little girl's watercolor portrait and writing reality style
The dream generated five images:
3,Background adjustments (optional)
If the background of the initial reference map is not satisfactory (e.g. the blank paper I took with the edge of the computer table and the generated background is automatically expanded), the function can be optimized by "background replacement":
Upload a reference chart that needs to be adjusted, enter a hint: replace the background with a flower rack, wrap the drawing paper, prepare the painting
Enter a hint: Change the background to a frame, wrap the drawing paper, prepare the painting, shoot the front
Once a satisfactory frame background map is generated, the image is regenerated using the spectroph above to ensure a more consistent scene。
STEP 2: GENERATE A "FIRST-END FRAME ANIMATED CLIP" (I.E. DREAM / START AI 2 OR 1)
With five spectroscopys, it is necessary to make a consistent animated clip of the adjacent two, which provides two ways:
Method 1: Created from a dream
Download all spectroscopys, open the first frame of the dream, upload "two specs back and forth" each time (e.g. the first blank paper + the second line) and enter a hint: draw quickly in a slow step-by-step manner with delayed filming, and paint a little bit more naturally with a touch of the pen, and do not exceed the pace of drawing
Duplicate, make five spectroscopys of four consistent animations。
METHOD 2: GENERATE WITH A PRECIPITOUS AI (IN CASE OF INADEQUATE DREAM POINTS)
MOMENTUM AI IS CURRENTLY AVAILABLE FREE OF CHARGE AND HAS BEEN DESCRIBED IN DETAIL:
MOVE AI ONLINE! THE WORLD'S FIRST BINARY/MARTICULAR AI, FREE OF CHARGE AND UNLIMITED, WILL BE USED FOR LOGIN
Similarly, the adjacent lens map is uploaded as the front frame, the animation generation function is selected, the parameters are adjusted and the fragments are generated, and the images generated by the test are fluid and meet basic needs。
LET ME SHOW YOU THE FIRST BIT EFFECT I'VE PRODUCED WITH A MOVING AI
Step three: Cutting into film
The last step is simple, and the process of making videos at peace is the same:
- OPENS THE CLIP PC END, IMPORTS ALL GENERATED ANIMATIONS, IN ORDER。
- Cut the extra part based on the length of the clip and add a switch to make the image more natural。
- Add background music, adjust the volume。
- CHECK THE CORRECT AND EXPORT THE VIDEO, AND THE RESOLUTION RECOMMENDS A SELECTION OF 1080 P TO ENSURE THAT THE PAINTING IS CLEAR。
The whole process came down, there was no complex operation, and the focus was on "Coherence of the spectroscopy" and "Natural transition of the animations." If you're interested, try it