Often a small group of people question thatDreamofLip SyncDigital people can only do 15 seconds yikes, so what if my audio is longer than 15 seconds? This article today teaches you!
Whisper a word, this tutorial a lot of details are very dry Oh (I can say that compared to some paid tutorials) ~ for example, my friend said that the dream generate digital people is very costly points, today's tutorial is to save points version, and the picture more vivid ~
Let's take the following case as a demonstration, jittery voice search "emotional rambling", come out all these videos ~
this kind (of)Emotional VideoRecent traffic is good, the video content is a pair of men and women, the woman cried to tell, said countless netizens, so the point of view of the collection of shivering is not bad ~ casually cut off a picture, 6 works have 4 more than 10,000 likes.
Tools needed: instant dreaming + projection cutting
The production process is in 2 steps:
→ ①Image generated by AI
→ ② i.e., dream number people lip-synching
Without further ado, let's get on the practical, and strive to be the simplest way to teach everyone, white can easily get started!
First, that dream AI generates pictures
That is, the Dream Computer web page:
https://jimeng.jianying.com/s/iyrmHgaU/
Mobile app store search 🔍 "Instant Dream AI"
Open and DreamAI, select "Image Generator" - Enter the "prompt word" below - Ratio 3:4 - Model "Image 3.0pro" - Generate now!
A pair of men and women, the woman occupies the main position, the woman looks at the man, the woman has long black curly hair, fair skin, wearing a white shirt chiffon, with tearstains hanging on her face, looking at the camera head-on, the girl's eyes look at the boy, her eyes are full of aggression and helplessness; the man is wearing a dark shirt, his back is turned to the camera, and the boy is in the foreground, occupying a very small position on the right side. Rainy night, city neon. 4k HD, realistic, digital illustration style .
Four diagrams will be generated at once, if you are not satisfied you can adjust the cue words appropriately and continue to generate again. Save the pictures you need for backup.
That is to say, the pictures or videos saved directly by Dream will have watermarks, and you can refer to this previous article of mine for the method of removing watermarks.
Pure dry goods! That dream video to watermark method to share, that dream AI raw pictures without watermark download, cell phone version of the computer version have, come to learn! (Super detailed tutorial)
II. i.e., dreaming of human counterparts
Digital people lip-sync, that is, make the character in the picture talk or sing. We have to prepare a voiceover first to make it easier for us to create a lip-sync video.
◇ Audio preparation:
The method I used for the audio part this time is to go directly to download the popular video and use the audio from it directly. I.e. dream lip-synching is only supported for 15 seconds, so we're going to go ahead and make a cut to the audio. This is the original video I downloaded, it's 19 seconds long.
Open the cutscene.
Import the video clip you just downloaded
Tap the cut at a location within 10 seconds of the video.
Delete paragraph 2 and export audio 1
Return to cut, step back, delete segment 1, export audio 2
This way we have the audio we need for this time.
◇ Use pictures to lip sync.
Next, it's time to get this woman to talk.
Open and DreamAI, select "Digital People" - Upload Image - Upload Audio 1 - Generate Now
[Costs 45 points]
After a few moments, we made our first video.
Boys talking along? It's okay, when editing, we can key out the picture boys - picture in picture, masking can be. The method is as follows (cell phones can also be operated):
Do the same for the second video [cost 50 points】
We went to cut the projection to put the two videos together with subtitles, rain sound effects, and a foreground rain effect.
A complete video is ready (if you're a RMB player and don't care about points, you don't have to scroll down!)
I thought you said you were teaching a points-saving version of the tutorial? This one video cost 45+50=95 points!
Next is the point!
This previous picture goes directly to the mouthpiece, is the method most people use, many careful partners will find that the middle of the spliced video will be stuck frame, frame skipping, this problem will also be solved here.
◇ Use the video to lip sync.
Going back to i.e. Dreams, let's start by turning that image into a video. How long the audio is, how long the video needs to be. The audio I've prepared is 19 seconds long, but Instant Dream can only generate videos up to 10 seconds long, so we're going to generate two videos.
Open and DreamAI, choose "Generate Video" - enter the prompt word below - upload image - choose 10S for duration - generate immediately [Costs 20 points].
The camera wraps around counterclockwise as the girl looks to open her mouth to speak, nodding her head from time to time. It's raining, the girl's hair strands flutter gently in the wind, the girl's tears slide gently down her face, simulating the frequency of a real person's closed eyes, the girl's hand holding the boy's arm
The first video is generated, let's go to the second video. Be careful here!
To eliminate stuck pins in the middle of the video, we can't use that image to generate the video anymore. Instead, we'll use the last frame of the first video to generate the second~
Open Cutscene and import this video above
-Timeline dragged to the last frame, upper right corner -Export frames
And we get this still image
Using this image as a base, go ahead and generate a second video
Enter the word below - Upload a picture - Choose 10S for the duration - Generate now [Costs 20 points].
The camera wraps around clockwise, the girl speaks and looks at the boy with her eyes open, simulating the frequency of a real person's closed eyes, the girl is sad and sheds tears, her hair is blown by the breeze, the camera focuses on the girl, the camera zooms in slightly
After generating it, drag the two videos together to the cutout, align our two audio clips, delete the audio, and export them separately.
Export the following:
Now use these two videos to lip sync! Upload video 1 and audio 1 to generate.
[Spending 4 points]
Do the second segment in the same way, uploading video 2 and audio 2 to generate.
We went to Cut Screen to put the two videos together, adding subtitles, rain sound effects, etc. Look at the finished product:
Let's do the math on how many points the video lip-sync cost, 20+20+4+5=49 points.
Saves almost half the points and has a more vivid picture~
Did you learn it? Go ahead and try it.