I was a virtual idolMVIt's a good feedback. I'm sure you're starting to practice
I was asked in the comment section about the mouth..
Honestly, it's just gambling。
It's crazy to draw a card and check the part that matches the lyrics, but it's still not right。
BUT THESE TWO DAYS! I'VE CHANGED MY MV TO CREATE A WORKFLOWSekoPerfectly solved the problem
This time, I've spent only a third of my time doing better and more efficient
Come and update the curriculum
In short, Seko is an Aigent from the merchant soup that can be used directly without downloading:https://www.1ai.net/39585.html
They've been upgraded to support the three-person mouth-to-mouth of the same image, and they've got a mirror canvas on it, and it's easier to modify it
I STARTED WITH A MAN AND A WOMAN SINGING MV, AND I GOT A REALLY GOOD RESPONSE FROM THE GROUP:
THIS LESSON WILL FOCUS ON MEN AND WOMEN SINGING MV CASES, WITH MORE COMPLEXITY AND MORE OPERATIONAL DETAILS
THE VIRTUAL MV 2.0 IS IN THE SECOND PART, SO IT CAN BE STRAIGHT DOWN
I'm not saying much, babysitting
MEN AND WOMEN SING MV
Today we focus on how to use Seko to produce good music for us。
My work stream is divided into the following parts: identification of the subject — giving the subject and the lyrics to Seko — confirmation of the lens — using the Seko mouth-to-mouth — generation of other videos that do not require the mouth-to-mouth type — export of video material — video editor re-alignment (reordering the image/tracking/mixing/showing/subtitting)。
If there is no subject, you can try to give Seko this direction:
In Seko, we have two options for making videos:
1. No subject, choice of style: the video is written throughout, and consistency is easy to disperse and less suitable for long video
2. Fixed subject, with no selection of style: The video will be generated in accordance with the subject of reference given, with a high degree of consistency, but if the material given to the subject is not good enough, it will also affect the production effect。
Seko's biographic model is quite the right thing to do, and I picked the style and put the lyrics on it, and I created the right man and woman on the page。
Of course, this step is not necessary, if you have the chosen character set to create the subject directly。
Now, we go back to Seko's front page, create these two subjects, and then update the hints, and bring the two protagonists to generate the video:
Click to send and come to the planning page。
Seko will automatically create fine art styles, scenes, set music styles, and generate dozens of spectroscopes by lot of words:
There are two points to check:
ONE, THE VOICE AND THE LINE ARE GONE: SOMETIMES AI WILL MISS THIS PART OF THE COMMAND. IT'S OKAY. YOU CAN ORDER IT OUT IN THE LOWER LEFT WINDOW
Two, you can choose the biographic model before creating the spectroscopy: when I make the film, I choose dream 4.0, and Nano, Seko IDX, Flux.1 Kontext Pro, you're interested in trying it yourself
I put the whole song on it for two minutes, and Seko directly generated me 77 spectroscopes -- the more you get, the bigger the image space you're going to have。
My video actually makes only half a song, and a few more lyrics can give me more relevant images, saving me time to write my own tips and to draw back and forth。
When you enter the video editor, you can see that each lens is currently in the form of a picture。
The system sometimes automatically produces subtitles, and if the last step does not remove the symphonies, there will also be a symphonies, which will close at two separate points:
Now, let's check if the picture has anything to change。
For example, there's a light on the hostess' face, and I'd like to take it off
The following are some of the reasons for this:
There is also support for local reprogramming, element additions, cropping and what to change。
You can also directly recreate the hint:
To make it easier for us to be broadly on track, here we are going to pass on the songs that have been produced:
NOTE THAT ONLY UP TO 5MB AUDIO FILES ARE CURRENTLY SUPPORTED。
An online search of compressed websites, many of them free of charge。
And then we're going to work on the mouth-to-mouth, and here we're going to use SekoTalk, the trade-to-suppos model。
On top of the photo spectroscopy, there are options for the mouth type to enter the edit page:
Cut the audio clips of the video, upload -- add -- fill in the image description -- and generate the video:
The same is true of the multi-person mouth: select a two-person symmetry image, upload the audio portion of the two-person performance separately and reorder it in the track:
For videos that do not require mouth-to-mouth, confirm that the spectroscopy material is fine and can be used at the left side of the video。
Here Seko will automatically write a video to generate a hint based on the spectroscope content, and you can also adjust the length and model of the video in this area:
REPEAT THE ABOVE STEPS AND TURN ALL SPECTROSCOPY MATERIAL INTO VIDEO, EXPORT IT AND ALIGN IT WITH THE ORIGINAL ARCTIC AXIS, REORDER THE IMAGE, SUBTITLING AND TRANSMUTATION, AND DOUBLE MV
VIRTUAL MV 2.0
AFTER EXPLORING THE TWO-PLAYING, I STARTED TO UPGRADE MY VIRTUAL MV
The whole step is the same
1. Introduction of a multi-person perspective
The 1.0 program has already taught you how to create images of people with a high degree of consistency. Once Midjourney has created this original iconic role, we can get these from Nano:
Heads close
Side half
All over
I observed that the logic of the creation of the Seko main body was to regenerate the spectroscopy, and the idea was that it was possible to enrich the spectroscopy perspective by making the material of the individual role the subject。
And because of the fact that the material came out of Nano, there was a basis for consistency in itself, so the whole video could be very consistent。
Creates the subject, as in the previous part, and quotes in the hint:
Attention, the body mass I use here is not very high, it's a little blurry, and people make up only one tenth of the picture, which leads to a deterioration in quality when I do the whole body video。
So you can do more on the raw materials and try to use the high-level materials that make up a higher percentage of the images。
2. Include detailed description of the mouth type
Although there are selected entries on the page, this description is valid:
It'll make your hair dance
3. Video clip acceleration
When you import video editing software for secondary processing, you can speed up the selected segment, and I will add it to 1.5-2 times the speed of the video/syllable card point。
SOMETIMES THE AI MIRRORS ARE A LITTLE SLOW, JUST AS FAST, AND IT'S MORE REAL。
Four, switch with more lenses
Generating a single image in Seko is by default five seconds, and more material is needed to fill it, as it is more sensitive to the high-intensity MV to switch mirrors。
You can choose to enter more lyrics than I just did, find appropriate material to fill in the front, or you can start another project in Seko to create a spectroscopy with the same hint, and you can pull up your video production efficiency at speed。
Final Thoughts
The day before the article was published by Veo3.1 and OpenAI launched a stronger Sora2。
At first, I saw that they were actually a little anxious..
ISN'T IT A LITTLE AWKWARD FOR ME TO DO THE MV THING AT THIS POINT
But after the case, I realized, not at all。
My workstream, my curriculum, still worth it。
The current maximum for automatic video generation is 25 seconds, and it takes $200 a month for official channels for ChatGPT members to run。
IT'S ALMOST IMPOSSIBLE TO USE THESE MODELS FOR A GIVEN SONG WITH A MV -- THEY ALL HAVE THEIR OWN SOUND, AND WHERE DO THEY MATCH MY MUSIC。
In addition, the creation of a digital divide in Chinese is a persistent problem for foreign models。
And Seko now has this video Agent program, which has access to multi-familitural model + video model, which is presented in the most interactive way possible to make video, so that I can do it more effectively。
The path that I have explored today will certainly help many to solve their problems。
We were talking about a circle, as if all the tools were in trouble now, and one-stop platforms would have to draw cards before we could get a long video like I did before。
Now, with Seko, you can make a video。
Of course, if you want better results, you need to learn some adjustment techniques, such as those mentioned in today's article。
BUT THERE IS A COMBINATION OF THESE TOOLS, WHICH IS REALLY MOVING THE INDUSTRIALIZATION OF THE AI COMICS, AI MV AND AI SHORTS TO REALITY, AND PRODUCTIVITY CHANGE MAY CHANGE THE COURSE OF MANY LIVES。
THAT IS WHY I WANT TO UPGRADE THE MV GENERATION TODAY。
It makes no sense to do an AI content without control, without a real scene。
so i don't want to be the one who just yells "xx out the bull" and i want to make you something really useful。
Slow down, but I hope these elements have long-term value。
So, if today's content is useful to you, it's great