A LOT OF PEOPLE HAVE SEEN A BEAUTIFUL IMAGE OF AI, AND THE FIRST REACTION IS TO ASK, "HOW DOES THIS HINT GO?"
But when you really do it for a long time, you'll find that it's not a magic hint, it's a creation that can be repeated。
ONCE THIS IS DONE, YOU CAN MAKE YOUR OWN AI Skill.
Its role is simple:
YOU DON'T HAVE TO RETHOUGHT THE HINTS EVERY TIME, AND YOU DON'T HAVE TO REPEAT THE STYLE, CHARACTER, CAMERA AND IMAGE REQUIREMENTS. JUST HAND OVER YOUR CREATIVE HABITS, YOUR WORK CASES AND YOUR USUAL TIPWORDS TO AI, AND IT WILL HELP YOU SORT OUT AN EXCLUSIVE PROCESS. FROM NOW ON, IF YOU WANT A PICTURE, A VIDEO, YOU CAN LET IT LEAD YOU THROUGH。
This section does not speak of complex theory, it does not speak of white: how to make its own Skill, how to test, how to modify it, how to use it in the AI and Tusheng videos。
I. Why now, Skill
WE USED TO CREATE IT WITH AI, AND MOST OF THE TIME IT WAS "WRITING A HINT."。
If you want a good character, you have to write about it。
If you want an advanced picture, you have to write light and a picture。
You want the video to be natural, you have to write about camera sports and character moves。
Each time, it takes time and results are unstable。
The same idea, to write one effect today, to change the story tomorrow, may be completely different。
SO THE BETTER WAY TO DO IT NOW IS NOT TO KEEP A NOTE OF THE WORD, BUT TO FIX A COMMON METHOD AND LET AI WORK FOR YOU。
That's what Skill means。
You can read Skill as an "exclusive creative assistant."。
It is not a single reminder, but a complete creation process。
You tell it what style you want, it'll break it down; you give it the case, it'll summarize the pattern; you upload pictures, it'll help you analyze the picture and then generate video tips。
WE USED TO RUN AI。
Now it's more like we trained an assistant who knows his style。
These images are not handwritten. They're created by Skill
ii. Making Skyll, it's not as complicated as it looks
When they heard about Skill, many people thought it was difficult to understand code, model and technology。
Actually, no。
The most important thing about doing Skill is not technology, but to make your needs clear。
For example, you want to make an old-fashioned image of Skill, then you tell AI:
"I'd like to do a Skyll to create images of ancient visual figures. The images will include Eastern aesthetics, dress details, natural expression, film-film, and will continue to generate video alerts.”
That's a lot clearer than "help me make a living Skill."。
BECAUSE AI DOESN'T KNOW HOW TO WRITE, BUT NEEDS TO KNOW WHAT YOU REALLY WANT。
The more specific your description, the easier it is to imitate your style。
In addition to stating what needs to be done, you can prepare some of the work you've done before。
For example, you're happy with pictures, you're familiar with graphics, you're familiar with video。
THE MORE THESE THINGS, THE EASIER AI IS TO KNOW WHAT YOU LIKE。
It draws a lot of patterns from it:
What's the quality of the person, what's the style of the costume, whether the picture likes cold or warm, whether the camera likes a view or a medium, whether the video should be larger or lighter。
Once these patterns are sorted out, they become the rules in the Skill file。
So the core of Skill's production is not "doing a lot of work," but let AI understand your creative habits。
iii. Let AI help you generate a Skill document
When you get your needs and materials ready, you can get AI to help you generate Skill documents。
You can just say:
"Please help me generate an AI Skill document that can be used directly according to my needs, work cases and commonly used tips. This Skyll is designed to guide users in the selection of the person's appearance, personality emotions, image style, and then generate graphic tips. After uploading a picture, the user will also be able to analyse the picture and generate video alerts
AI generated the document, your initial Skill version。
This document usually contains several core elements:
It'll explain what this Skyll does。
For example, it is appropriate to generate ancient figures, visual images, graphics and videos。
CONTENT OF AI OUTPUT
It will then prescribe the use of the process。
For example, you ask the user what person you want, then you ask the person's emotions and scenes, and you end up with a complete hint。
It also collates the texture format。
For example, the biographic hints need to include characters, costumes, scenes, light, graphics and graphics。
Video alerts must include the movement of the person, the motion of the lens, the change of light and the requirements of stability。
With this document, you don't have to rephrase your style every time。
ONCE IT'S UP TO AI, AI WILL MAKE YOU CONTENT ACCORDING TO THIS RULE。
Here's the thing:
The first generation of Skill may not be perfect。
It's just an initial version, a really good Skill, must have been made of it。
IV. The real key is testing and modification
A lot of people stopped at the moment they made the document。
But actually, Skill made it up after the test。
Because a document that looks complete does not mean it actually works。
After you uploaded Skill, for example, AI, which can produce a graphic hint for you, has a good sense of AI: people are too fake, their skin is too plastic, they're not very sophisticated, and they lack high quality in commercial video。
This situation suggests that the style requirements in the hints are not clear enough。
YOU DON'T HAVE TO DO IT AGAIN AT THIS TIME, JUST TELL AI THE PROBLEM。
You can say:
"Now this Skyll has a problem, generating pictures of AI that are more sensitive and less advanced. Please help me revise the tone of the hints, with a focus on commercial visual senses, advanced antiques, real person skins, fine clothing material, film-class photo and natural camera languages, so that the images are more like mature visual works than normal AI pictures.”
AI will help you to add these requests to Skill。
For example, it is not sufficiently sophisticated in graphics, which makes it possible to strengthen antique materials, embroidery, hairdressing, fabric level and make-up。
If a person looks too fake, it emphasizes the true skin quality, the natural expression, and the true five。
If the picture is too flat, it increases the film's luminous, front- and back-story, camera depth and commercial sense。
This is the most important place for Skill:
It does not end at once, but can be adjusted。
The more you use, the more you know what's bad。
The more you change, the closer it gets to your creative habits。
V. Completion of living maps and graphic videos with Skill
When Skill adjusts, it can be used officially。
You upload Skill and then enter, "Put this skill on," and AI will start to lead you in the direction of the image。
For example, the person is a man or a woman, the temper is gentle or cold, the mood is quiet or firm, the scene is in the yard, the palace or the house, and the image is warm or cold。
You just have to answer those questions。
You don't have to write your own full text。
AI QUESTION AND ANSWER SESSION
AI WILL SORT OUT A HINT FOR THE GRAPHIC TOOL ACCORDING TO YOUR CHOICE。
You can copy it to Banana, or dream, or gptimage 2.0, or some other biographic tool。
WHEN THE PICTURE IS GENERATED, SEND IT BACK TO AI。
And then Skill will continue to analyze the content of the pictures, such as the person's actions, the expression, the light, the image and the image, and then write you a video message。
prompt:
Please generate a video of the 5-second classic drama, not recreate it, not change the original person's face, five officials, clothing, hair, makeup, headdressing, background, image and overall style。
It's like selling flowers at a fair, and suddenly someone's staring at her from the other side of the crowd. She did not escape immediately, but in instinctively kept her emotions under pressure, still standing on the ground with her basket of flowers, calm on the surface, with her eyes full of vigilance and experimentation。
At the beginning of the video, the person kept his or her head on the left side of the picture, with his or her eyes on the left side of the picture, and the pupils moved softly, as if she had confirmed something; her eyelashes shivered slightly, her lips tightened slightly, her jaw squeezed softly, as though the alert mood had just appeared and she had been pushed back. Then the character naturally made a small move along the original lines: Her fingers were tightened with the flower basket, which she was very softly protected in her arms, and her body stopped half a minute behind and the eye remained vigilant towards the side. It's medium and small, but it's visible, it's not sudden, it's not a performance。
AT THE END OF THE EXERCISE, THE PERSON'S MOOD SANK, DISPLAYING RESTRAINT AND VIGILANCE, AND THE LOOK OF THE PERSON WHO WANTS TO TURN AROUND AND LEAVE, MUST CONTINUE TO ACT LIKE NOTHING. THE ENVIRONMENT IS A LITTLE BIT OF RESTRAINED: THE MARKET IS FAR AWAY, THE WALKER IS FADING, THE PETALS AND TWIGS IN THE FLOWER BASKET ARE SLIGHTLY SHAKEN BY THE WIND, THE SIDE-TO-SIDE LIGHT IS ON THE EDGE OF THE HAIR AND THE SLEEVES, AND THERE IS A LITTLE DUST IN THE AIR. THE CAMERA IS A FIVE-SECOND CLASSIC FILM SENSE, SLOW-PUSH OR LIGHT-SIDE MOVE, SHALLOW VIEW, WITH A FOCUS ON THE EYES OF THE PERSON, MICRO-EXPRESSIONS AND THE SMALL PLAY ACTION OF HOLDING THE FLOWER BASKET. THE WHOLE PICTURE IS LIKE A SECOND OF AN ANCIENT PLAY, NOT A FAKE, NOT A ROLE SHOW, NOT AN AI STAND-BY。
[negative hints]
Don't recreate new characters, don't change face, don't change face, don't change hair, don't change hair, don't change make-up, don't change face, don't change context, don't change scenery, don't change angle, don't turn into a picture, don't pose, don't wait, don't blink, don't move, don't dance, don't exaggerate, don't deform, don't deform, don't greasy skin, no plastic face, don't wear skin, don't wear hair, don't deform, don't melt, don't melt, don't flower, don't melt, don't melt, don't melt, don't melt, don't melt, don't melt, don't melt, don't float, don't float, don't rush, don't move, don't move, don't move, don't move, don's, don's, don's, don's, don's, don's。
The greatest benefit of this process is smooth。
This hint is much better than simply saying “let the person move”。
Because it is more specific and easier to control the stability of the picture。
wind up
So, Skill wasn't as complicated as I thought。
At the heart of it is to organize your creative needs, the style of your work and the usual hints into a set of methods that can be used repeatedly。
AFTER DOING WELL, IT'S LIKE AN AI CREATIVE ASSISTANT WHO UNDERSTANDS YOUR STYLE。
You want a graphic, it helps you organize the hints; you want to make a video, it helps you analyze the pictures and generate video tips。
It doesn't have to be perfect from the beginning。
The basic version, then the effect changes, and Skill gets better and better。