This is the most detailed version of Seedance AI Video TutorialOh, shit! Just startedAI VideoIT'S A MISTAKE TO THINK THAT A.I. PRODUCES A NICE SPECTROSCOPE OR A SHOT, AND IT'S A GOOD THING? IT'S OKAY TO COLLAGE DIRECTLY, AND THE RESULT IS THAT NO ONE WANTS TO SEE THE VIDEO, THE SCENES GO OFF, THE CHARACTERS GO THROUGH THE GANG, AND THEY CAN'T EVEN WATCH IT。
TODAY, WITH THE BIG WHITE WORDS AND EVERY STEP TO TEACH YOU HOW TO START WITH A FULL-FLEDGED VIDEO OF THE AI FILM. YOU CAN ALSO BE A DIRECTOR。
THE FOLLOWING STEP IS TO TEACH YOU, AND WHITE CAN DO IT ONE BY ONE: IF YOU WANT TO BE A DIRECTOR, IT'S FROM 0 TO 1. ALL OF THIS WILL MAKE AI GENERATE THE HINTS。
ONE: IDEAS GENERATE STORIES, AND THERE MUST BE AN IDEA OR INSPIRATION. PLAY IMAGINATION, OR LET AI GIVE YOU ONE OF WHICH YOU CHOOSE FROM. BUT I SUGGEST THAT WE THINK OF OURSELVES。
i'm not inspired. but there are references. it's the simplest, inverse. the video understands the story with manus, and the picture with the bean bag/dollarbow. one for the hint。
2: How to handle role design. 3: How to get the video emotional, camera and rerun
That's all you need to know, and then we're going to dream: there's no threshold。
1: AI NOVELS ARE RAPIDLY GENERATED. 2: AI FIGHTS QUICKLY GENERATE 3: SHORT PLAY METHOD。
Step 1: How to generate ideas? For example, looking at the opposite sex recently, with some inspiration, I thought maybe one of the women who failed to save her subordinates, worked hard, finally took off, and, uh, it was a very broad feeling, and I thought I could go a little deeper and define a general story。
For example, let me give you a little bit of detail: it is about the monologue of the heart: a female soldier who failed to save her friend, who was defeated, went to a deep mountain forest, where he refined, and decided to go back to revenge
I asked AI to help me perfect, and I used: gemi. Next: Enter a hint, according to my story: Generate 3-4 emotional beats. Of course you can generate more, 5-8 or 10-15。
YOU FEEL SOMETHING? THIS IS BASED ON AN IDEA THAT GENERATES A BROAD STORY STRUCTURE. SO WHAT'S NEXT
Turning every emotional clip into a scenario description: where is it, what is its main action, and what has happened to its emotional changes?
The effect: it should be very intuitive. It includes action, role and emotion. Then step two
STEP TWO: HOW TO HANDLE THE ROLE? IS IT NOW THAT SOMEONE'S GOING TO CREATE A PICTURE? NO, IT'S GOING TO KEEP AI ASKING QUESTIONS ABOUT OUR ROLE. ONCE AGAIN, WE USE A HINT:
Animation for this character: - Role in the story - character - physical characteristics - clothing and equipment - colour palette
See the effect: isn't it detailed and complete, but it's not too much, not too much, what should we do
LET AI STREAMLINE: ENTER A HINT: IT'S ACTUALLY A SIMPLISTIC POINT, AND IT'S NOT GOOD FOR AI TO PRODUCE SO MANY。
COMPRESS THIS TO A CHARACTER DESCRIPTION OF TWO TO THREE LINES FOR THE AI IMAGE. ALL KEY FEATURES ARE RETAINED。
LOOK AT THE EFFECT: IT'S DOWN TO JUST THE COLOR OF THE CHARACTER. DOES AI UNDERSTAND THAT? I CAN'T UNDERSTAND HOW TO FINE-TUNE IT. IT'S A FULL VERSION。
based on these characters, try to see the effect: i'm creating this role with banana pro. here's the thing: think what style? write the truth, the comics, i write sagepunk. and here's the cut:
Attention is drawn here to the high rate of failure and to the fact that you can find a picture of failure, or like me: failure
A search can be made: the data is based on a search: it's a nice suit, and it can be passed on to someone else. I smoked a lot。
here's a note: a stand-alone picture is that there's no way to make the details consistent. there will always be some deviation. or wear a gang, the best way to do it, or create a full rolesheet. how? i have a complete text that you can refer to: if you use gemini, translate it into english
A PROFESSIONAL CHARACTER DESCRIPTION FORM BASED ON AN UPLOADED REFERENCE IMAGE IS STRICTLY CREATED AND IS MAINTAINED AND FULLY COMPATIBLE WITH ITS ORIGINAL VISUAL STYLE. DO NOT CHANGE THE LEVEL OF WRITING OR RENDERING; MAINTAIN IMAGES OF THE SAME TYPE AS REFERENCE IMAGES ( FOR EXAMPLE, IF THE REFERENCE IMAGE IS THREE-DIMENSIONAL, IT IS THREE-DIMENSIONAL; IF IT IS PHOTOGRAPHIC, IT IS PHOTOGRAPHIC; IF IT IS PAINTING, IT IS PAINTING) WHILE MAINTAINING THE SAME OBSCURITY, TEXTURE, COLOUR AND OVERALL BEAUTY. • USE A SIMPLE, NEUTRAL COLOURED BACKGROUND TO MAKE THE IMAGE OF THE PERSON VISIBLE. THE DRAWING PAPER IS ORGANIZED SO THAT THE LEFT PART CONTAINS TWO LARGE-SIZE WHOLE-BODY PANELS, SHOWING THE EXACT ANATOMICAL STRUCTURE OR PROPORTION OF THE ROLE AS IT RELAXES IN POSITION A AND A CLEAR CONTOURS: ONE IS A POSITIVE VIEW OF THE WHOLE BODY AND THE OTHER A VIEW OF THE BACK OF THE WHOLE BODY, WITH THE MEASURE AND ALIGNMENT OF THE TWO. ON THE RIGHT SIDE, FOUR PORTRAIT PANELS WILL BE CREATED AND ARRANGED IN TWO ROWS: A POSITIVE PORTRAIT, WITH THE PERSON'S FACE FACING THE CAMERA; A LEFT SIDE, WITH A SIDE OR THREE-SIDED FACE FACING THE LEFT; A RIGHT SIDE, WITH A SIDE OR THREE-SIDED FACE FACING THE RIGHT; AND A VERY WELL-WRITTEN FACIAL FEATURE, WHICH CLEARLY DISPLAYS THE FINE SIDE OF THE EYE, EYELID OR EYELASHES, EYEBROW, EYEBROW (IF ANY), NOSE, MOUTH, SKIN OR SURFACE LINEN AND HAIR OR HEAD DETAILS. TO ENSURE THAT THE IDENTITY FEATURES ON EACH BOARD REMAIN PERFECT, SO THAT THE IMAGE OF THE PERSON ALWAYS APPEARS TO BE CONSISTENT WITH THE SAME BODY IN THE REFERENCE IMAGE, WITH THE SAME PROPORTION OF THE FACE IN THE THREE STANDARD PORTRAITS AND WITH THE USE OF EQUIDISTANCE AND CLEAR VISUAL SEPARATION TO DISTINGUISH THE DRAWING BOARDS. THE LIGHT EFFECT SHOULD BE CONSISTENT WITH THE ENTIRE PAINTING (IN THE SAME DIRECTION, STRENGTH AND DEGREE OF TENDERNESS AS THE ORIGINAL STYLE), WHILE THE SHADOWS SHOULD BE CONTROLLED TO PRESERVE DETAIL AND AVOID DRAMATIC EMOTIONAL CHANGES, SO AS TO PRODUCE A CLEAR REFERENCE MAP THAT CAN BE USED DIRECTLY FOR PRINTING, WITH CLEAR DETAILS。
Result: Bananas are best. Or else there will be a deviation. The effects are good。
STEP THREE: HOW TO GET THE VIDEO EMOTIONAL, CAMERA AND TURN AROUND, IN THE CONVERSATION BEFORE, JUST ASK AI TO CREATE IT。
You know my story, my role and these three scenes. For each scene, it is recommended that three to four of the best film footage to express feelings. For each shot, please tell me: the type of camera (widen angle, medium view, shoulder, close-up, insertion, etc.) the angle of the camera and the movement (low angle, high angle, handheld, static, slow-in and follow the emotions that the viewer should feel from the camera (single, guilt, calm, strong)。
WHAT IF THESE ARE TOO LONG? STILL THE OLD WAY, WE LET AI OUTPUT SPECIFIC HINTS。
Now that you've learned my story, my role and my three scenes, and you've suggested three or four film shots for each scene, including: lens type lens angles + the emotions to be felt by the sports audience, please convert each shot into a short hint that I can paste into an AI image or video tool. Every shot in my plan: keep the same type of camera you've chosen, cameras and emotions. Write a simple tip (1-2 sentence). It's about what we see, what we see, what we see, what we see, what we see, what we see, what we see, what we see, what we see, what we see, what we see, what we see, what we see, what we see. Don't invent new angles or scenes. Please write in simple, natural language suitable for Kling2.6 or similar image-to-video models, Higgsfield. Please organize your answer in the following format: scene 1-[situation name] camera A: [insert your original camera description] >Tip:[1-2 sentence AItip] scene 2-.
and then put our subject in the whisk, after all, free. so use it to generate. just keep smoking. i can't change the hint。
And then, based on the spectroscopy, you can generate video, you can use the Grok, and then you can draw a card, and of course you can use the Violin 2.0 character plus scene+action+ close-up+ lens, and you can output. That depends on your economic strength
And then when all the spectroscopes and videos are finished, they can be spelled. It's not like you know how to produce video quickly, from a full process of 0-1 to an idea of the idea, to a mirror, to a person's detail, that's a complete process of 0-1。
Here you gooh, goddance 2.0 GAMES: FOR EXAMPLE, AI NOVEL COMICS CREATE YOU WANT TO QUICKLY GENERATE A FIGHT PICTURE: TEACH YOU ONE MOVE, NOTHING, COPY A NOVEL: COPY THIS。
upload a character: enter a line into an instant dream, generate it with a feeddance 2.0
Look at the effect: it's strong, but you have to understand the principle. What if you change your action? Or what to do with the mirrors, look simple, but it's not that simple。
HOW DOES AN AI SHORT PLAY COME ABOUT
This is also an emotional reminder, based on the above method, of the person concerned and the spectroscope. There's no more line, and then there's @ in the dream. It's hard to understand. It's easy to understand。
Fight video generation? It's the easiest way to upload a reference video。
Methodology: A reference fight is used and the role is uploaded. The music has to be separated. It works
