{"id":58342,"date":"2026-10-20T09:37:48","date_gmt":"2026-10-20T01:37:48","guid":{"rendered":"https:\/\/www.1ai.net\/?p=58342"},"modified":"2026-09-20T15:33:40","modified_gmt":"2026-09-20T07:33:40","slug":"ai%e6%8f%90%e7%a4%ba%e8%af%8d%e5%88%9b%e4%bd%9c%e7%ac%ac%e4%b8%83%e5%8d%81%e4%b9%9d%e8%8a%82%ef%bc%9a%e4%b8%ba%e4%bb%80%e4%b9%88ai%e4%ba%ba%e7%89%a9%e6%80%bb%e6%9c%89%e5%81%87%e4%ba%ba","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/58342.html","title":{"rendered":"Section 79 of the AI tip: How do you write a Feedance 2.5 living tip? Why does an AI character always have a \"false sense\" of character, scene, and behavior"},"content":{"rendered":"

existingAI Video<\/a>IT IS ALREADY POSSIBLE TO PRODUCE VERY CLEAR FACES, NATURAL SKIN TEXTURES AND RELATIVELY COMPLEX PERSONALITY MOVEMENTS, BUT MANY WORKS STILL HAVE A PROBLEM THAT IS DIFFICULT TO IGNORE: TAKING ONE FRAME ALONE, THE PERSON DOES NOT SEEM TO HAVE ANY OBVIOUS ERROR, BUT ONCE THE PICTURE IS REALLY MOVING, THERE IS A CLEAR \u201cAI SENSE\u201d\u3002<\/p>\n

THIS PERCEPTION DOES NOT NECESSARILY STEM FROM INADEQUATE MODELLING CAPACITY, NOR DOES IT NECESSARILY BE RESOLVED BY ADDING WORDS SUCH AS \u201c8K, SUPER-HIGH RESOLUTION, FILM QUALITY, DELICATE SKIN\u201d. IN MANY CASES, WHAT REALLY AFFECTS A PERSON ' S REAL IDENTITY IS WHETHER THE PERSON ' S IDENTITY IS STABLE IN DIFFERENT LENSES, WHETHER THE PERSON ' S ENVIRONMENT IS COMPATIBLE WITH THE VISUAL EXPERIENCE OF REAL LIFE, AND WHETHER THE PERSON HAS ANYTHING IN THE PICTURE THAT MAKES SENSE\u3002<\/p>\n

IN OTHER WORDS, THE \u201cREAL\u201d IN THE AI VIDEO IS NOT JUST A QUESTION OF THE QUALITY, BUT A RELATIONSHIP BETWEEN A SET OF PEOPLE, THE ENVIRONMENT AND BEHAVIOUR\u3002<\/p>\n

IF AN AI VIDEO IS CREATED AS A VIRTUAL VIDEO, THE REFERENCE MAP IS MORE LIKE AN ACTOR AND A SITE, AND THE HINT IS ACTUALLY WORKING VERY CLOSE TO THE DIRECTOR. WE NEED TO NOT ONLY TELL THE MODEL WHAT'S IN THE PICTURE, BUT ALSO TO BE MORE CLEAR ABOUT WHO THE PERSON IS, WHERE HE IS, WHY HE DID SOMETHING, AND IN WHAT ORDER THEY HAPPENED\u3002<\/p>\n

IN THIS SECTION, WE ARE RE-ESTABLISHING A MORE HUMAN-FRIENDLY VERSION OF THE A1 MESSAGE FROM PEOPLE, SCENES, PROPS AND TIME AXES\u3002
\n
\nChapter I: Trueness is not a drawing, but a relationship<\/strong><\/p>\n

THE QUALITY ITSELF WAS AN IMPORTANT ISSUE AT THE BEGINNING OF THE POPULARIZATION OF THE AI VIDEO, SO MANY OF THE HINT TEMPLATES WOULD INCLUDE DESCRIPTIONS SUCH AS \u201c8K, HIGH-RESOLUTION, SUPER-REAL, FINE TEXTURE, FILM-LEVEL PHOTOGRAPHY\u201d\u3002<\/p>\n

These words are not completely ineffective as models develop, but they are already very difficult to be at the heart of determining a person ' s sense of truth\u3002<\/p>\n

When humans judge whether an image is real or not, they do not just observe whether the person's skin is clear enough. We are also unconsciously observing many relationships, such as the matching of people and the environment, the weight of physical actions, the lack of clarity of vision, the causal consequences between actions, and whether the person is not the same person after a different angle\u3002<\/p>\n

FOR EXAMPLE, WHEN A GIRL SITS IN A COFFEE SHOP, AND IF SHE KEEPS A STANDARD SIT-IN, SMILES AND LOOKS AT THE CAMERA FROM THE BEGINNING TO THE END, EVEN IF THE FIVE OFFICIALS ARE VERY SOPHISTICATED, THE PICTURE IS STILL EASY TO DISPLAY AS AN AI FIGURE\u3002<\/p>\n

But if she looks down at the glass on the table, stretches out her hand and holds the handle, takes a cup of coffee, then puts it back down, and then, because someone passes through the window, the person will immediately have more life\u3002<\/p>\n

So the real sense is not entirely from \u201chow perfect a person is\u201d, but from a lot of small, realistic details\u3002<\/p>\n

AND THAT'S WHY WE CAN'T JUST FOCUS ON THE FACE AND VISUAL STYLE OF THE PERSON WHEN WE'RE MAKING A REAL-HUMAN AI VIDEO, BUT WE SHOULD GIVE PRIORITY TO A FEW MORE FUNDAMENTAL ISSUES: WHETHER THE PERSON IS ALWAYS THE SAME PERSON, WHETHER THE ENVIRONMENT IS LIKE A REAL PLACE, WHY THE PERSON IS ACTING IN THIS ENVIRONMENT, AND WHETHER THESE MOVEMENTS TAKE PLACE IN A REASONABLE SEQUENCE\u3002
\n
\nChapter 2: Identify the person, then design the person<\/strong><\/p>\n

PERSONAL CONSISTENCY IS ONE OF THE MOST PROBLEMATIC LINKS IN THE LIVE AI VIDEO, BUT IT DOES NOT NEED TO BE OVERLY COMPLICATED IN THE ACTUAL WORKFLOW\u3002<\/p>\n

Usually we just have to prepare a four-view of the character (two person-story, one body-side, one back-story) as the basic reference material for the model to make it clear what the person looks like. In models such as Seedance 2.5, the reference map itself already has a strong capacity for identity locking, so the key is not to add more complex labels, but to clarify in the hints which role to use and which role to avoid confusion in multiple characters<\/p>\n

\"42f24430jlneg002ald000v901jp\"<\/p>\n

Four View Tips:<\/p>\n

The image of the person in the image is set in four views, the grey studio background, the overall make-up board for the professional human role, the layout, the clean, not the poster, not the big magazine, not the drama, not the complex scene\u3002
\nThe overall layout is strictly divided into two parts, top and bottom: the first half accounts for one third of the height of the picture and the second half for two thirds. The area above displays two fine face features, on the left, on the front, and on the right, on the side. Both features are focused on the head and the neck, and are well organized, with a clear display of the characters ' five officials, bone structure, make-up, hair, hairdressing and decorum. Personal facial features, hairstyles, make-up and head dressing must be strictly consistent with the reference image. The skin sense is real and natural, with porosity, stripes, fuzzy hair, eyebrows, nuanced hair and light on the skin, not grinding skin, not too sharp, not looking like a beauty, and a high-resolved make-up feel\u3002
\nThe area below shows two reference maps of body and clothing, arranged around\u3002
\nThe left-hand picture must be a larger, front-looking costume display with a standard A-pose position for the person. The focus of this picture is not on the whole body, but rather on \u201ca significantly larger number of people\u201d, so that the role is more proportional to the picture. The picture on the left side must be taken in a more proximate light, with the person starting to draw from the lower side of the shoulder neck, until the end of the shoe, and the head being completely cut off from the picture because the image was amplified. Please note that this is not an ordinary long-range body picture, but rather a positive one that is magnified at close range: the body is significantly larger in the picture, with the upper and lower parts being magnified and the overall visual effect being the main part of the dresser in the time frame. The person ' s shoulder, arms, hands, torso, waist, legs and shoes must be preserved in their entirety, and the edge of the image allows for only the head to be cut off and the hand, foot or body to be cut off. This figure on the left must look closer, more \u201cfull\u201d and more \u201cexpanded\u201d than on the back of the right to make the details of the dress and the body more visible. It is prohibited to make the left part of the body as a general small scale, to make the person too small in the left side, to keep the head in the picture and to cut off only half the head or leave a chin, mouth, nose, edge of the hair\u3002
\nThe right side of the picture is the entire body of the back, standard A-pose, which presents the whole body from the head to the back of the shoe. The person's back-to-back lens, with complete back, hair, back structure of the clothing, shoes and overall contour, is used to display the role-back shape. The right side of the picture is shown on the general whole back, with a normal figure ratio, which does not need to be magnified as the left\u3002
\nThe two charts below must be the same person, the same dress, the same set of accessories, the same body ratio. The entire body is maintained at a permutation rate, but do not exaggerate. Clothing design, fabric level, shoes, dress style, clothing structure and costume styles must strictly follow the original map. All clothing and fittings are characterized by a sense of true physical material quality, with natural wrinkles, crumblings and thin changes, with real reflections of metals, and with real natural expression of materials such as leather, silk and veils. If the original dress is a long skirt with a fork, the person on the left side can walk half-step by leg in order to show the fork structure and the effect on the leg, but the whole body will still need to remain in a stable position with a make-up and not a show\u3002
\nThe light is only polished using a three-point clean and integrated studio: the main light on the right, the auxiliary light on the right and the contours on the right. Don't mess with the light, don't color the light, don't get the vibe. The background is pure gray and seamless\u3002
\nStrictly require:
\nTHE TWO FACES ABOVE ACCOUNT FOR ONE THIRD OF THE PICTURE. THE TWO BODIES BELOW ACCOUNT FOR TWO THIRDS OF THE PICTURE. THE LEFT BOTTOM FIGURE MUST BE A \u201cMAGNIFIED FRONTAL COSTUME DISPLAY\u201d, WITH A SIGNIFICANTLY LARGER FIGURE IN THE PICTURE AND A FULL HEAD CUT OUT FOR CLOSE PROXIMITY. THE RIGHT BOTTOM FIGURE MUST BE A FULL BACK FULL BODY. IT IS FORBIDDEN TO USE THE LOWER LEFT-HAND MAP AS A SMALL GENERAL SCALE. DO NOT ALLOW ANY HEAD INFORMATION TO APPEAR IN THE LOWER LEFT CHART. THE RIGHT AND LEFT CLOTHING IS PROHIBITED FROM INCONSISTENT. THE NUMBER OF CHARACTERS IS PROHIBITED. IT IS FORBIDDEN FOR AI TO BE GREASY, PLASTIC SKIN, EXCESSIVE BEAUTY, CHEAP CG\u3002<\/p>\n

IN A MULTI-ROLE SCENARIO, IN ORDER TO FURTHER REDUCE THE PROBABILITY OF MISCALCULATION OF THE MODEL, A VERY LIGHT DISTINGUISHING MARK, SUCH AS C1, C2 NUMBERS (WHICH CAN BE MARKED WHEN THERE ARE MORE PEOPLE), CAN BE USED IN THE HINT. THE FUNCTION OF THIS NUMBER IS NOT TO REINFORCE THE IDENTITY ITSELF, BUT TO ENSURE THAT THE MODEL DOES NOT CORRESPOND TO THE WRONG OBJECT WHEN DEALING WITH MULTI-PERSON RELATIONSHIPS. IN OTHER WORDS, IT IS MORE LIKE AN \" INDEX LABEL\" THAN AN ADDITIONAL IDENTITY\u3002<\/p>\n

\"a360244jlnegd00bd000ullp\"<\/p>\n

At the same time, when identifying a person, it is sufficient to add a small amount of critical visual information, such as dress style, overall aromatic or emblematic features, without having to repeat a large number of facial details. Because, where four views already exist, excessive repetition of descriptions may interfere with the model ' s weighting of reference maps\u3002<\/p>\n

For example:<\/p>\n

FIGURE C1 (REF. 1), 22-YEAR-OLD FEMALE, WEARING A LIGHT-COLOURED TOP AND DARK-COLOURED LONG SKIRT, BLACK WITH LONG, MEDIUM HAIR AND A QUIET WHOLE\u3002<\/p>\n

FIGURE C2 (CORRESPONDING REFERENCE FIGURE 2), 24-YEAR-OLD MALE, DARK-COLOURED LEISURE JACKET, SHORT HAIR, AND OVERALL SEXUAL ORIENTATION\u3002<\/p>\n

\"6fc0a49bj00tlnego00g3d000v9000m6p\"<\/p>\n

Upon completion of this lightweight label, it is sufficient to give it directly to Seedance 2.5 for generation, and the model will be based on a reference map to achieve the main consistency of the subject, while the function of the hint is mainly to \u201cprevent the character from the wrong mark + from the basic visual information\u201d, rather than to redefine the person itself\u3002<\/p>\n

In such a work, the core of identity consistency has in fact shifted from a \u201cdescription of the person\u201d to a \u201cright reference figure\u201d\u3002<\/p>\n

Chapter Three: The scene is not just for beauty, but for real photography<\/strong><\/p>\n

When the identity of the person is stabilized, the second factor that determines the sense of truth is the environment\u3002<\/p>\n

IN PREPARING THE SCENES, MANY WILL GIVE PRIORITY TO VERY BEAUTIFUL INDOOR PHOTOGRAPHY, CONSTRUCTION PHOTOGRAPHY OR COMMERCIAL SPACE PHOTOGRAPHS. THESE MATERIALS ARE OF HIGH QUALITY IN THEMSELVES, AND THEY ARE ALSO VERY PROFESSIONAL IN THEIR DESIGN, LIGHT AND DEPTH, BUT THEY ARE NOT ALWAYS SUITABLE FOR ALL REAL-LIFE AI VIDEOS\u3002<\/p>\n

\"e9865cd9jlneh6022gd000v9012lp\"<\/p>\n

Photographic Works<\/p>\n

Professional photography is often carefully designed\u3002<\/p>\n

The windows are not exposed, the dark layer is rich, the furniture is organized, there is a beautiful deep separation between the subject and the background, and each visual element is in the most comfortable position\u3002<\/p>\n

\"416589c4jlnehk02mtd000v9015mp\"<\/p>\n

But mobile video in real life is usually not so perfect\u3002<\/p>\n

When ordinary people shoot with their hands in their bedrooms, the windows may be slightly exposed, parts of the unorganized bed may appear on the edge of the picture, and there may be charging wires, water glasses and books that have just been seen on the desktop, and the camera height will not always be in the most standard position\u3002<\/p>\n

These \u201cimperfect\u201d parts constitute a sense of life\u3002<\/p>\n

Therefore, if the goal is to produce more natural reality videos, the reference to scenes is best given to material with life record attributes\u3002<\/p>\n

What is really important here is not the need to film using a branded mobile phone, but the fact that the picture itself has no way of seeing ordinary people: natural light, normal hand-held heights, non-over-designed images, reasonable depths, and real signs of life\u3002<\/p>\n

In selecting the environment, several aspects can be highlighted\u3002<\/p>\n

First is light. It would be useful to know where the main light came from. If there is a window on the left side of the room, then the light orientation of the person should be broadly consistent with this relationship. Otherwise, the characters are easy to post like later\u3002<\/p>\n

\"5b412,305jlneif01nwd000v9000krp\"<\/p>\n

Next is the camera height. Living records are usually close to the person ' s chest, eyes or average handheld height, rather than the large-scale use of top-down slots or extreme overheads\u3002<\/p>\n

\"0f738faaj00tlneip01jdd000v90ksp\"<\/p>\n

It also depends on whether there are real traces of use in space. A truly inhabited bedroom is unlikely to be as free as a furniture model room. The wrinkles of bed sheets, open books, half-drinking water and charging mobile phones make space more credible\u3002<\/p>\n

\"a8b44dfdj00tlnej501f3d000v9000kpp\"<\/p>\n

Chapter IV: Solving with props and action chains: \u201cPersons don't know what to do\u201d<\/strong><\/p>\n

When the scene is ready, the next step is usually character behaviour\u3002<\/p>\n

There's a very common problem here: there's nothing around the person, so we can just repeat, \u201cThe person is sitting, naturally,\u201d \u201cThe person smiles\u201d, \u201cThe person looks out the window\u201d and \u201cThe person turns his head\u201d\u3002<\/p>\n

These moves are not inoperable, but if the whole video relies on this abstract behaviour, the characters will soon have a clear digital modeling sense\u3002<\/p>\n

Props can solve this problem well\u3002<\/p>\n

\"cfe42517jlnejg00eld000v9000opp\"<\/p>\n

Props can enrich the environment in the first place, but more importantly, they can provide the person with a clear addressee. The original text summed it up as \u201cdoing something for someone\u201d, an understanding that is very accurate\u3002<\/p>\n

If there's nothing in front of the person, we can only write:<\/p>\n

People naturally sit at the table\u3002<\/p>\n

\"9fa1b4ffj00tlnejw0140d000v90hfp\"<\/p>\n

But if there's a cup of coffee on the table, it can be designed:<\/p>\n

The person first looks down to the glass, holds the handle in his right hand, picks up the coffee cup slowly and then re-places it back on the table\u3002<\/p>\n

\"a62e0fa5jlnek5014ad000v9000hgp\"<\/p>\n

If the person has a book next to him, he or she can continue to turn pages, read and close books\u3002<\/p>\n

In the case of a mobile phone, a smaller act can be designed, such as a screen opening, a person looking down, hesitation and failure to pick up a cell phone\u3002<\/p>\n

So, the props are not simply making the picture look rich, but are addressing a more important issue:<\/p>\n

Why are people moving\u3002<\/p>\n

Devices suitable for reality video are usually common, such as cups, books, mobile phones, headphones, cameras, combs, keys, backpacks and umbrellas\u3002<\/p>\n

Their common denominator is that there is clarity in the way people interact and that, when a person carries out an action, the object ' s state will change\u3002<\/p>\n

On this basis, the individual actions will be further transformed into \" action chains \" \u3002<\/p>\n

For example:<\/p>\n

Girls drink coffee\u3002<\/p>\n

It's just a move\u3002<\/p>\n

But if it says:<\/p>\n

The girl had looked out of the window and the phone on the table was suddenly shaking, and she looked down and looked at the screen, but did not pick up the phone immediately, but held the coffee cup next to her, and then took a sip of it and put it back\u3002<\/p>\n

Video Cue Words:<\/p>\n

Using reference figure 1, which is the subject of the person, is strictly consistent with the person ' s five officials, hairstyle, clothing and body mass; using reference chart 2, which is the scene environment, maintains the original space layout of the water-bound coffee shop, window-by-window seats, wooden window frames, indoor warm environment and lake view outside the window; using reference figure 3, which is the coffee cup prop, using reference chart 4, which is the desktop mobile phone prop, and maintaining the appearance of all reference objects\u3002
\nA video of a real life record of about 8-10 seconds. The overall atmosphere was bright, sunshine, relaxing, like one afternoon when a friend sat across from her and recorded it with her mobile phone. The warm, natural sun comes in from outside the window and forms a soft and bright light on the person ' s face, hair and desktop, a natural reflection on the lake outside the window, a warm, woody indoor environment that integrates with the sun, and a permutation and temperature\u3002
\nThe person sits at the window, and his body relaxes. At first, she looked out of the window at the lake, with a slight smile in her mouth, and then, as she realized that a friend was filming herself, she turned her head towards the camera, had a short eye contact with the camera, showed a simple, natural smile, took a little bit of the sexy look that was taken by her friend, and then returned her eyes to the window\u3002
\nThe cell phone on the table was suddenly slightly shaken, and she noticed that she would then naturally look down to the cell phone screen, without reaching out immediately to hold the cell phone, but with a brief pause. She then looked at the camera again, as if she was responding with her eyes to the person who was filming her, with a little laughter, and then, naturally, reached out to hold the handle on the coffee cup\u3002
\nShe picks up the coffee cup and drinks a little, and then gently puts it back on the table. In the process of putting down the cup, she once again briefly looked at the camera, and her mouth was naturally up, like sharing this comfortable afternoon with her friends after the shot, and then she returned to the phone\u3002
\nThroughout the process, the person blinks, bends his or her head, adjusts his or her position and changes his or her vision, and there is always a sense of easy interaction between the camera and the person. Instead of continuing to stare at the camera or deliberately facing the camera, the person occasionally pays attention to the camera while looking out the window, looking at the phone, drinking coffee and giving natural eyes and smiles\u3002
\nThe camera uses a medium-to-half view, the seat is close to the perspective of a friend sitting opposite the person, with a slight handshake, and the camera is able to make natural adjustments and very slow approaches as the person moves. Don't be mechanically stable, don't be obvious commercial mirrors, don't keep people on camera. It's like a real, daily video of friends, sun, joy, lax, interactive\u3002<\/p>\n

People will be much more natural immediately\u3002<\/p>\n

The reason is that there is a causal link between the actions\u3002<\/p>\n

The cellular vibration drew attention, so the person bowed; she did not immediately take her cell phone, and there was a very short pause; the person then chose to drink coffee, and then a new behavioural target\u3002<\/p>\n

The logic of such acts is more important than simply increasing the number of actions\u3002<\/p>\n

And don't be afraid to stop\u3002<\/p>\n

In real life, a large number of conditions are very subtle, such as blinking, breathing, adjusting the seat, remaining on the handle with the finger, looking at the phone without moving immediately\u3002<\/p>\n

These brief pauses are performances in themselves\u3002<\/p>\n

the 10s video does not need to be visible every second. reducing the number of actions, increasing the links between them and stopping them tends to be more natural than continuing to increase complex behaviours\u3002<\/p>\n

Chapter V: Structure of the whole message and the time axis for real performance<\/strong><\/p>\n

In a stable way, you can break down the real video phrase into four parts\u3002<\/p>\n

The first part is a holistic way of filming, telling the model what kind of equipment and state it should be like, for example, mobile phone life records, natural light, light handheld, and not excessive commercial light\u3002<\/p>\n

\"929a5cfdj00tlnekv00izd000tv00q4p\"<\/p>\n

The second is the identity of the person, which clearly describes the number of the person, the characteristics of the core face and the elements that require continued consistency\u3002<\/p>\n

\"da7dd968jlnel300j3d000uc00r1p\"<\/p>\n

Part three is the prop<\/p>\n

\"bd4fbf9bj00tlnela00jbd000tv00rjp\"<\/p>\n

Part IV is the axis of time, with a real schedule for character behaviour, vision changes, props and lens movements\u3002<\/p>\n

\"81a11573jlnelk00h1d000td00r3p\"<\/p>\n

Chapter 6: Final re-processing of painting words and establishment of a fixed inspection method<\/strong><\/p>\n

The whole production process can be fixed into a simple workflow:<\/p>\n

First, prepare the material of the person, extract the anchor of the identity; then look for a frame of reference that fits the logic of life; then select two or three props that are truly capable of participating in the action; then design the person ' s behavior and action chain; and finally organize the process with a time axis\u3002<\/p>\n

In this way, \u201cto make a person more real\u201d is no longer a very vague goal, but has been reduced to a few issues that can be modified separately\u3002<\/p>\n

People change their faces, change their identities\u3002<\/p>\n

The environment is too fake to change the scene\u3002<\/p>\n

Action machine, change behavior\u3002<\/p>\n

It's a mess. Change the timeline\u3002<\/p>\n

It is easier to test and re-use such hint structures than to regenerate and take chances\u3002<\/p>\n

Conclusion<\/strong><\/p>\n

THE \u201cLIVING SENSE\u201d IN THE AI VIDEO IS NOT, IN ESSENCE, A PARTICULAR KEYWORD OR THE EFFECT OF WRITING \u201cREAL, NATURAL, FILM SENSE\u201d SEVERAL TIMES\u3002<\/p>\n

What really determines whether a person is credible is that the person remains stable from different angles, that the person ' s state can continue what happened before, that the environment is in line with a realistic photography logic, that props provide reasonable justification for the person ' s actions, and that there are natural causal consequences between actions\u3002<\/p>\n

When these relationships are clear, the hint is no longer just a description of a picture, but begins to take over the director's job\u3002<\/p>\n

SO WHEN YOU'RE GOING TO MAKE A REAL AI VIDEO, YOU CAN PUT DOWN THE WORDS \"THE TOP 8K\" \"SUPER-REAL\" AND ASK YOURSELF A FEW MORE IMPORTANT QUESTIONS:<\/p>\n

Who is this man? Where is she now? Why would she do that? After that, what happens next second<\/p>\n

When these questions are answered clearly, the person actually starts to move from oneAI Image<\/a>\"to become a person living in a picture\u3002<\/p>","protected":false},"excerpt":{"rendered":"

NOW THE AI VIDEO CAN ALREADY PRODUCE VERY CLEAR FACES, NATURAL SKIN TEXTURES AND RELATIVELY COMPLEX PERSONALITY MOVEMENTS, BUT MANY OF THE WORKS STILL HAVE A PROBLEM THAT IS HARD TO IGNORE: TAKING ONE FRAME ALONE, THE PERSON SEEMS TO HAVE NO OBVIOUS ERROR, BUT ONCE THE PICTURE IS REALLY MOVING, THERE IS A CLEAR \u201cAI SENSE\u201d. THIS PERCEPTION DOES NOT NECESSARILY STEM FROM INADEQUATE MODELLING CAPACITY, NOR DOES IT NECESSARILY BE RESOLVED BY ADDING WORDS SUCH AS \u201c8K, SUPER-HIGH RESOLUTION, FILM QUALITY, DELICATE SKIN\u201d. IN MANY CASES, WHAT REALLY AFFECTS A PERSON'S REAL IDENTITY IS WHETHER THE PERSON'S IDENTITY IS STABLE IN DIFFERENT LENSES, WHETHER THE ENVIRONMENT IN WHICH HE OR SHE LIVES IS A VISUAL EXPERIENCE OF REAL LIFE, AND WHETHER THE PERSON HAS ANYTHING IN HIS OR HER PICTURE THAT MAKES SENSE. IN OTHER WORDS, THE \"REAL\" IN THE AI VIDEO IS NOT JUST A MATTER OF DRAWING, BUT..<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[149,144],"tags":[802,3149,956,5321,192,6481],"collection":[],"class_list":{"0":"post-58342","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"hentry","6":"category-jiaocheng","7":"category-baike","8":"tag-ai","12":"tag-prompt","13":"tag-6481"},"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/58342","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=58342"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/58342\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=58342"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=58342"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=58342"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=58342"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}