BROTHERS, IT'S THE END OF THE YEAR, AND THE AI CIRCLE BEGAN TO FIGHT
It's too much..
First there's Google Nano Banana Pro, and the drawings are so strong, then we're going to expand the software. In particular, O1, just released by Clin, says, “One sentence can change the video”. There's still a lot of big factories that want to move, and a big wave of products on the road。
The fairies fight, and mortals enjoy themselves。
Question: How do you choose to fight with too many tools
IT'S THE KIDS WHO MAKE CHOICES. ADULTS HAVE TO RECOMMEND YOU A TOP-OF-THE-ART AI "FAMILY BUCKET":
i don't know
Nano-Bnanan pro, Midjourney, Flux; VideoVeo3.1, Sora2, Clin O1, dream, conch。
IT'S CALLED THE TOP OF THE WORLDLovart Everything
It's like, you want sugar, you want red sugar, you don't have to make choices。
enable loveart
open lovart.ai, register in. there's free points。
The interface is super clean, just an infinity canvas, a "+" on the left, an image generator or something。
I went in and started with a left-hand "+" new project
Here's to share with you a couple of Lovetts that make me "whoosh" moment:
1. First shock: How can the video be changed
I USED TO KNOW THAT AI CAN CHANGE PICTURES, MAKE VIDEOS. WHO'S HEARD OF VIDEO? YOU CAN DO IT WITH THE A.I., YOU CAN DO IT YOURSELF, YOU CAN DO IT。
That's what this is about. It's called high-tech for top hackers. It's only been a few years, and now the average man can do it
I uploaded a video of a very good-looking woman and a very good-looking woman's picture。
Upload or select a video (left), a picture (right)
Enter in the prompt box:
- "Replace the woman's clothes in the video with the woman's clothes in the picture, and the rest without change."
After a few minutes, it's changed。
It's really like, "A word for video."
It's not the kind of gibberish, it's the kind of thing that matches the light, the quality。
I was just a little confused, and there's only one thing in my head: Now that technology has evolved to this extent
2. Finding a new continent: this is the right position to make a video
After these two days of double-testing, I found out that in Loveart, using the Nano-Bnanan pro, then using the Veo, Sora, Cresian video, was the most powerful
The video process is shared below. First do a spectroscopy in Loveart and then call on some of the top-of-the-a-A's videos to make a little musket。
Use Nano-Bnanan pro to draw first
In lovart, dot left navigator+ number, select Image Generator, choose Nano-Bnanan pro, enter hints, set resolution (available at 4K!)
I lost a long list of hints to the East Asian girl who wanted the film to feel and look funny. As follows:
- The tip: "A woman of about 25 years of age sitting on the side of the open lawn, in the dark and in the natural light of the day, was caught on the side of her face. She had black pony tails, air, a few tatters on her cheeks and a natural light makeup (sweet cheeks, soft eyes, pink lips). She wears a fine silver necklace, with a clear neck and a chain line, and she has a tight body. The upper body wears a navy blue-banded garb, and the heart/micro-V collar (deep v), metal shoulder bands are visible; the lower part is a high-grain army green thong. She had white towels in her right hand, a small frontal body, eyes on the right, a soft, shy face, a light red colour and a soft bite on her lower lip. A friend on the right side was partially circumcised and debilitated. The colour is soft, the contrast is low, the focal lens, the shallow view (bokeh), the fine particles and the mist feel, the film feels, but remains real。
Pop, the picture's out. Nice
THEN! THE POINT IS! I'M GOING TO GET AI TO CREATE A NINE-GAUGE SPECTROSCOPE FOR THIS BEAUTIFUL WOMAN。
i don't even want to think about words, pick up a beautiful picture, upload a reference to loveart, and say:
- Reference diagram 1, 9-magic position, generating a 9-magic photo of the 2-magic, consistent with the 2-magic appearance."
No problem, the spectroscopy is out。
And you can use a long hint without a reference. I tried. It was too long. I put it to the end。
Let's see the effect:
With a spectroscope, you can take it and make a little movie。
- Tip: Convert the storyboard grid in the picture in order (left to right, top to bottom) to a continuous video. Each grid is a shot, smooth transition. Consistency, no text shown. Film quality, pure picture。
I got a musket shot:
In the front, she took the lead in creating a spectroscopy:
With this spectroscope, I'm calling Veo3, Sora2 and Colin O1 on the same page in Loveart, making video. The hint is simple:
- Phrasing: Generates a video around the spectroscopy of Figure 1 that requires each shot。
The video is as follows:
At that moment, I really thought it was too convenient... and the best part was here -- you don't have to struggle with what's best, just test, one comes out the right one。
if there are new models in the future, loveart will soon be integrated into the platform. like yesterday's optimum-sync 2.6, and i think it'll be coming up today or tomorrow。
It was you who moved the tools, and now they're around you, and it feels so good。
Real climax: Touch Edit, “Manufactured map”, point
In addition to the integration of dozens of tops in AI, Loveart, there are a lot of living things。
If it's just "facilitation," then it turns out that when Touch Edit does this, I'm really gonna shoot my thighs
WHAT'S THE MOST PAINFUL THING WE'VE EVER DONE WITH AI? IT'S NOT EASY TO FIX THE DRAWINGS。
It's a problem when the models are stronger。Nano Banana Pro it's already a very powerful model, but there's a problem: good use, good control, you keep changing prompt, regenerating, smoking cards, expensive。
1. Editorial Elements
loveart has an "editing element" function. loveart can identify each of the elements of the original map, including background, outlook, product, text, all of which are separate layers! all elements can be selected, dragged, replaced, and even text can change content and font styles directly
2 Touch edit
It's the new Touch edit, which makes me completely satisfied
Use a simple button to hold the ctrl/ cmd (Mac), click on any picture. The selected element will automatically jump to the dialogue window and you will send a command to the dialogue window。
Precision control
WE'RE USING AI, AND WE'RE HAVING A LOT OF TROUBLE WITH "NEED TO THINK ABOUT A PART OF THE PICTURE WITH OUR HEAD."。
For example, in order for me to get a "bite" in a poster, I had to smoke a dozen cards, and people had to crack。
Now, I'm just going to click on the eggnut and say, "Here's a bite," and it's over。
This function allows for precise and partial modifications, modifications, replacements, and controlled extensions: objects, scenes, environment and light。
Integration living map
simply put, you can upload a lot of drawings on the lova platform, drag them all in, then click on the parts you want, and then they can combine them into pictures。
IT'S VERY POWERFUL, IT'S PIXEL-LEVEL, IT'S A GIRL YOU CLICK, IT'S A GIRL'S FACE OR A GIRL'S FACE, IT'S A NOTEBOOK, IT'S AN APPLE COMPUTER, IT'S AN APPLE COMPUTER, AND IT'S A MANUAL SELECTION。
Looks like I've uploaded a bunch of pictures
The recommendations are as follows:
Get a nice mouth girl:
When I saw this picture on it, I laughed — it's not a living picture anymore, it's a puzzle. Later on, you can see it, fix it, change the light, change the background, change the face, change the word, click on it, and then you can see it, and then you can do it through a chat。
There are, of course, other functions such as changing posters and making PPTs. Direct mark text can be modified. You can try it yourself. It's very convenient。
One generalizes, and it feels like Lovart finally solved the century's dilemma of how the reference diagram works。
NOW AI IS YOUR EMPLOYEE. YOU HAVE A TOP DESIGNER WORKING FOR YOU。
Agent
WE USE THE ORIGINAL AI TO MAP, AND THERE'S OFTEN A SENSE OF WEAKNESS, AND YOU HAVE TO USE ALL KINDS OF HINTS TO FIGHT WITH AI。
and loveart, with its own smart body, with design, with context memories, plans and steps every time it gets a command。
Look at the way it's thinking. It's the way it's thinking every time, so it's very good. If the drawing is wrong, it automatically adjusts the hints and policies and recreates them。
There was also support for several consecutive rounds of change, which would not be inconsistent。
To summarize.
To be honest, the depth of the last few days is that I feel like the bottom of Loveart's intelligence, Touch Edit, and so on, turning a strong model into a strong work stream, i.e., see what you get, with a little mouse point, which is the way to work in the most intuitive and user-centered way。
Previously, the precision of 80% had been modified by a mere hint, using the mouth to direct Nano Banana; now, the lovart smart body plus the Touch edit function has increased to 90%。
DON'T UNDERESTIMATE THE EFFICIENCY OF 101 TP3T, WHICH REALLY SAVED PROFESSIONAL USERS' LIVES。
Anyway, with Loveart's Nano Banana Pro, all kinds of videos, AI, it's like it's like a tiger's wing
Here's your address:https://www.1ai.net/35218.htmlI don't know. Seriously, try that Touch Edit. You'll come back to thank me
HYDRACH PUNISH 1
I'm sorryrole>You're a movie specter. To expand a reference to a series of stories of film。< /role>I'm sorryinput>Users provide a reference map。< /input>I'm sorryi'm sorry>1. Analysis of all subjects (persons/objects/environment) and their relationships2. Strict consistency: same role, clothing, environment, light running through all images3. Do not speculate about real identity or location, create based on visible content4. Do not introduce new elements that do not exist in the chart5. The depth of the view changes with the lens: the vision is deep, the profile is shallow< /i'm sorry>I'm sorryplease>generate a 3x3 storyboard grid map (4x3 if more lenses are needed)Narrative order (left to right, top to bottom):- Box 1: Vision for Environment- Box 2: Panorama presents the whole of the subject- Box 3: Middle View Intake- Grid 4: Central periscope focus- Box 5: Periphery/interface- Grid 6: Exterior view details- Cell 7: Dynamic angle (on or above)- Cell 8: Emotion/reaction lens- Box 9: End shotRequire:- ** No words, labels, numbers, labels appear in the image**- Grid is separated by a thin black line only- Every cell is a pure film- There's a clear emotional progression- Unified movie style< /please>I'm sorryoutput>Only one clean storyboard grid, no text。< /output>
HYDRAX PINT 2
3×3 Movie Spectroscope Contact Sheet Pitt Plugin (Chinese engineering version)Mission objectiveImage analysis based on input images, locking the same subject in the same environment and generating one 3×3 Contact Sheet Grid 9 A picture of film footage that covers the full range of focus and space angles。Global ContraintsIt has to be the same personMust be the same costumeMust be the same lightIt has to be the same environmentreal photographycinematic color i don't knowAt the bottom of each image, a modern sense, film-style Chinese word label must be addedFont requirements: very simple, professional, modern film sense fontNo ventilationBAN GAME WIND UIBan Cheap FontsFirst row: Environmental constructionELS LABEL: SUPERVISEDThe subject is in an extremely broad environmentVery small numbers of people, highlighting the size of space and the overall atmosphereVISION LS LABEL: FULL BODYHe's got a full head to footStandard station or natural dynamicsMEDIUM VISION MLS LABEL: KNEE CHARTThe characters enter from above their kneesA balance between environmental narrative and personal performanceSecond row: Core CoverMIDVIEW MS | LABEL: ABOVE WAISTFrom the waist aboveEmphasis on character postures and interactionsCENTRAL NEARVIEW MMU LABEL: ABOVE CHESTFrom the chest aboveFocus on people's emotions, state and expressionNEARVIEW CU| LABEL: FACE CLOSE-UPPure facial close-upHighlight eye, emotional tension and psychological stateThird row: Details and anglesDestails & AngelsSUPERVISE EUU LABEL: DETAILSDepth focus body parts or accessoriesAllow: eyes, ears, earrings, fingers, necklacesBan: Pure fabric texture, non-material surfaceLow Angle label: Low angleFrom the ground upA strong sense of visionThe characters have a high-profile, oppressive, heroic visionIt's very movingHigh angle High Angle label: High angleLook down at the subject from top to bottomHighlighting space relationsCreating a vulnerable perspective or a sense of environmental oppressionFinal output image request3×3 Full mirror gridEach cell must satisfy:Correct lens languageCorresponding Chinese camera tagsReal lightFilm-level unified coloringPeople and the environment are the sameModern professional Chinese font layout