Recently, I saw a lot of accounts on the jittery voice in the hair of this fairy group of video, looks fairy, but a closer look can be found, basically AI generated, but this does not affect the data of the video at all ah, basically every one of them has thousands of likes.
See here not only to lament, love of beauty (SP) is really a major power source for the development of AI technology ah.
Here's how to make a similar video together
AI mapping
A set of reference cue words are given first, but of course you can adapt them to your own needs
Main Description:
Five curvaceous Oriental beauties, their skin glistening with mother-of-pearl, their collarbones draped with fine gold tassel chains. The modified jacket is made of liquid silk, the upper part of the body is covered only by gilt peony embroidered breastplate, the waist is wrapped with gradient transparent shark silk sash, and the high split skirt reveals the garter stocking hoop full of crushed jade with the movement. Her feet were on gold silk shoes with pointed tips, and her ankles were wrapped around red rope bells with vermillion sand. Long hair with meteorite iron hairpin half pulled, a few strands of green silk sticking to the sweaty neck, the end of the eyes depicted malachite powder stained tear makeup, lip color like drops of pomegranate juice like full of drops.
Scene Description:
The Mirror Flower Water Terrace in the night, eight bronze appraisals surround the central hot spring pool, the water surface floating gold leaf and peony petals. Background mist suspended in the faded Dunhuang flying satin, broken carved corridor hanging down the wisteria holographic projection, white jade railing perched on the gilt chi dragon beasts, mouth spit out curls of incense in the moonlight with a light purple halo.
Light Shadow:
Amber moonlight intertwines with spa vapors to create a honey-glazed, luminous transition on the skin. Rembrandtian triangulation of light is used in key areas, with iridescent polarization hidden between the folds of silk, and a pale pink halo of subcutaneous capillaries appearing in the tulle passages. Reflected light from the surface of the water traces a flowing golden line down the side of the leg, interspersed with a veil of Tyndall effect light in the mist.
Atmosphere:
The forbidden beauty of the floating fragrance perfectly harmonizes the classical sense of abstinence with the modern erotic tension. Like a slightly drunken fairy after a banquet in the Yao Chi, both divine and blasphemous, but also with the temptation of flesh and blood of human warmth, every fold and undulation of the clothes are in accordance with the golden ratio of the human body.
style:
The new school of oriental erotic aesthetics, using 8K ultra-high definition skin rendering technology, sweat beads to see the rainbow ring formed by surface tension, tulle needs to show the silk protein fiber structure. The scene is combined with digital Chinese ink and wash, the edges of the petals are pixelated and dissipated, and the remnants of the ancient buildings retain their weathered texture. The overall effect is a fashion blockbuster like Vogue's Ancient Style feature, emphasizing ergonomic aesthetics and the ultimate expression of material translucency.
Of course one can also ask the AI to help with the rewriting, like this
Below we feed the written prompts to the graphing software, in this case the choice is Instant Dream AI
Fine detail is pulled full and the image ratio is selected as 9:16
The result is okay, we choose the last one, go to the canvas to expand the picture
We get a picture that looks like this
High Definition
Next we open up Starstream AI for HD image outing
https://www.xingliu.art/
Click on Create Now and copy in the image you just generated from Dreams!
Select HD zoom, fine at 0.3 is fine
Once the HD zoom is complete, we can download the image
Generate Video
Let's start by selecting the instant dream to see the effect
The generation speed of the instant dream is still good, the effect is a bit too neat and tidy
Try Koring again, now that the Koring experience has also been improved, and the speed of generation for non-members is very impressive!
This is good, there are some special moves that seem more realistic
Let's try again with the Conch AI, which is very rich in dribbling, except that it is relatively slow to generate.
Finally try Tongyi Wanxiang, this generation time is too touching, never thought of ah
These two tools, rolled a few times are not particularly satisfied, in general is the video of the character movement amplitude is too large, resulting in the character out of shape, some even originally 5 people, the back becomes 3 people, AI in the number of characters in the control still needs to be improved.
Video Editing
The last is to generate a good more satisfactory material into the editing software editing can be, with dynamic background music, a more attractive girl group video is completed!
Final Thoughts
In fact, now the use of AI to produce video has been very convenient, to video fire, the main thing is still need to have a person's IDEA, a good idea, can create a legend.
Then it's about not getting too hung up on whether some content will catch fire and how you'll cash in when it does; thinking too long term is sometimes not always a good thing, and often gets in the way of your own actions.
Do it first, and when you become skillful at it, everything will go your way!
