{"id":43375,"date":"2025-10-05T17:28:52","date_gmt":"2025-10-05T09:28:52","guid":{"rendered":"https:\/\/www.1ai.net\/?p=43375"},"modified":"2025-09-30T12:58:26","modified_gmt":"2025-09-30T04:58:26","slug":"ai%e8%a7%86%e9%a2%91%e5%88%b6%e4%bd%9c%e5%85%a8%e6%b5%81%e7%a8%8b%e6%95%99%e5%ad%a6%ef%bc%8c%e7%94%a8%e8%b1%86%e5%8c%85%e5%92%8c%e5%8d%b3%e6%a2%a6ai%e5%88%b6%e4%bd%9c%e8%a7%92%e8%89%b2%e4%b8%80","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/43375.html","title":{"rendered":"IA VIDEO PRODUCTION OF FULL-PROCESS TEACHING, AND IA SHORTS WITH SOYBAGS AND DREAM AI FOR ROLE CONSISTENCY"},"content":{"rendered":"

Bean curd<\/a>,Dream<\/a> after the update, seedream4.0 has made a big breakthrough in graphic creativity and is very useful for the production of AI videos, especially for role consistency\u3002<\/p>\n

Today I'm going to share my dream and bean buns and make one as quickly as possible AI short film<\/a>And the whole process didn't cost a penny\u3002<\/p>\n

It's a four-step specs, a graph, a video, a clip. It's easy. It'll work\u3002<\/p>\n

i. Mirror production<\/strong><\/p>\n

Let's start with a set of shots, and you can think about the structure of the story\u3002<\/p>\n

You can also use an AI supplement, like not knowing what to do with the subject, so that Ai can help us recommend something\u3002<\/p>\n

I'd like to make a short film in about 30 seconds, and I don't know what to do, give me 20 themes, and give me a reason<\/p>\n

\"953665e73j00t2q7lr0059d000m800vzm\"<\/p>\n

PICKING OUT SOMETHING I LIKE, LIKE, I PICKED OUT \"SUNDAY OF THE STRAY CAT,\" AND THEN I ASKED AI TO HELP US WRITE THE SPECTROSCOPY:<\/p>\n

You're a professional video director, who's going to do a 30-second video about an orange cat walking through an alley in the afternoon, jumping up a window and drinking milk, being gently touched by an old man in the sun and telling stories<\/p>\n

The content of each shot is not complicated, but it's fine to list it in terms of scene + scene + main + action\u3002<\/p>\n

\"e27403e9j00t2q7lq002nd000mb00dvm\"<\/p>\n

THE SPECTROSCOPY CREATED BY AI IS MAINLY A SOURCE OF INSPIRATION, SO YOU CAN FILTER, DELETE, AND SORT OUT THE DESIRED FRACTION\u3002<\/p>\n

I'm mostly here to do a demonstration, and I haven't made any changes\u3002<\/p>\n

BESIDES, MY SHORT FILM IS BYSTANDER, SO HERE'S TO GET AI TO HELP ME PRODUCE A BYSTANDER BY THE SPECTROSCOPY\u3002<\/p>\n

Based on these spectroscopes, write me a story, keep it consistent, talk it out, and time is in about 30 seconds<\/p>\n

\"c6923b62j00t2q7lq000wd000kp006mm\"<\/p>\n

The content is also likely to be sufficient, because the video is still highly random, and the subsequent citation needs to be rewritten on the basis of the video\u3002<\/p>\n

IF YOUR SHORT FILM IS DIALOGUE-LIKE, YOU CAN ALSO ALLOW AI TO REFER TO SCRIPTS TO GENERATE DIALOGUEZERO-BASIC AI TEACHING, FREE-OF-CHARGE, ANIMATED AI PODCAST VIDEO WITH BEAN BUNS<\/a>!<\/p>\n

Some video generation and editing techniques are also mentioned and have some reference\u3002<\/p>\n

II. VINCENT<\/strong><\/p>\n

Now, we're going to send this seven-point lens directly to the bean bag, so it can generate pictures\u3002<\/p>\n

There'll be an accident<\/p>\n

According to the spectroscopy below, a continuous set of images is produced in the form of animated films by Pixar, with consistent and consistent figures, animals, props and props, while maintaining the same scenery\u3002<\/p>\n

\u2026<\/p>\n

although the bean bag has been updated, seedream 4.0, it's got pictures, role consistency as before 2.0, bad<\/p>\n

\"4e373638j00t2q7lr00avd000dv00jhm\"<\/p>\n

I thought I had a problem with the way I asked, so it didn't work well to change the hints, or to upload the pictures for the bean bag\u3002<\/p>\n

If a friend knows why, tell me about the comment section<\/p>\n

\"84ebe0d0j00t2q7lq004bd000lc00b6m\"<\/p>\n

it's good to seedream4.0 the same way<\/p>\n

The same hint, the consistency of the role that produces the picture is completely fine and its effects are stable\u3002<\/p>\n

\"0772c95aj00t2q7l00gkd000tc00bm\"<\/p>\n

\"058d335ej00t2q7lr00fzd000tl00bdm\"<\/p>\n

You can brush it a few times, you can't be satisfied with a picture, you can recreate it<\/p>\n

If you have a problem with details, you can also click on a picture to change the big chart mode\u3002<\/p>\n

\"3f6879adj00t2q7ls00hod000uedm\"<\/p>\n

OR REUPLOAD THE IMAGE AFTER DOWNLOADING IT AS A REFERENCE TO RECREATE AI\u3002<\/p>\n

Finally, the resulting images are sequenced to see if the lens is missing and if there are any better ideas to update the images in time\u3002<\/p>\n

III. Toussaint video<\/strong><\/p>\n

Now a bean bag with a account number, a computer can produce 10 videos free of charge in one day, a mobile phone, five<\/p>\n

THE DREAM NON-MEMBER IS 720P RESOLUTION, 5 SECONDS MINUS 10 POINTS, WITH A HIGHER DEMAND FOR VIDEO QUALITY AND MEETING MEMBERS\u3002<\/p>\n

\"a7cb6a33j00t2q7lq0015d000nq008hm\"<\/p>\n

It's basically enough for our daily output, and if you have multiple cell phone numbers, you can add more than one\u3002<\/p>\n

Just upload the pictures and simply describe the next camera and the main movement\u3002<\/p>\n

for example:<\/p>\n