After the recent update of GPT4o's multimodal graphs, the whole circle has entered a period of AI exuberance, and everyone is crazy about the actual operation of the graphs, and the feeling of "opening up a new world" has come back in the previous two years when I first came into contact with AI.
As a blogger who focuses on practical AI cases, of course, not just to do a few charts to highlight how much GPT4o raw charts are fluff and done, but more hope that it can be used to land on practical AI projects to help people realize.
I recently noticed this magnetic male voice over philosophical short story, produced through AI for graphics, AI video, and film and TV mixing, which easily gets 100,000 likes and is perfect for utilizing the latest GPT4o to speed up the graphics.
Account Status
The blogger distributed the content to the jitterbug, video number, the account are up, jitterbug fast 30w fans, and recently distributed the content to the oil tube SHOTS, for the opening of the YPP, a fish more than one to eat!
How do I cash in?
1️⃣ Both platforms have trailers with goods, as the users are male-dominated, books, men's products, and in-vehicle products are more appropriate
2️⃣ Open a member's area to produce and customize higher quality videos and courses to realize knowledge payment cash.
3️⃣ platform traffic revenue, the account has huge traffic, in the domestic dual platform already has a certain base traffic revenue, if in the oil pipe SHOTS account into the traffic pool, YPP revenue is more fragrant!
The next step is to take you by the hand and make this AI philosophy with GPT4oShort Videos, after reading it will surely have will surely have gained! If you are unable to use GPT4o, you can also try to use DeepSeek ➕ i.e. Dream Finish operation.
Production Process
1️⃣ Generating Stories
2️⃣ Generate copy in English and Chinese
3️⃣ Generate corresponding image
4️⃣ Voiceovers
5️⃣ Clips
I. Generating stories with GPT
- Tool: CHATGPT
- Website: https://www.chatgpt.com (magic required)
1️⃣ Copy Extraction
By extracting the story copy of multiple videos that you want to benchmark, organize them into a word document and use them as a GPT information feed, if you still can't extract the video copy of the students can read this issue: "Farewell to charges! Four free AI video watermark removal artifacts + all platform short videos without watermark extraction (recommended collection)".
2️⃣ Getting Gpt to Learn Storytelling Style
Send the word document to the GPT and it learns the structure, style, emotional pacing, etc. of the story.
Gpt will analyze the characteristics of each story and come up with a copywriting publicity proposal for the story:
3️⃣ Let Gpt create stories!
Next, we let Gpt follow the examples it has learned to create similar stories for us to choose from, I chose the "story two" is the story in the hands-on work.
II. Gpt generates Chinese and English texts
After selecting the story, we had Gpt create both the English and Chinese copy for the story and that was it.
III. Gpt creation of corresponding pictures
The point is, if we follow the Deep+Image story picture production method, to this step is generally to let DeepSeek create a corresponding story of multiple pictures prompt words, and then we go to Image Dreams to generate, and in Image Dreams need to be used many times with the technique of matting in order to maintain the consistency of the characters, the scene.
This operation would require dual-platform coordination, repeated matting, and wouldn't work too well, and would be easily confused involving multiple character AIs:
But now, we just need a little guidance inside the Gpt dialog that we just finished creating the story, and it can directly generate the corresponding picture in the story just now, and the picture effect texture is very great! Let's see how it works.
1️⃣ Ask the gpt to unpack the story into multiple scenes and tell the prompt words for the scenes.
2️⃣ Once we have the image descriptions for all the scenes, we can have Gpt start generating images
In response to the man's first scene where he is lost in the balcony, we can compare the effects of Gpt and i.e. Dream:
Gpt comes out with a more cinematic quality to the images, character emotions that fit the story better, and a more advanced color scheme that's great!
3️⃣ We continue to "use the mouth of the picture", asked Gpt to keep the characters, style, location and other elements, continue to generate all the scenes of the picture can be, encountered do not like it can let it generate again, it is a very high rate of film!
This will give us all the pictures we need for this story, so let's take a look at them:
Character likenesses, balcony scenes, chairs, and other elements have really held up well! If you've tried to maintain consistency through MJ, i.e. Dream, and other mapping software you should understand why I'm so excited!
Because the story is created by Gpt, it is more likely to know what kind of image it should use to present its story, and for us creators it also eliminates the tedious operation of repeatedly padding images across multiple platforms!
IV. Dubbing with NoizAI
- Tool: NoizAI
- Website: https://noiz.ai/landing
Before we start, why don't I just use the "clone tones" or the tones that come with the projector? It's because the voice reading is more optimized for Chinese, if you use English, it will have a heavier accent, you can feel it at .......
Ok, let's continue with the tutorial, open Noiz and click Start Now ➡️ New Clone Sound
Upload your favorite sound, it must be 5 to 10 seconds.
This will have the sound of our narration, so you can try it out and re-upload it if you're not satisfied.
Then we take the English text we just wrote and separate it into two to three paragraphs or so and just let it generate the voiceover!
Just click on the bottom right corner to save to local when you are satisfied.
V. Editing Video with Cutting and Mapping
1️⃣ We open the cutout and select it's "Picture as Text".
2️⃣ Click on "Edit Copy Freely"
3️⃣ Copy our Chinese text in, yes, Chinese, it will not be recognized by the cutscene if you copy English in. After copying, choose a tone (the voiceover will be muted later) and click Generate Video!
4️⃣ Select "Use local material".
5️⃣ Cutting the movie will help us to set out the template at the pace of the Chinese dub, which contains the dub track, subtitles, and BGM.
The audio track we delete, the BGM can be listened to to see if you're happy with it, and if you're not, you can delete it.
6️⃣ Import all images and voiceovers into the cutscene in order.
7️⃣ Pull the pictures and voiceovers in order to the corresponding subtitle positions, and copy the English subtitles under the Chinese subtitles.
8️⃣ Appropriately add some picture effects, transitions effects, you can look at their own feelings to choose, do not necessarily have to choose my
9️⃣ Add Movie and TV Mix Clips
- Tool: 33 Search Frames
- Website: https://fse.agilestudio.cn
Through a simple search, you can get the corresponding picture of the movie and TV clips, we download them and then add them to the end of the cut screen can be!
🔟 Export video
When you're done, we'll click Export and we're done!
Project Advance
1️⃣ As I mentioned before, if you want to go deeper into your project, try converting the images to video for a higher overall viewing experience!
2️⃣ Subtitle, watermark control, my work is only as a process display, so the subtitle can be adjusted better, such as the proportion of Chinese and English, the focus of the word highlight, etc., we can go to refine.
3️⃣ Transitions, special effects effects, you can feel more about the plot, style of the story, each story made differentiation adjustments.
4️⃣ BGM optimization, if you have the conditions, you can adjust the BGM stomping, volume, so that the BGM with the advancement of the plot and change.
The disassembly of the whole project is here, immediately on the practice up!