Used to make a productpromotional videoWhat process is needed
Creative planning, script writing, mirror drawing, film preparation, late production, finished output
Now, there it isSoraThis process became..
AI WRITE SCRIPT - INPUT SCRIPT → ADJUSTING PHRASES → GENERATE
It's not science fiction, it's reality
So how does it work? Come down and let's try it。
Generate drafts
Tools: Generate drafts of dream 4.0
I'll start with a dreamLine Chart
Enter the prompt
Zenium
with iced tea as the main subject, a set of advertisements was generated. it contains nine graphs, linked to the spectroscopy, logical, with different schematics, ending with a product close-up containing logo. the specs are in black and white。
Nine graphs are generated here, and if only four are generated at times, try more than once, or change to the Agent model。
Then we're going to synthesize these nine drawings, and we're going to use PS here, or other puzzle software, or PPT, Word。
Generate File
Before that we could use a little verb material
it is possible to create a synonym for our mirrors by using the general question (https://chat.qwen.ai/)
Uploads the image of the lens that has just been assembled, and then enters the hint "Based on this photo specs, provide me with the appropriate Chinese lingo”
Generate Video
It's time to get the bibliographies
we can use sora for video generation and start sora
Opensora (https://sora.chatgpt.com/)
Upload a combined lens picture below, and then enter a hint
based on the drafts of the spectroscopes, a tvc commercial for pepsi coke products is produced
“From the mountains, pick the first fragrance of tea。
With heart, for pureness。
Thermally agitated, awakening the odour。
Add ice crystals, and it cools in the air。
After the exercise, it's the most timely reward。
Be alone and be quiet。
When we get together, we share happy times。
There's no burden in getting out of work。
Master Kang, iced red tea, had a good summer, and walked."
Video size set to screen
and then we'll generate a video, and we'll wait for the video, and that's it。
We found several problems:
1. The first map was lost
2. The first frame became a synthetic draft
3. Only half of the cases are documented。
the reason for this is that the 10s video is a little short and the case is a little long, resulting in the video being closed and half the case。
Come down and adjust the hints and simplify the text。
based on the contents of the nine spectroscopy drafts, a tvc commercial for ice-cream tea products is produced, written in real style, in chinese:
"From the depths of the mountain
Be careful。
It's delicious when it's hot。
When you drink cold, it cools your heart。
It's a reward after the exercise。
Stay alone。
Happy to be together。
Inter-work help。
Master Kang, iced red tea, had a good summer, and walked."
The overall effect is still possible, but the picture is still fast and the case is slow。
if we want to do better, we can do it twice, with 4-5 images in 10s, and finally merge。
We split it into two parts, one for video。
The whole effect is not as good as it is, and then we'll adjust it with the big cut。
The whole is much better than the first two, and one small problem is that video is generated in two segments, and the tone and background music are not consistent。
All right, that's it for today. Interested students can try it on their own。