[Journey] Growing AI Influencer farm

[hexagonal]

Growing AI Influencer farm

So i've finally find how to distribute videos on tiktok and instagram without getting shadowbanned or accs flagged.

I have tried so much stuff during the past years and finally made a really simple flow
Following flow
1) Finds viral videos on social media
2) Generate similar videos for all the accounts that have a similar character
3) Upscales to 4k
4) Publish and repeat

So far I've good results, here's an example of one of those videos

 

[OvO]

Nice. Which AI image/video generation tools are you using to create these?

 

[hexagonal]

Nice. Which AI image/video generation tools are you using to create these?

Hi! I run an img2img model to generate the model holding the product or in the scene that I wish, then run wan 2.2 to generate the final video

 

[OvO]

Are you using something like Z Image or Flux for the base image to start? Also are you using character/face consistency?

 

[Sofia445]

Wish you the very best! I will be following along. I myself would like to try this someday but all this SDXL, Wan etc etc just confuses me lol.

 

[hexagonal]

Update
Working in a dashboard to keep track of the influencers for each of the campaigns, having 3 agents running auto setting up everything.

Future is sick, placeholder values so far in the dashboard​

 

[WarlordYT]

how are you going to manage API cost of wan 2.2 , its like 0.25$/second

 

[xHaYwiRe]

wish you the best, keep update

 

[TechEinstien]

Looks amazing. How are you monetizing it?

 

[hexagonal]

Improved the model, prompts and videos now look even better without human work

https://streamable.com/bbzkhv
https://streamable.com/bbzkhv

how are you going to manage API cost of wan 2.2 , its like 0.25$/second

Hey I self host it thanks for following all my journeys hehe

Looks amazing. How are you monetizing it?

Promoting own stuff

 

[kaniko]

what is 2.2 ?

 

[desperado596]

wan 2.2

its ai model here : https://github.com/Wan-Video/Wan2.2

 

[jelenaitmui9]

Using a combination of image and video generation models can help create more realistic content, running the generated videos through a upscaling process also helps to avoid detection. Distributing the videos across multiple accounts with similar characters seems to be the key to avoiding shadowbans.