AI Influencer Video Content Creation Question

[Tinzo]

Hi all,

I am currently on my journey in creating AI influencers and models. I've used Stable Diffusion to successfully create the avatars and get a consistent character.

My question is, what are the best programs (locally or non-local) to create short reels, SFW and NSFW?

Here is an example of an account I wish to emulate: https://www.instagram.com/urasianivy__/reels/

Any help is appreciated in advance!

 

[shijin]

For NSFW projects, look into ComfyUI workflows with ControlNet for consistency

 

[lillygreen]

Wow it's really impressive!

 

[Tommy_account]

Looks like you’re moving in the right direction - starting with Stable Diffusion for the base influencer/avatar is a solid foundation. The challenge is mostly about consistency + movement.
ComfyUI and ControlNet will help you to contol camera angles.
You can also try to train LoRA on your influencer's face/body to keep the character consistent.
D-ID or HeyGen is for making tailored talking-head videos (here you can work with lip sync and voice sync)

The best thing is to train LoRA or DreamBooth model. I can even say it's a must if you want to keep your avatar recognizable frame to frame.

Could you pls clasify what is more interesting for you - NSFW or SFW models?

 

[V]

For NSFW projects, look into ComfyUI workflows with ControlNet for consistency

Looks like you’re moving in the right direction - starting with Stable Diffusion for the base influencer/avatar is a solid foundation. The challenge is mostly about consistency + movement.
ComfyUI and ControlNet will help you to contol camera angles.
You can also try to train LoRA on your influencer's face/body to keep the character consistent.
D-ID or HeyGen is for making tailored talking-head videos (here you can work with lip sync and voice sync)

The best thing is to train LoRA or DreamBooth model. I can even say it's a must if you want to keep your avatar recognizable frame to frame.

Could you pls clasify what is more interesting for you - NSFW or SFW models?

Both of you seem very knowledgeable in this field. Nice share, guys.

I am more interested in NSFW models, @Tommy_account.

 

[Tommy_account]

I am more interested in NSFW models

In case of NSFW models the only key is using local pipelines. Cloud tools will probalby block you.
I bet the best combo is
ComfyUI + AnimateDiff + ControllNet: using for smooth motion
LoRa is a must to keed face identical
You may also try to add VAE finetunes for keeping fixed facial drift

In the end of nicely done workout do upscale/clean. You may use Topaz Video AI in case you don't know more
NSFW content is little bit harder to produce

 

[uncensoredvlad]

Interesting! I'd love to hear more about your journey

 

[Tinzo]

Interesting! I'd love to hear more about your journey

Thanks, it's been fun learning generative AI and stable diffusion. I was able to earn some followers/subs via Fanvue, but I recently got banned.

Not sure if anyone knows of any other subscription-based platforms that would work with AI-generated models.

 

[ltesocksio]

How did you manage to keep the character consistent? Would you mind sharing your prompt?