AI videos

Anyone tried messing around with Wan 2.1 on their own computer? It's crazy hard to get good results.

Been playing with it but I hate ComfyUI. I do get some okayish results but for that I mostly need loras. Biggest pain in the ass is that the updates keep breaking either comfy itself or the nodes.
 
Been playing with it but I hate ComfyUI. I do get some okayish results but for that I mostly need loras. Biggest pain in the ass is that the updates keep breaking either comfy itself or the nodes.

Yeah I need to start messing with loras. Also hate ComfyUI.
 
Been playing with it but I hate ComfyUI. I do get some okayish results but for that I mostly need loras. Biggest pain in the ass is that the updates keep breaking either comfy itself or the nodes.

Okay I got Loras going last night and the results are night and day better. I have no idea how CivitAI makes money but they're doing the lord's work hosting all this shit.
 
Okay I got Loras going last night and the results are night and day better. I have no idea how CivitAI makes money but they're doing the lord's work hosting all this shit.

I think they make money by people buying credits to make images, loras and checkpoints with their servers. But yea loras make all the difference 👌
 
Been trying some of the free ones. I wonder how realistic it is to fix the problems with inconsistencies, and I want some tools to help specify, I can never make exactly what I want. I feel its harder to get that done than we expect.

fPcD88jmBz0oKcPk.gif
 
I can't stop revisiting the basement.

VJHGJK0k6FCNFkJz.gif


these tools are going to be integrated into realtime VR like the holodeck before long and we're not ready for it

GyKTfO.jpg
 
Last edited:
> Look like Deliriant trip mimic varying from dark(DPH/Benadryl) to surreal humour(Muscimol).

Love how others look more like Autistic psychosis(Think Ketamine + Shrooms) than actual Schizopherina. Since It like was the AI in another reality then showing what It saw situation.
 


Just one example of using a high-realism image model paired with a different video generation model, which illustrates just how realistic the outputs can be. These videos are just about at the threshold now where no one will be able to tell the difference anymore.
 
Top Bottom