Using clips from the backrooms,
we trained our Hunyuan Video Lora (13 hours, H100) using Uniform Frame Extraction at 504x360 resolution
Prompt Template:
"a video of KBK backrooms, [prompt] , found footage in the style of KBK"
Description
FAQ
Comments (8)
Ohhhh yes. There goes my weekend.
This lora really seems to appreciate high resolutions. I struggled to get good output below 720x480p
really cool! I made one for SD1.5 about a year ago https://civitai.com/models/400181
if you made backrooms model before, i'm glad you like my video model, it was trained for 14 hours on H100, just to see how deep into the backrooms we can go!
this looks awesome, and model hallucinations looks artistuc lol
yes and you can prompt stuff you want to "glimpse" too, it can be quite accurate
This lora is going to break the internet, or maybe my brain, if I keep prompting, do I noclip into the backrooms? Thank you for making this, bestjohnson!
Yes.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.