wavymulder

I make stuff.
Released models/LoRAs/hypernets - @StabilityAI @visualelectric

[wavymulder @wavymulder]

New #Dreambooth model: Analog Diffusion - Get it on @huggingface: https://t.co/PD3EXODBxu

#stablediffusion

[wavymulder @wavymulder]

@meltycanon that "do the most" vibe

[wavymulder @wavymulder]

@fofrAI what gets me is the clothing. fashion is 100% on brand for CES-goers

[wavymulder @wavymulder]

@HelloCivitai @huggingface Thank you to @FAL for the compute🙏

[wavymulder @wavymulder]

Overlord Style LoRA for FLUX.1-dev is now available! This style makes for a great general-purpose anime LoRA. Get it now on @HelloCivitai https://t.co/VN9KmF8L3C and @huggingface https://t.co/jleOvwjF8g

[wavymulder @wavymulder]

one of the first things I trained (nearly two years ago!) was a Hypernet for SD1.5 on the style from the anime Overlord - and now I've trained a flux lora with that same dataset! Will post soon :)

[wavymulder @wavymulder]

trained a FLUX lora on the tarot cards from Cyberpunk 2077. Makes cool images but had heavy losses in prompt adherence. Probably won't retrain/recaption, but may still share the lora because the style is sick.

[wavymulder @wavymulder]

@huggingface @HelloCivitai thank you to @FAL for the compute 🔥

[wavymulder @wavymulder]

ASCII art LoRA for FLUX.1-dev! Get it on @huggingface https://t.co/hxz0bWVy0v or @HelloCivitai https://t.co/H8uh8PUuB5 - my first LoRA trained for flux 😁

[wavymulder @wavymulder]

@FAL

[wavymulder @wavymulder]

flux lora🔥goofing around with an ascii art dataset. it can't always create letters/characters, but even when it fails it gives a really cool dither/halftone effect. may share! ty @FAL for the compute

[wavymulder @wavymulder]

flux is pretty neat

[wavymulder @wavymulder]

@rileyj_s @visualelectric i can dig it (using the Fashion Photography style)

[wavymulder @wavymulder]

@jfischoff the random seed is an offering to the Diffusion Gods

[wavymulder @wavymulder]

@jfischoff @isidentical Panda Express is better imo