Views
No views yet
linoyts/emoji-krea2-r32-default3d emoji. Describe the subject in plain words and the LoRA renders it
as a glossy, isolated-on-white 3D emoji / avatar.diffusers main, so install from source:pip install -U git+https://github.com/huggingface/diffusers1import torch
2from diffusers import Krea2Pipeline
3
4# Load the LoRA onto Krea 2 Turbo (the 8-step distilled inference model).
5# The LoRA was trained on Krea 2 RAW but expresses strongly on Turbo.
6pipe = Krea2Pipeline.from_pretrained("krea/Krea-2-Turbo", torch_dtype=torch.bfloat16).to("cuda")
7pipe.load_lora_weights("linoyts/emoji-krea2-r32-default")
8
9# Turbo recipe: 8 steps, no classifier-free guidance (guidance_scale=0.0).
10# Resolution-aware timestep shifting is applied automatically by the scheduler — no `mu` needed.
11image = pipe(
12 "a wise wizard with a long white beard and a tall blue pointed hat, 3d emoji",
13 num_inference_steps=8,
14 guidance_scale=0.0,
15 height=1024,
16 width=1024,
17 generator=torch.Generator("cuda").manual_seed(0),
18).images[0]
19image.save("emoji.png")krea/Krea-2-Raw (validation on krea/Krea-2-Turbo), via the Krea 2 DreamBooth LoRA trainer.linoyts/3d-emoji-1024 — 306 emoji upscaled to 1024px and re-captioned (subject description + 3d emoji anchor), following the guide's style-LoRA captioning advice.3e-4 constant, 8-bit AdamW, 1600 steps, 1024px, bf16, gradient checkpointing + cached latents.other.