Prompt

Icon Kit, An animated image of a yellow and purple jar. The jar has a white lightning bolt on it. The background is a bright blue.

Prompt

Icon Kit, An eye-level view of a vibrant orange jack-o-lantern against a backdrop of a deep blue sky. The pumpkins face is angled towards the left, with a black outline around it. The eyes are squinted with white lines, creating a striking contrast with the orange and yellow of the pumpkin. The orange is adorned with black lines, adding a pop of color to the scene.

Prompt

Icon Kit, An animated image of a red canister with a black cap on the top. The canister has a black stripe on the side and a yellow hose attached to the cap. The background is a bright blue color.

Prompt

Icon Kit, Captured from a low-angle perspective on a vibrant blue backdrop, a black and white graphic is depicted in the center of the frame. The graphic features a black round object with a white outline of a skull in the middle of it. The skull has three black circles around it, and a black circle around it. A small orange carrot is sticking out of the top of the object, adding a pop of color to the scene.

Prompt

Icon Kit, An animated image of a rustic brown leather satchel with a brass buckle. The bag has a small green vine growing out of its side pocket. The background is a gradient of light green and beige, adding an earthy vibe.

Prompt

Icon Kit, A side-view animated image of a flaming red and orange torch with flickering flames. The handle is wooden with black iron bands. The background is a dark, smoky gray, making the flames stand out vividly.

Model description for Flux-Icon-Kit-LoRA

Image Processing Parameters

Parameter	Value	Parameter	Value
LR Scheduler	constant	Noise Offset	0.03
Optimizer	AdamW	Multires Noise Discount	0.1
Network Dim	64	Multires Noise Iterations	10
Network Alpha	32	Repeat & Steps	25 & 3100
Epoch	20	Save Every N Epochs	1

Labeling: florence2-en(natural language & English)

Total Images Used for Training : 40 [ Raw 8-bit ]

Best Dimensions & Inference

Dimensions	Aspect Ratio	Recommendation
1280 x 832	3:2	Best
1024 x 1024	1:1	Default

Inference Range

• Recommended Inference Steps: 30–35

Setting Up

import torch
from pipelines import DiffusionPipeline

base_model = "black-forest-labs/FLUX.1-dev"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)

lora_repo = "strangerzonehf/Flux-Icon-Kit-LoRA"
trigger_word = "Icon Kit"  
pipe.load_lora_weights(lora_repo)

device = torch.device("cuda")
pipe.to(device)

Trigger words

You should use Icon Kit to trigger the image generation.

Download model

Weights for this model are available in Safetensors format.

Download them in the Files & versions tab.

---