Lexi-Images-v1-pixel

Lexi T2I

Base: stable-diffusion-v1-5/stable-diffusion-v1-5
Recipe: t2i-style
Base license: CreativeML OpenRAIL-M (base) + MIT (LCM-LoRA)

License and usage restrictions

This fine-tuned adapter/weights are licensed by Reallexi LLC for personal, non-commercial use only. Enterprise or commercial use (including internal business use, paid products, or services built on this model) requires a separate commercial license from Reallexi LLC - contact https://llm-shop.reallexi.io before using it commercially. This restriction applies to this trained artifact; the base model's own license (above) governs the unmodified base separately.

Recommended inference settings

Sampling at any other step count or guidance scale is untested: this adapter was trained only at the timesteps a call with these exact settings visits.

sd-cli -M txt2img -m lexi-t2i-q8_0.gguf -p "your prompt" --cfg-scale 1.0 --steps 4 -W 512 -H 512

A higher step count or guidance scale does not improve this model: it was distilled and trained specifically for four-step, guidance-free denoising, and pushing either setting past its trained range degrades output rather than improving it.

Prompting notes

• Short, concrete prompts work better than long descriptive ones: the adapter was trained on the caption style shown in the samples below.
• This is a LoRA-scale adaptation, not a from-scratch model: expect it to shift style/subject within the base's existing capabilities, not introduce concepts absent from both the base and the training data.

Samples

Generated by the trained model at its shipping step count and resolution.

a pixel art character with square dark green glasses, a onion-shaped head and a peachy-colored body on a warm background

a pixel art character with square green and blue glasses, a weight-shaped head and a foggrey-colored body on a cool background

a pixel art character with square green and blue glasses, a doughnut-shaped head and a rust-colored body on a cool background

a pixel art character with square black sunglasses, a doughnut-shaped head and a darkpink-colored body on a cool background

Who, where, and what platform trained this?

Produced by Reallexi LLC on Reallexi AI Model Builder, a local-first training platform (https://llm.reallexi.io). Hugging Face repository: Reallexi-llc/Lexi-Images-v1-pixel. Copyright (c) 2026 Reallexi LLC. All rights reserved.