Sign In

SD Forge Neo Lora Control - SD Webui Loractrl Extension

Updated: Oct 10, 2026

tool

Download

1 variant available

Archive Other

sd-webui-loractl.zip

5.72 MB

Verified:

Type
Other
Stats

27

Reviews
Published

Oct 10, 2026

Base Model

Other

Hash
AutoV2
C642EE73B8
default creator card background decoration
Followers - 7

7

Likes - 14

14

00148-2171385951.png

SD-WebUI-Loractl (Fixed for SD WebUI Forge Neo)

A working fix of the Loractl extension for Stable Diffusion WebUI Forge Neo. While I primarily use this with anima models, it should works across other models too.

What Does It Do?

Standard LoRA loading applies a static weight across 100% of the generation steps, which often leads to styles bleeding into each other or concept LoRAs fighting for priority.

Loractl gives you dynamic step-by-step control over when a LoRA activates, ramps up, fades out, or turns off completely. This allows you to:

  • Blend multiple LoRAs smoothly without prompt "baking" or visual clashing.

  • Lock in a concept/subject LoRA during early steps (e.g., 0%–40%), then fade it out so a style LoRA can take over the fine details.

  • Trigger a LoRA exclusively during the middle steps of sampling.

Installation

  1. Download the zip and extract.

  2. Drop the folder directly into your WebUI's extensions/ directory: stable-diffusion-webui-forge/extensions/

  3. Restart WebUI completely.

Syntax & Usage

The basic prompt syntax is:

<lora:lora_name:weight@step_start,weight@step_end>

  • Weight: LoRA strength (e.g., 1.0, 0.5, 0).

  • Step: Generation timeline expressed as a decimal from 0 to 1 (e.g., 0.4 = 40% of steps).

  • Multiple Keyframes: Chain as many comma-separated points as you need to build curves, tapers, or mid-generation bursts.

Example -

1) Standard generation with lora strength 1 :

<lora:scr4pbook_anima:1> <lora:KishimotoMasashi_AnimaBaseV10_byKonan:1>

2) Dynamic hand-off with Loractl:

<lora:scr4pbook_anima:1@0,[email protected]> <lora:KishimotoMasashi_AnimaBaseV10_byKonan:0@0,1@1>

  • scr4pbook_anima: Starts at full strength (1.0) to establish initial composition, then gradually ramps down to 0 by 50% of the generation.

  • KishimotoMasashi: Starts at 0 and gradually scales up to 1.0 by 100%, letting the art style take over the detailing without corrupting the initial layout.


You can also define multi steps as to activate a lora in middle of the step or change strength at each step manually.

Example of a three step -

<lora:Mash_Kyrielight_(Dangerous_Beast)-Fate_(Series)-variant-10546-anima-v1:1@0,[email protected],0@1>

  • 1@0: Starts at full strength (1.0) during early composition.

  • [email protected]: Tapers down to half strength (0.5) at the midpoint (50%).

  • 0@1: Drops completely to 0 by the final step (100%), keeping the finish clean.

This was the original extension from which you can get more info - https://github.com/cheald/sd-webui-loractl