Skip to main content

LORA's / Fine Tune : for image models

Feature RequestsLoRAs

Add the option to add custom Lora’s from web libraries like CivitAI - Hugging face : this give’s user a wider option when generating images… that matches more a style of their liking.

This should be unlocked for every image model, not just Flux. as it seems that most LORA’s / Fine Tunes’s are using Flux models….. even Stable diffusion, or any model that supports a LORA / Fine tune.. should have the option

( This gives user immense creative potential that fit more they’r personal needs…. a thing that standard Image style’s cannot ever achieve )

IMAGE MODELS

Ps. even most image models out there should be on the model list:

Just to name a few:

Flux ultra = ( more megapixels ) / Flux RAW’’ = ( more natural look )

Redpanda, Ideogram, ecc………..

/more features in the image section can be discussed…, but this could be a fair upgrade/

as image section is still obsolete… further improvements can not only attract domestic users… but even new user to the blockchain

if they keep adding stuff like this…. Venice will be main competitor in all-rounder AI

Status: Backlog4 comments

Log in to comment and vote

Comments4

  • Tan Pine

    •

    Oct 30

    Support has now confirmed this in writing on a ticket:

    • "the API does not provide a way to list which models use LoRAs or which specific LoRA is attached"

    • "There is also no endpoint that exposes LoRA metadata"

    • "clients cannot programmatically detect when lora_strength is active"

    • "The parameter simply exists for models that use additional LoRAs, and it has no effect on models that do not"

    So lora_strength ships as an undetectable no-op. There's no way to know which models honour it, which LoRA is loaded, or whether a value did anything at all.

    Minimum fix that makes the existing parameter usable: an endpoint listing models with an attached LoRA and its identity. The fuller fix is a lora field so users can supply one. Neither needs open upload — a curated allowlist is enough.

  • Tan Pine

    •

    Oct 30

    This is the one that counts; it's already tagged LoRAs and sitting in Backlog.

  • Sapphire Jam

    •

    Mar 26, 2025

    •

    Merged request

    •

    1 vote

    Add fine-tuning functionality

    Would be good to be able fine-tune LLM models and use then in API

  • Aquamarine Teacup

    •

    Mar 25, 2025

    Above all: The Pony Realism Enhancer LoRA