Qwen Image LoRA trainings Stage 1 results and pre-made configs published

Question

Qwen Image LoRA trainings Stage 1 results and pre-made configs published

FurkanGozukara posted Sep 4 3 min read

Qwen Image LoRA trainings Stage 1 results and pre-made configs published - As low as training with 6 GB GPUs - Stage 2 research will hopefully improve quality even more - Images generated with 8-steps lightning LoRA + SECourses Musubi Tuner trained LoRA in 8 steps + 2x Latent Upscale

1-click to install SECourses Musubi Tuner app and pre-made training configs shared here : https://www.patreon.com/posts/137551634

Hopefully a full video tutorial will be made after Stage 2 R&D trainings completed
Example training made on the hardest training which is training a person and it works really good. Therefore, it shall work even much better on style training, item training, product training, character training and such
Stage 1 took more than 35 unique R&D Qwen LoRA training
1-Click installer currently fully supporting Windows, RunPod (Linux & Cloud) and Massed Compute (Linux & recommend Cloud) training for literally every GPU like RTX 3000, 4000, 5000 series or H100, B200, L40, etc
28 images weak dataset is used for this training
More angles having dataset would perform definitely better
Moreover, i will make a research for a better activation token as well rather than ohwx
After Stage 2, I am expecting hopefully much better results
As a caption, i recommend to use only ohwx nothing else, not even class token
Higher quality more images shared here : https://medium.com/@furkangozukara/qwen-image-lora-trainings-stage-1-results-and-pre-made-configs-published-as-low-as-training-with-ba0d41d76a05
Image prompts randomly generated with Gemini 2.5 in Google AI Studio for free

How to Generate Images

In the zip file of this post : https://www.patreon.com/posts/114517862
We have Amazing_SwarmUI_Presets_v21.json made for SwarmUI
Import it and i am using Qwen Image 8 Steps Ultra Fast to generate images and then apply Upscale Images 2X to make them 4x resolution (1328x1328 to 2656x2656)
Of course in addition to preset don't forget to select your trained LoRA - I used LoRA strength / scale = 1
This tutorial shows it : https://youtu.be/3BFDcO2Ysu4

image/png

image/jpeg

If you read this far, tweet to the author to show them you care. Tweet a Thanks

chevron_left

Andrew Mewborn · Answer 1 · 2025-10-10T12:38:04+0000

Appreciate the detailed breakdown here—it’s clear a lot of testing went into this stage of R&D. The results already look promising, especially considering the small dataset. How much improvement do you expect once Stage 2 introduces better activation tokens and expanded image angles?

	Wan 2.2, FLUX, FLUX Krea & Qwen Image Just got Upgraded: Ultimate Tutorial for Open Source SOTA Imag FurkanGozukara - Aug 19
	Hunyuan Image 2.1 by Tencent Full Tutorial and 1-Click to Install Ultra Advanced App to Use Locally FurkanGozukara - Sep 10
	Nano Banana (Gemini 2.5 Flash Image) Full Tutorial, 27 Unique vs Qwen Image Edit FurkanGozukara - Aug 27
	Qwen Image Edit Full Tutorial: 26 Different Demo Cases, Prompts & Images, Pwns FLUX Kontext Dev FurkanGozukara - Aug 23
	Ovi is Local Version of VEO 3 & SORA 2 - The first-ever public, open-source model that generates bot FurkanGozukara - Oct 11

Qwen Image LoRA trainings Stage 1 results and pre-made configs published