CivArchive
    Qwen Enhancer - Higher Quality - v1.0
    NSFW
    Preview 104866157
    Preview 104866159
    Preview 104866155
    Preview 104866160
    Preview 104866162
    Preview 104866149
    Preview 104866140
    Preview 104866151
    Preview 104866153
    Preview 104866147
    Preview 104866148
    Preview 104866154
    Preview 105733138
    Preview 105733385
    Preview 105733556

    ❤️ An All in One LoRa ❤️

    Enhance any image of Qwen to better quality, more beauty, more details, better anatomy.

    Description

    FAQ

    Comments (19)

    koto2091187Oct 8, 2025
    CivitAI

    It leaves visible lines and tiles on the image, how do I fix it?

    vladulidloOct 9, 2025

    I also see the grid pattern and I see this in many Qwen Image outputs. I only focus on the pattern and it is very disturbing.

    takomli2013979Oct 9, 2025

    I think try reduce the strength of lora.

    amin230
    Author
    Oct 9, 2025

    the prompt you use is important to look. for photo view you should just use photo related prompt, any artistic prompt like "abstract" or "illustration" can change the image totally.

    the other thing is you can play with the weight of lora to reach close to your wish.

    zz5Oct 9, 2025· 1 reaction

    Try the Qwen-Image-Edit-2509 model. In my tests it did not produce the grid/lines. Instead it will give you a fine grained dithering pattern, which looks much better IMHO. I used GGUF_Q6_K from QuantStack.

    Edit:

    It seems the lines are not completely gone in some cases, but still they are much less pronounced.

    shalegriOct 10, 2025· 2 reactions

    use flux as refiner on denoise 0.15, steps 5+ and you will be happy)))

    vladulidloOct 12, 2025

    @shalegri I mainly use Flux because Qwen is much slower for me and seems lower quality output. Refining using Flux misses the point as this will need multiple models and slow it even more.

    @shalegri i do that all the time too. LOLLL

    jmxbfOct 8, 2025· 1 reaction
    CivitAI

    thanks!

    JellaiOct 8, 2025· 3 reactions
    CivitAI

    Interesting. It seems to like symmetry more. I had been working on an image for a while attempting to have a dynamic camera angle, and it was going okay. I put this lora in for a bit of a boost, but it flattened the view in every generation.

    I do like the general look though, aside from the camera angle. It's a cool lora. Thank you.

    amin230
    Author
    Oct 9, 2025

    the prompt you use is very important to look. for photo view you should just use photo related prompt or just "photo" without any more prompt. Any artistic prompt like "abstract" or "illustration" can change the image totally to artistic view.

    if the problem is still there, there's no choice to reduce the weight.

    JellaiOct 10, 2025· 1 reaction

    @amin230 Yes, I was definitely going for a realistic photo in the prompt. But yeah, in the end, I just reduced the strength. Just wanted to let you know where you leans in case you make a version 2. I do think the result looks good.

    peeledkotOct 9, 2025· 3 reactions
    CivitAI

    changing the asian lady to european was brutal

    amin230
    Author
    Oct 9, 2025

    sorry I didn't mean it. the prompt I used is just "1girl" without any identity. flux decide to consider it as Asian & the lora consider it as European. if you want Asian girl, just add to your prompt o their nationality, lora just makes their look better.

    amin230
    Author
    Oct 9, 2025

    I meant Qwen, not flux :)

    AIBeauty4SeenFeb 12, 2026

    I laughed at the earlier comments. But..I now I have to say this. Your Lora file is Race Changer

    ZannarkOct 9, 2025· 3 reactions
    CivitAI

    who is quen?

    zerocool22Nov 15, 2025
    CivitAI

    It can be used with QWEN EDIT 2509 & normal creation?

    amin230
    Author
    Nov 15, 2025

    as far as I know, they have different loras. If there's a way to use a Qwen lora for Qwen Edit, I don't know. maybe others can help us.

    LORA
    Qwen

    Details

    Downloads
    1,685
    Platform
    CivitAI
    Platform Status
    Available
    Created
    10/8/2025
    Updated
    5/13/2026
    Deleted
    -