CivArchive
    Add More Details - Detail Enhancer / Tweaker (细节调整) LoRA - v1.0
    NSFW
    Preview 995787
    Preview 995798
    Preview 995801
    Preview 995802
    Preview 995806
    Preview 995812
    Preview 995813
    Preview 995814
    Preview 995816

    Add More Details - Detail Enhancer / Tweaker

    Do you like what I do? Consider supporting me on Patreon 🅿️ or feel free to buy me a coffee ☕

    I've been using the (great) Detail Tweaker by CyberAIchemist for a while now, and I was really curious. I really just wanted to know if I was capable of doing something similar.

    I had a lot of fun doing this, and I think it came out pretty great. I wil ldefinitely incorporate it in most of my work going forward.

    You should use this between 0.5 and 1 weight, depending on your preference. You can go lower than 0.5 for a more subtle effect, of course.

    I also found out that this gives some interesting results at negative weight, sometimes. See the examples to see what I mean.

    How to use LoRA's in auto1111:

    • Update webui (use git pull like here or redownload it)

    • Copy the file to stable-diffusion-webui/models/lora

    • Select your LoRA like in this video

    • Make sure to change the weight (by default it's :1 which is usually too high)

    Description

    FAQ

    Comments (62)

    jomceyJun 1, 2023· 4 reactions
    CivitAI

    You kept teasing us with this. Glad you finally uploaded!

    Lykon
    Author
    Jun 1, 2023

    it just wasn't ready :D

    DutchgrasssJun 1, 2023· 3 reactions
    CivitAI

    it's hard to tell these days if this is embedded into a checkpoint so tricky when or not to use it. Did disable it when i was trying your latest realistic model, will retry in some moment with this activated. Anyway, thanks for your work and that you share them.

    Lykon
    Author
    Jun 1, 2023

    Absolute Reality is great on its own. I advise you only use detailer enhancements if you feel you need them or to achieve a specific style.

    dekejiJun 1, 2023· 3 reactions
    CivitAI

    uff... an interesting LoRa, I didn't know I needed it until I saw it hehe... I will definitely use it when generating images

    LDWorksDavidJun 1, 2023· 5 reactions
    CivitAI

    More interested in the -1 value than 1 but good job here.

    Lykon
    Author
    Jun 1, 2023

    yeah, it's a way to reduce detail :)

    LeonTheLordJun 2, 2023· 1 reaction
    CivitAI

    so how was ur work different from CyberAIchemist? hope u not just simply copy and submit.

    Lykon
    Author
    Jun 2, 2023· 2 reactions

    they're quite similar but it's not a copy. CyberAIchemist's one is great, you can keep using that.

    puma0000Jun 18, 2023· 1 reaction

    The fact this one has a lower range makes it easier to use for me, so I prefer it.

    ktiseos_nyxJun 2, 2023· 10 reactions
    CivitAI

    lays at Lykon's feet LORD LYKON WHAT MUST WE PAY YOU FOR SUCH GREAT CONTENT - coughs

    SolarisJun 2, 2023· 2 reactions

    I've been in the same state since Lykon released CLAMP and Yoshitaka Amano lora lol

    lays on the side too

    FloorPuddingJun 3, 2023· 1 reaction

    Lykon a real one

    NikJoshJun 2, 2023· 7 reactions
    CivitAI

    So what are the differences between this and the other one? How is it even done? To my feeble eyes, they both seem to have the same function.

    Lykon
    Author
    Jun 2, 2023

    they do. If you read the description, this mostly started as an experiment just to see if I could make it.

    NikJoshJun 3, 2023

    @Lykon Yeah I did, that's why I wanted to ask if there was any distinctions for specific use cases between them, sounds like they're basically the same so thanks for clarifying.

    JoeanAmierJun 8, 2023· 1 reaction
    CivitAI

    这种细节调整的lora是怎么训练的？要怎样的数据集？

    454564Jun 19, 2023· 1 reaction

    bilibili 【完美炼丹术，差异炼丹法】最新LoRA训练进阶教程10

    idolJun 23, 2023· 1 reaction
    CivitAI

    有sd2.1的版本嗎?

    Lykon
    Author
    Jun 23, 2023· 1 reaction

    not that I'm aware of.

    Bibi_AIJun 23, 2023· 1 reaction
    CivitAI

    Why do I always get this ERROR (below) when I use this LoRA? Please point out for me the solution, I'm not a tech person so I don't understand how to solve it. Thanks! The ERROR as below (it always appears no matter when I use this LoRA solitarily or in combination with other ones. As long as I use this LoRA in my prompt, the ERROR shall appear):

    "WARNING:tensorflow:6 out of the last 6 calls to <function Model.make_predict_function.<locals>.predict_function at 0x000001D92AF5E290> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has reduce_retracing=True option that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for more details."

    Please help me. Thanks!

    Lykon
    Author
    Jun 26, 2023· 1 reaction

    this is a normal lora, maybe the file is corrupt? Try to redownload it.

    PechistaJun 25, 2023· 4 reactions
    CivitAI

    Any plans to port this to v2.1 model?

    Lykon
    Author
    Jun 26, 2023· 1 reaction

    no.

    AbelRussJun 28, 2023· 1 reaction
    CivitAI

    what are the trigger words for this lora? nothing is listed in the profile

    joparebrJun 30, 2023· 4 reactions

    <lora:more_details:1>

    klhscottJul 13, 2023· 3 reactions

    I am actually quite curious why the trigger words are needed since we usually add <lora:**:1>

    WeisseningBlitzJul 15, 2023· 8 reactions

    @klhscott Some lora do more than one thing and need the triggers to fine tune. Though, for some, yeah, just a weight value sufficient.

    juanluixNov 25, 2023· 5 reactions

    [@klhscott]

    Note: Translation from Spanish (Spain) to English.

    The text may contain semantic errors.

    ───

    I've been testing the tags.

    Whenever I can I look at all the tags whether they are power 50 or 1

    When you only put the instruction <lora:more_details:1> you are not really giving it the power it needs.

    It is as if I received a light part of the lora, even though it is at 1

    There are times that it is not even activated, because it acts like any other word from the prompt and it is possible that it will not even be used when generating.

    The typical thing to put a word but it doesn't really appear in the image.

    I suggest you see more information about lora and how stable diffusion actually uses it. (Stable Diffusion or any other, I guess)

    Note: If there is no tag or you do not want to use one from the list, you can use the file name itself as a tag

    So if it doesn't have a tag, what you should put in the prompt (this info is what Stable Diffusion itself recommends when you use lora)

    <lora:fileName:Power> tag/FileName

    Option 2 (no tag): <lora:more_details:Power> more_details

    <lora:fileName:Power> This is normal and depending on the configuration you have, it is possible not to use the file name since in Automatic1111 you have 2 options to call the lora.

    tag: Normally you use the tag that comes with the lora and choose if there is more than 1 tag.

    If it does not have a tag, you put the file name again

    ────

    For this lora, it doesn't seem like he needs much.

    With 1 tag it is the minimum to activate the lora 100%.

    So that you understand it in broad strokes.

    It's like a person missing an arm.

    Yes, it is the person, but not the whole person.

    WolfenrisJul 2, 2023· 7 reactions
    CivitAI

    I tested it with a lot of model. This is insanely good, great work thanks :)

    rufustheruseJul 14, 2023· 12 reactions
    CivitAI

    🦋 Butterflies. 🦋 Butterflies everywhere. Is there a good way to use this and not get butterflies all over the image? My negative prompt skills seem lacking in getting results here.

    ohtobeaone111Aug 10, 2023· 3 reactions

    Mine is filled with birds. Swarms of birds.

    97BuckeyeJul 27, 2023· 7 reactions
    CivitAI

    Will you update this for use in SDXL?

    Lykon
    Author
    Jul 27, 2023· 3 reactions

    can't be updated, needs to be redone :)

    jopixaJul 28, 2023· 1 reaction
    CivitAI

    I use StableDiffusion directly in Google Colab.

    With Hugging Face I use:

    pipe = StableDiffusionPipeline.from_pretrained( .......

    Then I run the prompt

    images = pipe(prompt,.....

    How can I load something from Lora to use it this way without using webUi?

    sunatte1Aug 17, 2023· 2 reactions

    you may use comfy UI to use loras

    StartAISep 4, 2023· 1 reaction

    Sorry for the stupid question but how this pipe prompts work? Any Space there or article that can make me understand?

    sunatte1Sep 4, 2023· 1 reaction

    @StartAI what do you mean by pipe prompts?

    Do you mean to use models using pipes or do you want to loop prompts?

    StartAISep 22, 2023· 1 reaction

    @sunatte1 Just saw this that you wrote: "Then I run the prompt

    images = pipe(prompt"

    sunatte1Sep 28, 2023· 1 reaction

    @StartAI please share your complete code with me at [email protected], I will correct it for you

    StraitjacketAug 5, 2023· 1 reaction
    CivitAI

    This is a great Lora, but it seems to disable my Lycoris models when I use it, does anybody know why that might be??

    And when I say 'disable' I mean it makes it so I literally cant even re-add any lycoris to my prompt (and have it function) even after turning this Lora off again, I have to reload the UI or change models to fix it! Very weird, any ideas??

    juanluixNov 25, 2023· 1 reaction

    You most likely have some conflict with the extensions.

    I guess you already know that. but Lycoris uses a different directory.

     

    But just in case:

    LyCoris: ...\models\LyCORIS

    LoRa: ...\models\Lora

    z0_0d713Aug 19, 2023· 7 reactions
    CivitAI

    Buddy, you save my life.

    Latent Hires. fix always changes the image composition too much because it only work over 0.4. This LORA is a good alternative in this situation.

    qazxswOct 7, 2023· 7 reactions
    CivitAI

    This Lora does added more details to my image. However, it also makes my image look "dirtier" if you know what I mean.

    I tested it with absolutereality, probably works better with non-realism image generation.

    codegixJan 20, 2024· 1 reaction

    It does add more detail, and also make it less detail with minus value

    qazxswJan 25, 2024· 1 reaction

    @codegix I was referring to recommended setting, but you are right about lower the weight

    civinootFeb 23, 2024· 2 reactions
    CivitAI

    I'm sucha newbie dummy. Can this work with img2img? I'm playing around with the weights and at best, it'll create a similar character to my image (text-wise, but not visually), but also turned the other way/angle/etc.

    I can't for the life of me... get the same picture.. just with more detail.

    terasineFeb 24, 2024· 3 reactions

    It does, yes. You'll still have to get the prompt you need for the image you want, and use the image itself as a base, then adjust the weight how you need it

    hnico2193Feb 26, 2024· 10 reactions
    CivitAI

    I'm getting a huge performance boost when prompting this: "<lora:more_details:1> <lyco:more_details:1>" like 6 sec. instead of 20 sec. only works if lyco is 1 and im getting the exact same image without lyco. is this commonly known?

    NeatWolfMar 21, 2024· 1 reaction

    For real? I'd love to test this out - Could you please share your sampling method+base settings?

    MiloslavMar 29, 2024· 2 reactions
    CivitAI

    Круто!!!

    MaratekMay 12, 2024· 7 reactions
    CivitAI

    Great! Brings pleasant little things into every Prompt!

    one4allplugin145May 15, 2024· 9 reactions
    CivitAI

    Makes most males, female, needs some tweaking!

    tobleronDec 11, 2024· 1 reaction

    If you use it with EpicPhotogasm instead of Dreamshaper8 you will avoid this problem.

    Robert_SinclairJul 1, 2024· 8 reactions
    CivitAI

    very good.

    PersonopijoijOct 18, 2024· 7 reactions
    CivitAI

    this is frustrating, what is the trigger word?

    Gunn3r71Oct 21, 2024· 1 reaction

    It's a Lora, just add the lora to the prompt and weight it.

    FUCKUASTARIDec 23, 2024· 1 reaction

    @Gunn3r71 how do you go abt that?

    xxxpasteles167Jan 22, 2025· 4 reactions

    <lora:more_details:1>

    add it at the end of your positive prompt

    LuminalArtMar 17, 2025· 4 reactions

    @Gunn3r71 If you go to your LoRA tab in the WebUI, you can click the LoRA you want to add and it'll add the word into the positive prompt. Then adjust the "1" to what you prefer.

    KazSanApr 5, 2025· 3 reactions
    CivitAI

    The drawing becomes more detailed, but unfortunately the eyes of the person become strange.

    RaimmondJun 11, 2025· 1 reaction

    There is a node for faces that actually do a good job (Face Detailer). Can't remember if works with SD or Flux.

    LORA
    SD 1.5
    by Lykon
    Download (Beta) View on CivitAI
    detail
    tool
    landscape
    portraits
    lora
    enhancer
    detailer
    details
    backgrounds
    detail tweaker

    Details

    Downloads
    401,175
    Platform
    CivitAI
    Platform Status
    Available
    Created
    6/1/2023
    Updated
    5/11/2026
    Deleted
    -

    Files

    more_details.safetensors

    Size:
    9.11 MB
    SHA256:
    d9cf2f88dec53a8dd8c09aba2b1d67963b265ebdf8255660f5bcd7f1c8edba33

    Mirrors

    HuggingFace (187 mirrors)
    CivitAI (5 mirrors)
    CivitasBay (1 mirrors)
    ShakkerAI (2 mirrors)
    TensorArt (1 mirrors)

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.

    CivitasBay
    Add More Details - Detail Enhancer / Tweaker (细节调整) LoRA - v1.0