CivArchive
    Warspite (Azur Lane) - ウォースパイト - v1.0
    NSFW
    Preview 703289
    Preview 577214
    Preview 576801
    Preview 576890
    Preview 577198
    Preview 583134
    Preview 576625

    A LoRA of Warspite (specifically the Azur Lane one). Should generally work on different models and cooperate nicely with other LoRAs etc. (at least from what I've tested).

    I'll iterate on the current model in the future and hopefully refine it a bit more, but this version seems to work well enough for now.

    Description

    Trained on 149 images with 2 repeats for 7 epochs at a network dimension of 16 and alpha of 8. Seems to me like it's overfitting slightly. Works well enough on AOM3, especially in hires renders.

    The model trained too quickly for my liking, I'll try adjusting learning rate and maybe fiddle around with splitting the dataset into a character and costume subset. Experimented with it a bit but couldn't get it to work properly yet. Look forward to that in a future version (maybe).

    FAQ

    Comments (17)

    LapiinApr 23, 2023
    CivitAI

    hey what model and vae is first picture?

    pensive_penguin
    Author
    Apr 23, 2023

    It's AbyssOrangeMix 3 A3 and it's respective VAE. I got it from huggingface (https://huggingface.co/WarriorMama777/OrangeMixs), I guess the resource viewer drops the specifc version for some reason

    LapiinApr 23, 2023· 1 reaction

    wow thank you and looks great

    LapiinApr 23, 2023

    did you also used fix,hires, somehow ? when resizing?

    pensive_penguin
    Author
    Apr 23, 2023

    @Lapiin Yeah, for all of them. Latent nearest exact for the AOM3 ones I believe and ESRGAN for the other one. 2x scaling each. Denoising Ratio of 0.7 and hires steps equal to generation steps, at least for the AOM ones

    LapiinApr 23, 2023

    @pensive_penguin thank you for that infooo can i add you on discord maybe ? ♥

    pensive_penguin
    Author
    Apr 23, 2023

    @Lapiin I don't really use discord currently, but I'll let you know when I do!

    LapiinApr 24, 2023

    hmm for AOM2 Latent nearest exact Denoising Ratio of 0.7 change picture too much and on lower scale its blury picture can you help me out maybe with this?

    pensive_penguin
    Author
    Apr 24, 2023

    @Lapiin I'll try to test this on AOM2 later, but I currently only have access to my shitty graphics card so I might have to put that off for a bit. But I'd recommend using AOM3 A3 if you're aiming for the same style as the sample images I posted.

    I'll probably test this a bit more throughly and work on a better revision of the model at some point soon-ish, so if you find anything else that seems off feel free to let me know!

    LapiinApr 24, 2023

    @pensive_penguin ohh sure so in AOM 3 in this settings,picture not changing?

    or do you used also control net or something?

    pensive_penguin
    Author
    Apr 24, 2023

    @Lapiin No, it's just vanilla AOM3 A3.

    I wanted to render it again but I just realized that the thumbnail image doesn't have a seed for whatever reason, so I'll just test with some other image and see if there is something off with the results while I'm on a completely separate PC to my usual one.

    Actually, I also might have had xformers active when I rendered that, which might fuck up results as well, or so I heard at least.

    pensive_penguin
    Author
    Apr 24, 2023

    @Lapiin OH WAIT which sampler are you using? with AOM3 in my experience Euler doesn't work well and usually results in blurry images, use DPM++ 2M Karras or DPM++ SDE Karras

    LapiinApr 24, 2023

    @pensive_penguin DPM++ 2m Karras mosty but like i said on 0,6-07 image keep changing position from seed one like other legs or diffrent clothes

    pensive_penguin
    Author
    Apr 24, 2023

    @Lapiin Yeah, I also just rendered a test and it looks different from the sample image as well. I guess I have to check my setup on the PC I rendered that on, no idea where the difference is coming from

    LapiinApr 24, 2023

    @pensive_penguin ohh haha okay so please let me know when you find out how to do same image with fix,hires but better colors and shadows

    loli_collegeMay 6, 2023
    CivitAI

    What are the promt, width, height used for first pic?

    pensive_penguin
    Author
    May 6, 2023

    Here is the exact image info dump:

    masterpiece, best quality, 1girl, <lora:azur_lane_warspite_v1:.9>, warspite

    Negative prompt: nsfw, (worst quality, low quality:1.4), (lip, nose, tooth, rouge, lipstick, eyeshadow:1.4), (blush:1.2), (jpeg artifacts:1.4), (depth of field, bokeh, blurry, film grain, chromatic aberration, lens flare:1.0), (1boy, abs, muscular, rib:1.0), greyscale, monochrome, dusty sunbeams, trembling, motion lines, motion blur, emphasis lines, text, title, logo, signature,

    Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 8, Seed: 502696844, Size: 360x640, Model hash: eb4099ba9c, Denoising strength: 0.7, Clip skip: 2, Hires upscale: 2, Hires upscaler: Latent, Hashes: {"vae": "f921fb3f29", "lora:azur_lane_warspite_v1": "fd533bd1b7", "model": "eb4099ba9c"}

    LORA
    Other

    Details

    Downloads
    1,681
    Platform
    CivitAI
    Platform Status
    Available
    Created
    4/23/2023
    Updated
    5/13/2026
    Deleted
    -
    Trigger Words:
    warspite

    Available On (1 platform)

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