CivArchive
    Preview 106465161
    Preview 106465272
    Preview 106465284
    Preview 106465369
    Preview 106465385
    Preview 106465395
    Preview 106465505
    Preview 106465545
    Preview 106465648
    Preview 106465663
    Preview 106465686
    Preview 106465757

    If ever a model deserved for me to say that it is the best I have created, maybe it is this one, what makes this model special is that it is a DMD model made to create super fast images but unlike others, it does not lose quality despite not using Hires.fix, of course you can use it but the results still seem great to me, without going into terms of hardware because I know that we all have different ones, I managed to create sharp images in 2 seconds, using Hires.fix in 6 seconds, I know that some may find details to improve but I really hope that you can try it, it will surprise you.

    • The VAE is BAKED

    • Upscaler: Im using 4x-ClearRealityV1 upscale by 1.5 Steps 5 denois 0.2 Hires CFG 1.1,

    • Width and height: 896*1152

    • CFG Scale: 1.2

    • Steps: 4

    • The sampling method i use is LCM-Karras

    It is important to follow the recommended parameters since not all samplers work correctly with DMD models.

    This model accepts negative prompts, so for greater realism I recommend:cartoon, illustration, anime, painting, CGI, 3D render, low quality, watermark, logo, label

    I truly hope everyone can download it, use it, and share their examples, as this is the only way to continue improving. Greetings to everyone.

    Description

    After reading carefully all your comments and messages, knowing that there is always room for improvement, I decided to make a new version of this model. How did it improve? Better skin textures, trying to get away from the plastic texture mentioned in LCM samplers. I added a few more realistic but artistic elements at the same time. If you notice a tendency towards dark images, that is a personal taste, but you can make them as light as you want. What makes this model stand out? Textures, details, realism. Very good for portraits and landscapes. I really hope you can try it and share your images, since that way we will all continue to improve.

    • The VAE is BAKED

    • Upscaler: Im using 4x-ClearRealityV1 upscale by 1.5 Steps 5 denois 0.2 Hires CFG 1

    • Width and height: 896*1152

    • CFG Scale: 1.2

    • Steps: 8

    • The sampling method i use is LCM-Karras

    It is important to follow the recommended parameters since not all samplers work correctly with DMD models.

    This model accepts negative prompts, so for greater realism I recommend: cartoon, illustration, anime, painting, CGI, 3D render, low quality, watermark, logo, label

    FAQ

    Comments (10)

    AkalabethOct 18, 2025· 1 reaction
    CivitAI

    Wow, this one is really good! Nice job!

    orfeus2010441Oct 19, 2025· 1 reaction
    CivitAI

    Nice deep shadows.

    Kid_Glitch_63Oct 22, 2025· 1 reaction
    CivitAI

    I think Sateluco's models deserve thumbs ups for sure. But in my external gen workflow I need to use the neg DMD trick to get nice images.

    Can one of you genius Bros actually tell me what is happening when I use neg DMD though? I would think it is subtracting out the accelerative effects of DMD, but I don't seem to need to add any steps. Thx.

    terong_belandaOct 24, 2025· 2 reactions
    CivitAI

    right now this is the best dmd model

    NekhOct 25, 2025· 2 reactions
    CivitAI

    This is by far the most realistic model I stumbled upon !!!!! Actually impressive !!! Most of other models were not that good at raw realism to this level , specially when it comes to generating south asian people (Indian, Bangladeshi ,Sri Lankan etc). So As s beginner I don't know much about these stuffs yet but you have done some magic work there !!!!

    .......I need a help with this..the one thing I couldn't achieve 90% of time is accurate hand , finger details, Maybe I am doing something wrong ( I used comfyui default workflow) so is there a way to do that ? I used negative prompt but it didn't seem to have much effect on outcome, And I noticed that by default when i ask for a picture of a "girl, woman or man" it gives a western white skin person and the finger details are actually good but for prompt like "indian girl , Bangladeshi man ", it happened to missing finger details.and even extra limbs and deformed ones (not always though) Can you tell me how I can overcome that problem? Maybe I am missing some node / lora or something in my workflow ?

    charlyOct 29, 2025· 3 reactions
    CivitAI

    I've been playing with this model about a week now. The level of detail it can create it's something i didn´t see on a XL model before. Also I find it very creative and with good prompt understanding. Good lights, textures... I don't know how you achieved that, but good work, it can be the best XL model so far, at least for my taste. Many thanks!

    Sateluco
    Author
    Oct 29, 2025

    many thaks for your comments, i love to see your example images

    charlyOct 29, 2025

    I can add that I prefeer the v1, don't know why the complaints. I'll upload some images ;)

    steaktreeDec 7, 2025· 3 reactions
    CivitAI

    This model is truly incredible. Have tested many SDXL models, and this one has a very natural, lifelike diffusion.

    Sateluco
    Author
    Dec 7, 2025

    Ty, i really like to see your creations

    Checkpoint
    SDXL 1.0

    Details

    Downloads
    984
    Platform
    CivitAI
    Platform Status
    Available
    Created
    10/17/2025
    Updated
    6/11/2026
    Deleted
    -

    Files

    xlSAT4stepsImitationOf_xlSATDMDIorv2.safetensors