CivArchive
    Preview 135047454
    Preview 135047280
    Preview 135047356
    Preview 135048958
    Preview 135055826
    Preview 135070116
    Preview 135070185
    Preview 135070362
    Preview 135070819
    Preview 135081449
    Preview 135081589
    Preview 135160293
    Preview 135163342
    Preview 135254925
    Preview 135286915
    Preview 135287041
    Preview 135287081
    Preview 135287145
    Preview 135326902
    Preview 135326925

    --Krea2
    Actually i tested with different combinations raw and turbo. For best quality i use raw with cfg4, for fast result i use raw, but witth cfg1 and turbo lora (it really gives better quality then kreo2 turbo model). U can use with other loras freely, also works good with nicegirls lora and nsfw loras. recommended weight 0.8 for raw and 1.2-2 for turbo

    --Ideogram4
    Works good with realism engine lora for nsfw images.
    Also updated my repo for better lora control https://github.com/DanrisiUA/ComfyUI-LoRA-Block-Filter
    One more. Made cool nodes with cursor (i have no clue how some of them to use) for ideogram https://github.com/DanrisiUA/ComfyUI-Ideogram4-Enhanced.git

    --Anima1.0

    Trained on top of my last ultrareal_finetune_base1 checkpoint for a slight realism boost. I also "vibecoded" a few experimental nodes for Anima (pretty sure something similar already exists, but here we are): https://github.com/DanrisiUA/ComfyUI-AnimaDynamicCFG

    Consider this a temporary realism upgrade until I train the next major fine-tune.

    --Anima0.5
    Trained on anima3. Recommended strength: 0.6.

    Works best with block filtering — I created a ComfyUI nodepack for this: ComfyUI-LoRA-Block-Filter. Examples generated alongside my other LoRA — nicegirls.
    Experimental version. Results may vary depending on prompt and block weight settings.


    --Flux2
    I use it with res2m+beta57, guidance 4, and it works very well. The quality is similar to my Qwen Lenovo, but I noticed that Flux2 is more flexible and has better prompt adherence.
    Recommend to use with strength 0.85-0.9

    -- Z-image Turbo
    Actually i'm not sure that my workflow is correct, i believe the result could be better
    Recommend to use with strength 0.80-0.9
    I used euler + sgm_uniform with 4 guidance
    P.S: at this moment for ComfyUI u need this pack to load lora properly https://github.com/PGCRT/CRT-Nodes

    --Chroma
    My first try to train lora that enhance realism in Chroma.
    I'll upload my workflow today a lil bit later (or atm u can just take it from images) cause there some nuances about settings. But from my observations i noticed that for fast gen dpmpp2m + beta with cfg 3 works good . More experimental settigns are res_2s_rkmk2e + beta57 and cfg 3, more accurate gens, better limbs and fingers, but gives a bit weird grain on very small details + plus this method is pretty long

    --Qwen
    Works perfectly and much better then other versions. Gives extremely "live" images with amateurish quality with dramatic light and shadows, some blur and with vibe of 2000s street photography
    You can take workflow here https://huggingface.co/Danrisi/Lenovo_Qwen/resolve/main/Qwen_danrisi.json
    Thanx to fox23vang226. i understood that it's possible (and even have better effect) to use lora with values 1.5 - 2.0.
    P.S.: almost all example images generate with both loras (NiceGirls)

    --Wan
    This is the first release — mainly a test to explore Wan's capabilities.
    Note: In this initial version, facial details are somewhat lacking. For better results, I recommend using it alongside the Instagirls LoRA by my friend, which significantly enhances facial features.

    --Flux
    Who needs a fancy name when the shadows and highlights do all the talking? This experimental LoRA is the scrappy cousin of my Samsung one—same punchy light-and-shadow mojo, but trained on a chaotic mix of pics from my ancient phones (so no Samsung for now)

    Description

    FAQ

    Comments (28)

    myneucontaJun 27, 2026· 1 reaction
    CivitAI

    PEAK!!

    heavy_battle_archJun 27, 2026· 4 reactions
    CivitAI

    Yasss! The last piece of the puzzle is here👏

    larrylangJun 27, 2026· 1 reaction
    CivitAI

    Whats this turbo lora than you are referring to?

    Danrisi
    Author
    Jun 28, 2026

    i remember someone post on redditt with link to HF. i'll try to search

    Agent_SmthJun 28, 2026

    i was just wondering too, can't find it (except one guy posting turbo lora 256 here on civit but i don't trust)

    Danrisi
    Author
    Jun 28, 2026

    @Agent_Smth krea2_turbo_lora_rank_64_bf16.safetensors

    this one. strange that i can't find it on HF by name

    BlackpepperflakeJun 28, 2026· 2 reactions

    https://huggingface.co/Comfy-Org/Krea-2/blob/main/loras/krea2_turbo_lora_rank_64_bf16.safetensors

    Danrisi
    Author
    Jun 28, 2026

    @Blackpepperflake ty kind man

    sissdfgaJun 28, 2026
    CivitAI

    I want to know how this lora's ability to draw Asians, Or does the author have any intention of optimizing it

    Danrisi
    Author
    Jun 28, 2026· 3 reactions

    if u talk about lenovo, it just improve details, textures, shadows and light. what about my another nicegirls lora - yes, it makes everyone like from eastern europe

    sissdfgaJun 28, 2026· 1 reaction

    @Danrisi thank u

    diesel2834Jun 28, 2026· 2 reactions
    CivitAI

    Thank you Mr. Lenovo

    Danrisi
    Author
    Jun 28, 2026· 2 reactions

    Haha. Hope lenovo will pay me

    OrangeJuiceAlienJun 28, 2026· 3 reactions
    CivitAI

    can you share your training settings?

    LuxariaJun 28, 2026
    CivitAI

    @Danrisi can you link the abliterated text encoder and the model you used please?

    Danrisi
    Author
    Jun 29, 2026

    https://huggingface.co/ahmed22xa/Huihui-Qwen3-VL-4B-Instruct-abliterated-comfy/tree/main

    i cant find the same i used in older generations. but tthis one works the same

    TensaizJul 4, 2026· 1 reaction

    I don't think an abliterated text encoder makes any difference at all

    woofdoggoJul 5, 2026

    @Tensaiz Maybe not as a text encoder but it's also used as a prompt enhancer and it can refuse if prompt deemed too spicy. You can either disable prompt enhancer, use ablated as both text encoder and enhancer or use regular qwen3vl as text encoder and load ablated qwen separately for prompt enhancing.

    SD_STORMJul 1, 2026
    CivitAI

    Does ComfyUI-LoRA-Block-Filter support krea2?

    Danrisi
    Author
    Jul 1, 2026

    I didn't make for krea2, cause its working okay. But I can make

    iVeo3Jul 3, 2026· 2 reactions
    CivitAI

    Best lora for krea2. Prompt adherence is 100x better.

    OfflyfansJul 6, 2026· 12 reactions
    CivitAI

    Make no sense to use the raw model for inference... It is actually no recommend by Krea them self

    ObsidianDreamsJul 6, 2026· 2 reactions

    Raw with the turbo lora gives superior quality to Turbo by far. It also gives a much higher amount of variation between images.

    @ObsidianDreams this

    Danrisi
    Author
    Jul 7, 2026· 6 reactions

    i dont care what devs are not recommend, if it works - i'll use it, especially when raw gives extra detailed textures. The Ministry of Health also recommend against drinking and smoking, but still

    jnkmusique426Jul 15, 2026

    @ObsidianDreams what szettings are you using for this so ? all my tests were trash with raw

    PatuwaJul 8, 2026
    CivitAI

    Very nice LoRa for Krea2 (i use turbo one), i tested with filter LoRa and works great. Another thing is that doesn't affect ethnicity but improves lighting and skin texture.

    Edit: Tested against other 4 Realism LoRas and most of them changes too much all the output, this one not, so it's better.

    Layla_FactorJul 12, 2026
    CivitAI

    Almost Real