Variant 6
I'm currently only distributing the optimized inference formats (INT8/ConvRot version) to keep performance high and protect the core workflow. The BF16 master weights won't be released publicly for this version. Thanks for understanding!
The V6 isn't a major overhaul, it's just a series of adjustments to the V5.
V5 available in bf16 & Int8 ConvRot
More face details
For those wanting even more detail on close-up portraits and tight face framing,
using FaceDetailer (Ultralytics face and eye bbox) is, of course, essential.
My settings
For achieving natural-looking skin renders with V4.
first pass : 10 steps / rk6_7s / linear quadratic / denoise 1
second pass : 4 steps / abnorsett_4m / bong_tangent / denoise 0.2
My advice, use:
V2 & V3 for graphic design, illustration, and painting
V1, V4 & V5 for photorealism
V6 for everything
You can also use Kreation for drawing, graphic design, and painting.
V4 (bf16 & Int8 ConvRot)
There's no point in chasing two hares at once, Variant 4 is oriented exclusively towards the photorealism, The pictorial aspect will be considered exclusively in subsequent versions..
V4 has been completely uncensored, if any censorship persists, it stems from the Qwen3VL text encoder, though this remains very rare.
The colors have been kept to a very neutral and natural palette.
GGUF Q8 & Q4 https://civarchive.com/models/2774280/kreamania-v4-gguf?modelVersionId=3123839
V3 (bf16 & INT8 ConvRot) :
Improved skin texture for greater realism
Optimized lighting effects, more subtle and with better contrast
Enhanced painterly visual style
Corrected color shift toward warm tones, color rendering is more neutral and realistic
Feedback on V3
V3 is a lot more restrictive
Yes, the "unlocked" feel has been diluted; I’ll need to revisit that in future versions.
Colors are less realistic.
Yes, that aspect still needs fine-tuning in subsequent versions.
Less realistic than V1
That’s possible, given that I’m aiming to finalize a versatile model, one that is realistic for photography yet also graphic in style.
KREA 2 TURBO Vs KREAMANIA
V2 (bf16 & fp8) : Improvement of version 1 by merging an unpublished personal LoRAs to optimize lighting, textures, and graphic effects.
V1 (fp8 & INT8 ConvRot) : The removal of certain restrictions compared to the model turbo fp8 scaled, along with improved realism and nudity rendering. The goal wasn't to focus on an NSFW model (even though it is capable of that), but rather to unlock the full range of creative possibilities.
Recommended ComfyUI extensions:
https://github.com/ethanfel/ComfyUI-Krea2TextEncoder
Description
Version features :
A new fine-tuning approach quite different from that adopted for previous versions.
covers both realism and graphic style
uncensored model
I do not recommend using Krea2T Enhancer Advanced with Kreamania V5.
Trigger : Bradhamel art style
You can use this trigger if you wish, but it is not mandatory.
FAQ
Comments (16)
My experience: Kreamania v1 is good, but each new version is worse and worse. Version 5 loses realism and adds overexposed and burnt images. Brighter images do not mean better realism. But that's just my experience.
That’s your point of view, our expectations differ. I was aiming for a specific look, and I achieved it. As for the slightly accentuated lighting, it’s a matter of taste, and while it doesn’t bother me, it certainly doesn’t amount to overexposure.
As for the realism of V5, I don't think I've yet encountered a Krea model that matches it, which leads me to believe that we aren't managing to fully leverage ComfyUI's capabilities to achieve that level.
You should view the different versions as variations rather than as an evolution of the same version. Thanks for the feedback anyway, even if I find it a bit excessive and narrow in focus.
@Adel_AI I apologize if my post is too strict. If I take all versions as variations, then yes, everything is fine and I am most satisfied with V1. Anyway, thank you for your work.
@DelightofVision Without going into too much detail about the direction I’ve taken across the various versions, since you mentioned liking V1, here is what led me to explore other avenues:
V1 is essentially just the base model unlocked via a merge with one of my LoRAs. I don’t like that version at all; the lighting is handled very poorly, the image looks completely flat, the color palettes are skewed in a way that feels unrealistic and distinctly "AI-generated," and, above all, it lacks any real graphic or illustrative quality.
So, as you can see, we started out with different visions and expectations. Still, it’s great that at least one version caught your eye.
Thanks for the feedback.
Are the random boils from body gone in this new version? :P
I replaced the boils with pustules, for a change.
Hey, could you please create an int8 for Variant 2?
Love your models but how come the sizes in your images are listed as 512x512? are there steps in the workflow that aren't included in the metadata? Does this model require swarmui or something?
Don't pay attention to what's displayed on Civitai, it has become very difficult to find nodes in ComfyUI that output parameters correctly, in a format Civitai can read. I often tweak them myself. My renders are usually 768x1024, but with direct inference, and without any loss of quality, I can go up to 1440x1920.
Thank you for your feedback.
@Adel_AI I see. Thank you! Huge fan, I still use Pixel Alchemy a lot.
@martius72 Pixel Alchemy ... OMG🫡
Thank you for your works! Very good!
thanks for your feedback
Thank you very much for the beautiful model.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.