Nunchaku Fluxmania Legacy (by Spooknik)
Select Pruned model fp16 (6.3 Gb)
KREAMANIA
I don't consider Kreamania to be a truly new model, it's just a slight improvement on Legacy, injecting a tiny bit of Krea (which, in my opinion, isn't a benchmark for realism) and adding some of my recent LoRA models. It deserved the name Legacy+ more than Kreamania.
I find this version provides more detail and improved lighting. The composition remains almost identical to that of Legacy.
Note : Don't use "freckles" in the prompt
Oddity : Unexpected appearance of text
https://civarchive.com/models/1808412/wan-refine-fluxmania
https://civarchive.com/articles/17346/wan-refining-fluxmania-for-more-realism
Quantized versions of the Fluxmania Legacy model:
https://huggingface.co/belisarius/FLUX.1-dev-Fluxmania-Legacy-gguf/tree/main
Showcase images are in PNG and include metadata, prompt & workflow.
All licensing conditions for the Flux Dev model apply. You can find the details here.
Model orientation : Photography - Realism - Portraits - Artistic nude
Fluxmania III :
File 22.07 GB : unet fp16 without VAE & clip
File 11.16 GB : unet fp8 without VAE & clip
File 15.91 GB : checkpoint fp8 (not bf16) with VAE & Clip L full fp32
Settings : dpmpp_2m - sgm_uniform / cfg 3.5 / steps 25 - 30.
I recommend using this version of Clip L with the Fluxmania model
Fluxmania I & II : Unet fp8 (no VAE, no clip) / Settings : dpmpp_2m - sgm_uniform / cfg 3.5 / steps 25.
Genuine : Unet fp16 (no VAE, no clip) / Settings : Euler Simple / cfg 3.5 / steps 30.
Description
Fluxamnia v1 merged with
Improved light, Improved background and more details. Less anatomical aberrations.
Gives good results with the LoRA Fluxartis Photography
FAQ
Comments (26)
Thanks for the update! Fluxmania is one of my favourite Flux Dev checkpoints. Can you kindly explain in a few words what's new in v2? I'm eager to learn. :)
I tried to improve the rendering of the hands, make the backgrounds more interesting and get better light rendering, hence my choice of models with which I mixed Fluxmania v1. I didn't have time to do a lot of testing, so your feedback will be useful.
The settings remain the same as V1. I noticed that associating the LoRA Fluxartis Photography enhanced the renderings more
@Adel_AI Oh, nice! Thanks!
Your models look great !
Can you add a GGUF version ?
I honestly don't think I'll do it for the moment, due to lack of time and especially because I plan to develop this model into other versions. I will probably decline it in several versions, notably GGUF once I consider that it is finished.
If you decide to add GGUF versions and you have python in your system, you can also use this GUI to convert.
https://github.com/city96/ComfyUI-GGUF/discussions/145
It seems to be easier to use, but I haven't tried, because I only have python in my portable ComfyUI install.
@jr81 I use this GUI more for my conversions, I find it practical: https://github.com/kevkid/gguf_gui
please add fp16 for your Fluxmania version 2 please
I can't do the merge in fp16, I'm out of memory. I know it's possible to convert a fp8 to fp16 with a script, but sorry I don't know how to use this kind of script
No you can't you have to merge from fp16 and cast it down to fp8 if you want fp8, it's the correct way. If you can give me your merge method maybe I can do for you.
@Wardensc the 3 models that make up the merge exist only in fp8 version, so it was impossible to introduce the fp16 format. When trying to combine Fluxmania V1 with Flux 1.D fp16, my RAM & VRAM crash. If you want to try it, no worries, the fusion ingredients are listed in the version comments
@Adel_AI i really want fp16~~ may be can fix it later
@gx_ground136 I will try
@Adel_AI thanks guy~~
Can do dark scenes✅
No excessive headshots✅
Fast generation✅
the best flux model, gives realism and lighting a real boost
Merci l'ami
Thanks, will you upload a fp16 unet version?
for V2, there will be no flp16, I explained the reasons. It is possible for future versions
@Adel_AI Ok, thank you for the reply.
Fluxmania V2 is this already include vae , clip_l, t5xxl_fp8_e4m3fn ?
It is clearly stated in the name of the model that it is a unet and not a checkpoint, so neither vae nor clip
cool
What the specilality of this model?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.