Please consider buying my two albums Sisters of the Wicked and Danger Ahead, my book Dracula and also please check out my YouTube channel. Civitai won't support creators, so any purchase is a real help. Thanks.
DRACULA https://www.amazon.com/dp/B0FM4DXYPL
Sisters of the Wicked https://amazon.com/music/player/albums/B0DSJY3ZHY?marketplaceId=ATVPDKIKX0DER&musicTerritory=US&ref=dm_sh_CdyeuxIBNBDOjZ1hpB1PPJxT0
Danger Ahead https://www.amazon.com/music/player/albums/B0DB2DV354
Music In The Machine https://www.youtube.com/@musicinthemachine/videos
Lora Creation: https://www.upwork.com/services/product/design-flux-lora-for-creation-of-ai-art-1909718528813429477?ref=project_share
For the De-Distilled Model, Distilled CFG Scale should be set to 0. Then use actual CFG. For me 3.5 to 8 seem good.
If you like the models, please follow. The page is getting crowded as I keep releasing my model in as many formats and options as I can.
V1.1 Hyper 8-Step is now available and works in ForgeUI as well as Comfy. Schnell V1 Much faster, getting good results with 1-4 steps. Quality is not quite on the level of the Dev model, but much faster. So far it seems text is more consistently good though. New GGUF version of the Schnell model up now. Thanks to pkmngotrnr for converting them for me. Much appreciated. So far works in Comfy only at least for me. DEVfp8V1.1GGUF is now up as well. V1 version has European Flux mixed in at what seemed like a pretty low percentage, but it seems like it is a bit too strong. If you like this look, feel free to use this one.
This is my initial FLUX model. It is base on the flux1-dev-fp8 model. This is a merge only in the sense that I merged in some Loras that I like to make it more realistic. As with all of my models, the goal is the best photo realism that I can achieve. If it will do other things well, that is a bonus. Until I can learn a lot more about FLUX, and see if at some point true merges are possible, this is a pretty good start. I don't know about compatibility with all software. I use ForgeUI and I am able to run it fine. Most gens take about a minute and a half for me at 1024 X 1328. I use Euler with the Beta scheduler or the Simple. I haven't experimented much yet with other samplers. I typically use 24 to 28 steps.
Description
NF4 version of the Acorn Is Spinning Flux V1.5 Hyper 8 Step. This version in my limited testing seems to work best with DPM2 Normal. I usually generate with distilled cfg at 0 and CFG 1. 8 steps usually works well, but sometimes if you are doing a lot of text it can help to up it to 10 or 12 steps. As before, it works well to use negative weighting with my Acorn Improved Nipples when doing images that you don't want nipples or nudity. I enter it like this: <lora:nipples:-0.99>
FAQ
Comments (10)
This was a good model in the past, now its just producing worse and worse images each iteration.
And it seems to be a trend with other models to, that new iterations is just creating worse images.
Sorry to hear that. It is still making great images for me. As things get added, you may have to learn to prompt it a little different. Of coarse you can still use the original model as well which should continue to produce exactly as it always has. The latest version changed nothing but adding in the improved nipples for those who like making nudes. The model I use the most is still my Acorn Is Spinning Flux V1.1 Hyper 8 Step.
This very latest NF4 version was requested by those whose machines need a model that uses less memory. It doesn't produce as good of images, but it is better than if you can't produce images at all. The newer versions aren't meant to replace the others, but just give more people options to meet their needs. If someone requests a certain type, be it GGUF or NF4, I like to provide it if I can.
Your dev 1.1 has been the nicest in terms of Lora compatibility for me so far, any change we will see a non hyper updated version ?
I do have the normal version of V1.5 that I haven't yet uploaded. I will try to get it put up soon.
@Seeker70 awesome ! i on forge so that might be some of it but I can never get as detailed skin and images with hyper.
@Gentlemanbunny I use Forge also. I don't really notice much difference in the outputs between the regular and the hyper version. Maybe slight differences. For me, the main difference is hyper get very close to the same quality in a quarter to a third of the time.
@Seeker70 yeah no idea, not saying its the model. Just for me sampler/ settings I tried changing everything I get a very like "overbaked" skin look or sub sd.15 quality xD
@Gentlemanbunny The full version is up now. Enjoy.
Not whining but this latest 6gb flux model is nf4 right and I've never used nf4 but I do know it should be loaded into a checkpoint loader, however when loading it with a checkpoint loader, I get this "ERROR: Could not detect model type of: C:\Users\intel\Downloads\ComfyUI_windows_portable\ComfyUI\models/checkpoints/acornIsSpinningFLUX_aisfHpr8stpv15NF4.safetensors"
I tried loading it with a unet loader but that didn't work either.
If I remember correctly, Comfy has an NF4 specific separate loader. I haven't used this new checkpoint in Comfy but I did use the V1.1 NF4 in Comfy so I'm sure it has a compatible loader. I think if you search for NF4 it comes up.