What's in it?
I've taken my Ally's Mix II: Churned, cut in a dash of my favourite model, ReV Animated, then taken the result of that + Noise Offset - SD 1.5.
STOP AND READ, LoRA FANS!
This is mixed with the OG Noise Offset model by Nicholas Guttenberg at Crosslabs
You cannot - CANNOT - extract a LoRA from this model and get similar results to the model. The noise offset does not extract. You have to train the noise offset in, or cut it in with a difference merge.
If you extract this model to LoRA, it will look like shit, and kids will point and laugh at you in the street.
Recreating my sample images?!
Some of my sample images are created with ControlNet, to get poses and compositions from my previous generations - you'll never be able to reproduce those. Others, it's possible, but some do use Patron exclusive embeddings. The sample images are meant to show what's possible with a model.
Description
FAQ
Comments (34)
Your skill with models never ceases to amaze me
Beautiful looking model! I'm downloading now!
Looking forward to play with it on Monday ❤️
She did lora makers dirty kekw
what upscaler do you use? everytime i use a upscaler there comes an extra leg or something , it will be deformed
Does the Noise Offset also effect mixing if it effects Lora extraction?
Yep, it transfers into new mixes, but I'm not sure how much the effect is diluted by mixing.
@theally have you thought about releasing an offset-stripped, unpruned version? giving the ability to add the offset as diff at the end as to not distort it when average mixing other models
@adobi that's a good call. I have uploaded the noise offset file created by the Crosslabs researchers (https://civitai.com/models/10391/noise-offset-for-true-darkness-in-sd), so one could diff it with 1.5, but there's definitely room for development here. The other option is to add the noise offset while training a LoRA (or fine-tuning), to add the effect.
Can this great blend be reduced to a 2gb pruned fp16 safetensor version ? Waiting for your promise: "I am making a LoRA from this model, the correct way, ty".
This looks sexy in all the right aesthetic ways from the previews and reviews. Great work Ally! Excited for this one.
Thanks for crafting great prompts! I have tested one with Babes and I had to make it into a cover image.
My question is...how do you control the darkness levels? If I've used one of your prompts, but edited it a bit, and I'd like to lighten it up....any suggestions?
I'm begging the author of this beautiful model for an inpainting version
If you're using a1111 you could make one. SD 1.5 inpainting model in box A, this model in box B, SD 1.5 in box 3 then merge Add Difference with a weight of 1.
I use this for inpainting in automatic1111, it does really great facial expressions
@randallaware And what about this?: https://i.imgur.com/mOl72AQ.png
@zerocool22 I've always left it default, worked perfectly for me.
Will you upload this model on hugginface with diffusers?
PLeaseeee diffusers versions for Dreambooth
seriously one of the best models i have seen to date. your work provides spot-on quality content by controlling the lighting (or heh the darking). thank you for your contribs!
Need VAE?
Use whichever you want. Most of these generations in my samples didn't use any.
Do you consider making a version to include abyss orange mix 3 (AOM3), for nsfw and nsfw hard support ? Your model is the best , but when trying to do nsfw hard , it does not do a good job (it fuses body parts).
I'm testing some merges! I'm tempted to do an adult/mature content merge!
@theally i'm testing too, but using supermerger is hard for someone who is merging for the first time. i wanted to do a model with the style of your model or rev animated but with the accuracy of aom3 when generating hardcore nsfw.
my favorite model, the only one.
Note: ReV Animated requires that any merges using it share the same permissions as ReV does, which is less restrictive than indicated on this model.
I'm loving this model!!
" If you extract this model to LoRA, it will look like shit, and kids will point and laugh at you in the street. "
- Some legendary guy giving the best line about merging.
Details
Files
theAllysMixIII_v10.safetensors
Mirrors
10752_theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
theAllysMixIII_v10.safetensors
















