In Norse mythology, Ratatoskr (Old Norse, generally considered to mean "drill-tooth" or "bore-tooth") is a squirrel who runs up and down the world tree Yggdrasil to carry messages between the eagle perched atop Yggdrasil and to the roots of the tree to Nidhogg.
This model is for Animals / Creatures and Furries but can do way more :
Because it's an Flux Kontext model its main use
is image to image keep that in mind.
This model can do:
Photorealistic
Digital Art
Anime
Comic
Vector Graphics
Line art
Stickers
can write short text
furry
etc
This is the Flux Kontext variant of Ratatoskr Flux
The use case is for example for editing any picture made with the ratatoskr model series you can find in my profile but its working with any given pictures too.
Changelog:
V1: First Release Version
this is a block merged model made with the Original Kontext Model, and Parts of Ratatoskr Flux and Sleipnir Flux
Note: this is "Just" V1 and not directly trained on the Ratatoskr Dataset yet but i will use this block merge for the starting point for an actual finetune (V2) so there will for sure some Updates in the future.
V2:
This is the first trained Version that is using like i mentioned the V1 Block merged version
Its trained now at 175 paired images i created on the Ratatoskr Database. Because of the difference of captaining of kontext to my standard model database i did write all the caption by hand that alone took over 9 hours the training time itself did took some days...
the training helped a lot to "repair the damage" of the block merge with the standard flux kontext model
way more consistent
better in sfw and nsfw
the model is compatible also with my furry enhancer to some point tested with 0.7 strength
Made by Freek22
Please check out my other models too and give like to support my work ;-)
Description
GGUF Model Variant:
This is the first trained Version that is using like i mentioned the V1 Block merged version
Its trained now at 175 paired images i created on the Ratatoskr Database. Because of the difference of captaining of kontext to my standard model database i did write all the caption by hand that alone took over 9 hours the training time itself did took some days...
the training helped a lot to "repair the damage" of the block merge with the standard flux kontext model
way more consistent
better in sfw and nsfw
If you like to support me please give a like!
FAQ
Comments (6)
Running into issues with the workflows, rather than the model. I'm 100% sure I'm doing something wrong, but can't figure out what. Hoping someone has some ideas for me...
* If I load Ratatoskr (any version/base) and/or the Enhancer into A1111/Forge/reForge, it creates perfect images. No issues at all.
* If I use the Ratatoskr workflows on Comfy (Flux.1, SDXL/HEL, Krea) I get awful eyes. It's like someone put a blob of black ink over the eyes, then dribbled a dot of the eye-color on top of the black ink.
* I've tested with identical settings. Same model, loras, CFG, clip step, samplers, seed, etc. Issue persists anytime I use the official workflows. I've tested by removing the upscaling and loras, and the issue persists. It's definitely something my setup doesn't like about the workflow.
* Here's the weirdest part...if I use the same models/etc and generate the base image in A1111, then bring that to the Kontext workflow, it stays fine. It's only if the base image is created via the workflows that things go south.
Any idea what I'm missing here? Anyone else run into this?
The question is which workflow exactly never heard anything like this works 💯 fine I only know on sdxl models comfy changed overtime and it may conflict the used sampler there when the option a1111 is activated in this old workflow but you mentioned all workflows what is impossible at all because they work fine
@freek22 I may well be using the wrong workflow, but that's why I wanted to ask. I've tried with RatatoskFluxWorkflowVarB-GGUF and RatatoskrfluxV11.39Krea workflow. Both seemed to have the same weird eye issue. I then went back to RatatoskrNoobV1.2 workflow and had the same issue, but when I loaded the same model into Forge, I had no such issue. Not sure if there's something weird in my ComfyUI causing it, but figured I'd ask in case there was an easy fix.
@LemurT the problem when it in your comfy install it's hard figuring out what's is wrong
@freek22 I was afraid that'd be the answer, but had to try and see if anyone had ideas. Thanks anyway.
I'm getting black images on comfy or no changes at all
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.



















