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 Fine-tuned for Animals/ Creatures and Anthro and Furries but can do way more and can be used as an general use model in general use:This model can do:<ul><li>Photorealistic</li><li>Digital Art</li><li>Anime</li><li>Comic</li><li>Vector Graphics</li><li>Line art</li><li>Stickers</li><li>can write short text</li><li>etc</li></ul>The model can do SFW and NSFWThis is the <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen-Image">Qwen Image</a> Version of my <a target="_blank" rel="ugc" href="https://civarchive.com/models/192854/ratatoskr-animal-creature-and-furry">Ratatoskr</a> Model Seriesfor more information about Owen Image please use the link above.The model its trained on my own ratatoskr database and parts of the furry enhancer database . This its trained on my Ratatoskr Database and the Furry Enhancer Database so no real pictures or celeb people pictures are used for training.for settings:I use personal this setting:Sampler: EulerScheduler: linear_quadratic/ beta57Steps between 25 and 35CFG from 3-4Simple Workflows are in the images but i also will provide some in the Training folder.i also will make some new workflows public in the future (WIP)Changelog:<h3 id="v1.0-first-version">V1.0 First version</h3><h3 id="v1.8-(-edit-model)">V1.8 ( Edit Model)</h3>It's alive ........This <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen-Image-Edit-2509">Qwen Edit 2509</a> Model version is an "Frankenstein Model" it is made by replace entire layers of the original model with Ratatoskr Qwen Image V1 ones. layer by layer so there is only a skeleton left of the original one i also have added a small lora on it too. (6800 steps on 110 paired images)this model perform better with anthro and furry images than the original one.Like the name suggests this model is not designed to create images only edit ones.can do sfw and nsfw images.do to (maybe) destroyed layers this model runs very fast at 4 steps to edit an imagei highly recommend using the workflow that i have attached to this modelthe settings are:4 stepseulercfg1.3beta57 or bong_tangendImportant this provided workflow use 2 custom comfy nodes made by me that only can be found attached in the training folder<h3 id="v2.99-qwen-image">V2.99 Qwen Image</h3>For this version i used the same Database but added some new pictures that got generated with V1 and use more pictures of the ratatoskr database in general and changed the training method slightly with V1 because its now trained on my own system even that did take some time longer to save costs but the training took about 3 weeks....I trained on the V1 Model this time for Natural Language and e621 so it now "understand" prompts better. I also tried to get rid of the female bias a bit.overall i added around 90k steps in training<ul><li>more detailed output</li><li>better prompt following</li><li>reduced the female bias</li><li>better with nsfw stuff in general</li><li>better colors</li><li>more realistic but also keeping its drawing abilities</li></ul>for settings this model version didn't changes much from V1Sampler: EulerScheduler: linear_quadratic or betaSteps between 25 and 35CFG from 3-4I highly recommend using the advanced workflow that is in the training foöder to get rid of the qwen grids by using z-image for upscaling.if you like to support my work please give a thumbs up i also interested what people can do with that model i also can highly recommend using it together with the <a target="_blank" rel="ugc" href="https://civarchive.com/models/310964/furry-enhancer">furry enhancer</a><h3 id="v3.3">V3.3</h3>while 2.99 was the new "raw" trained model and got plenty of new things implemented (include bugs)this versions changelog is relative short. Its more or less a bugfix of V2.99. So what did changed?The motto was "get rid of the grid"<ul><li>reduce the strength of the "Grid"</li><li>sharper images</li></ul>Note that on qwen images there is always this grid so the only way to remove it completely is running the model with the refiner like before.for settings this model version didn't change's much from V2.99Sampler: EulerScheduler: linear_quadratic or betaSteps between 25 and 35CFG from 3-4Workflow and examples are in the training folder.<h3 id="v3.51-(qwen-edit-image)">V3.51 (Qwen Edit Image)</h3>This version of Ratatoskr Qwen Edit is the first fully trained Version of it.Its trained on the newest Qwen image edit 2511 mixed precision by <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI/tree/main/split_files/diffusion_models">comfy-org</a> . This model is trained so far at 638 paired images that all got all generated by the Ratatoskr Models so its not trained on any real images or people ;-) so don't complain about that .....Note that i also implement some split blocks of the <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v/Qwen-Image-Edit-2511-Lightning/tree/main/qwen_image_edit_2511_fp8_e4m3fn_scaled_lightning_split">qwen lightning lora</a> by <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v">lightx2v</a>Because no one wanna a stupid slow model that needs 40 steps to function correctly ;-)Settings of this model:so because of the split blocks partly implemented the model has a range between4 - 16 steps between 1cfg and 2.5 cfgif it should stick to the image my personal setting is:cfg1 and 7 stepsfor creative edit or high quality cfg2.5 and 16 steps.i will upload an showcase with some examples too<h3 id="v4-qwen-image-(based-on-2512)">V4 Qwen Image (based on 2512)</h3>This is the first Ratatoskr model based on the new <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen-Image-2512">Qwen Image Model V2512</a> so for technical details look there ;-)This model is a Block Merged model Between My Fine-tuned Qwen Image V3.3but its way more complex that it sounds ;-)this Version will also used by me for further updates.Changes:<ul><li>Way more realistic</li><li>but still saved its drawing ability</li><li>i got rid of the grid now completely ;-)</li><li>a bit less horny like but this is a good thing in my opinion ;-)</li></ul>for settings this model version didn't change's much from V2.99 onwardsSampler: EulerScheduler: linear_quadratic or betaSteps between 25 and 35CFG from 3-4in general the model gets more and more creative with higher step and cfg count.so it really depends on you what settings are better for you.in all images is the workflow but you will find it in the training folder too with some example images.so have fun with edit your images but keep in mind you are responsible what you do with that edit model don't blame others.IMPORTANT:Training such models is very time and cost intense so please help me by.<h3 id="please-checking-out-my-other-models-too-and-give-like-)-when-downloading-this-model!">Please checking out my other models too and give like ;-) when downloading this model!</h3>Disclaimer:This disclaimer is not here without reasons. please have in mind this is like a virtual showroom and posting extreme stuff is like 💩 on my desk in real live 😉 and you are responsible for your own images that you generate keep that in mind.made by <a target="_blank" rel="ugc" href="https://civarchive.com/user/freek22/models">Freek22</a>

Ratatoskr (Qwen Image 2512 / Edit 2511)

This version of Ratatoskr Qwen Edit is the first fully trained Version of it.Its trained on the newest Qwen image edit 2511 mixed precision by <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI/tree/main/split_files/diffusion_models">comfy-org</a> . This model is trained so far at 638 paired images that all got all generated by the Ratatoskr Models so its not trained on any real images or people ;-) so don't complain about that .....Note that i also implement some split blocks of the <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v/Qwen-Image-Edit-2511-Lightning/tree/main/qwen_image_edit_2511_fp8_e4m3fn_scaled_lightning_split">qwen lightning lora</a> by <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v">lightx2v</a>Because no one wanna a stupid slow model that needs 40 steps to function correctly ;-)Settings of this model:so because of the split blocks partly implemented the model has a range between4 - 16 steps between 1cfg and 2.5 cfg if it should stick to the image my personal setting is: cfg1 and 7stepsfor creative edit or high quality cfg2.5 and 16 steps.i will load up a showcase for some examples tooin general the model gets more and more creative with higher step and cfg count.so it really depends on you what settings are better for you.in all images is the workflow but you will find it in the training folder too with some example images.so have fun with edit your images but keep in mind you are responsible what you do with that edit model don't blame others.Note: Such models need a crazy amount of time to make so please support this model with a like and look at my other stuff too.<a target="_blank" rel="ugc" href="https://civitai.com/user/freek22">made by freek22</a>

ratatoskrQwenImage2512_qwenImageEditV351.safetensors

ratatoskrQwenImage2512_qwenImageEditV351_trainingData.zip

Qwen Image Edit V3.51

Qwen Image (2512) V5.4

Qwen Image (2512) V4.99

Qwen Image (2512) V4

V3.3 Qwen Image

v2.99 Qwen Image

v1.8  Qwen Edit Hybrid

v1.0 Qwen Image

This version of Ratatoskr Qwen Edit is the first fully trained Version of it.Its trained on the newest qwen image edit 2511 mixed precision by <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI/tree/main/split_files/diffusion_models">comfy-org</a> . This model is trained so far at 638 paired images that all got all generated by the Ratatoskr Models so its not trained on any real images or people ;-) so don't complain about that .....Note that i also implement some split blocks of the <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v/Qwen-Image-Edit-2511-Lightning/tree/main/qwen_image_edit_2511_fp8_e4m3fn_scaled_lightning_split">qwen lightning lora</a> by <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v">lightx2v</a>Because no one wanna a stupid slow model that needs 40 steps to function correctly ;-)for setting of this model:so because of the split blocks partly implemented the model has a range between4 - 16 steps between 1cfg and 2.5 cfg i will load up a showcase for some examples tooin general the model gets more and more creative with higher step and cfg count.so it really depends on you what settings are better for you.in all images is the workflow but you will find it in the training folder too with some example images.so have fun with edit your images but keep in mind you are responsible what you do with that edit model don't blame others.Note: Such models need a crazy amount of time to make so please support this model with a like and look at my other stuff too. for download visit my main page: <a target="_blank" rel="noopener noreferrer nofollow" href="https://civitai.com/models/2147697/ratatoskr-qwen-image-version">https://civitai.com/models/2147697/ratatoskr-qwen-image-version</a><a target="_blank" rel="ugc" href="https://civitai.com/user/freek22">made by freek22</a>

Ratatoskr Qwen Image Edit V3.51

second version of this model 
4 - 16 steps between 1cfg and 2.5 cfg i will load up a showcase for some examples too

in general the model gets more and more creative with higher step and cfg count.

so it really depends on you what settings are better for you.
for download please go on my civit page  https://civitai.com/models/2147697?modelVersionId=2549343

Files with SHA256: 5012b162e663152eea2ab016b5c550f20588220d8360dc3f342cbf5f6c44dc6e