---
license: creativeml-openrail-m
library_name: diffusers
pipeline_tag: text-to-image
tags:
- stable-diffusion
- stable-diffusion-diffusers
- diffusers
- text-to-image
- sadxzero
- WarriorMama777

---

![image.png](https://cdn-uploads.huggingface.co/production/uploads/63239b8370edc53f51cd5d42/KI0lYzyzPePBLFWNr6gWV.png)
icon of cute little white duck, round frame, blue glow, purple background

# SXZ D.U.C.K. - Digital Units Creation Kit

This LORA has been merged with Luma 0.98VAE, I think it improves over it, don't believe me? Watch this!

![image.png](https://cdn-uploads.huggingface.co/production/uploads/63239b8370edc53f51cd5d42/4qslPnDVbNqIBH2-kRYM2.png)

# D.U.C.K. & D.U.C.K.LumaOrangeMix

D.U.C.K. and D.U.C.K.LumaOrangeMix models have also been provided on the repo. D.U.C.K. is a merge at 1.0 strenght, it's the version you'd want to mix to your models. 

D.U.C.K.LumaOrangeMix is... it's complicated... So you know to merge a LORA so that it's not as strong as D.U.C.K., you mix it at 1.0, then you mix it again with the original model at 0.75 so you get mostly D.U.C.K.? Well, what if instead of mixing it back with the original you do it with another model to add its flavor?

That's what I did, merging it back with OrangeRemix. In fact, that was the plan all along and the reason I created OrangeRemix on the first place! Alas, while D.U.C.K.LumaOrangeMix does outperform it at cute output, it's at the cost of detail, so DigitalUnitsCreationKit remains at the hosted inference API, here's a pic:

![image.png](https://cdn-uploads.huggingface.co/production/uploads/63239b8370edc53f51cd5d42/6qjsZufvtTYRyGLJ9NPh9.png)

If you like this content, support Luma and D.U.C.K.'s creator at https://boosty.to/sadxzero

Original links:

https://civitai.com/models/44726/sxz-duck-for-game-assets-concept

https://civitai.com/models/25831?modelVersionId=68200

https://huggingface.co/Yntec/OrangeRemix

https://huggingface.co/WarriorMama777/OrangeMixs/tree/main/Models