File size: 3,661 Bytes
e3a52bd 1ff15f1 e3a52bd 1ff15f1 270f3d9 1ff15f1 344abcf 1ff15f1 344abcf 1ff15f1 e34d392 1ff15f1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 |
---
license: creativeml-openrail-m
tags:
- coreml
- stable-diffusion
- text-to-image
- not-for-all-audiences
---
# Core ML Converted SDXL Model:
- This model was converted to [Core ML for use on Apple Silicon devices](https://github.com/apple/ml-stable-diffusion). Conversion instructions can be found [here](https://github.com/godly-devotion/MochiDiffusion/wiki/How-to-convert-ckpt-or-safetensors-files-to-Core-ML).
- Provide the model to an app such as **Mochi Diffusion** ([Github](https://github.com/godly-devotion/MochiDiffusion) / [Discord](https://discord.gg/x2kartzxGv)) to generate images.
- `original` version is only compatible with `CPU & GPU` option.
- Resolution is the `SDXL` default of `1024x1024` UNLESS NOTED OTHERWISE.
- This model requires macOS 14.0 or later to run properly.
- This model is quantized to `8-bits`.
- This model was converted with a `vae-encoder` for use with `image2image`, but it may be a while before support for SDXL i2i is added to Core ML and Mochi..
- Descriptions are posted more or less as-is from original model source.
- Not all features and/or results may be available in `CoreML` format.
- This model does not have the [unet split into chunks](https://github.com/apple/ml-stable-diffusion#-converting-models-to-core-ml).
- This model does not include a `safety checker` (for hiding NSFW content).
- This model can not be used with ControlNet.
<br>
# coreml-juggernautXL-v6_SDXL_8-bit:
Source(s): [CivitAI](https://civitai.com/models/133005/juggernaut-xl?modelVersionId=198530)<br>
## Juggernaut XL Version 6 + RunDiffusion
### This is an SDXL base model converted and quantized to 8-bits.<br><br>
## A bit about my new "partnership" which you've probably already noticed ;)
Last week, RunDiffusion approached me, mentioning they were working on a Photo Real Model and would appreciate my input. As we progressed, we compared Juggernaut V6 and the RunDiffusion XL Photo Model, realizing that both models had their pros and cons. Both I and RunDiffusion thought it would be nice to see a merge of the two. RD then entrusted me with their unpublished checkpoint, and I began working on a JuggernautXL/RunDiffusionXL Photo merge.
The result thrilled both me and RunDiffusion. After many discussions, we concluded that we'd like to collaborate more closely on projects in the future. RunDiffusion offered to publish this merge and also offered me their resources for training.
The total Step Count for Juggernaut is now at 1.37 Million Steps. Don´t forget that this Number is for the Base and all the Sidesets Combined. So its obv not 1.37 Million Steps on 1 Set, that would be useless :D
## Finally, a few recommendations for the settings:
- Sampler: DPM++
- Steps: 30-40
- CFG: 3-9 (3 looks a bit more realistic, but has slightly fewer details)
And now, have fun trying it out. As always, I'm eagerly waiting for your pictures in the Hugging Face Gallery :)
### Last but not least, I'd like to thank a few people without whom Juggernaut XL probably wouldn't have come to fruition:
- ThinkDiffusion
- Dreamlook.AI (Trained 3 Side Sets)
- Chillpixel
- SilasAI6609 |