Zankuro - LyCORIS LoCon - PonyDiffusionXLV6版本7.2 - 000052 (ID: 434656)

LoCon trained on Zankuro's artstyle for ponydiff: https://civitai.com/models/257749

Artist: https://fantia.jp/fanclubs/4384

LoRA Version: https://civitai.com/models/276071

Because of my lack of experience and time, the version number may be based on changes that have made certain aspects worse. As such a previous version may perform better than a later version.

I'm still learning the ins and outs of the model and LoRA training in general, please leave reproducible feedback or tips so that I may improve over time.

Check the "About this version" section for any quirks or pitfalls I'm aware of or have experienced.

Some Q's:

1. What's your workflow?

   • Sampler from images is done with Restarts with segments of [3,2,0.06,0.30], which are values from the paper on Restarts

   • Initial resolution is typically 832 x 1248 or vice-versa and upscaled by x(20/13), I try to keep the pixel count around 1024 x 1024 for initial gens and adjust the upscale ratio as needed

   • Upscaler is usually 4x_fatal_Anime_500000_G or 4x_NMKD-Siax_200k

   • Second sample pass is the same as the first (params are on civit's meta interface) but at ~.35 denoise

   • A facedetailer is used at .5 strength

   • I use ComfyUI, sorry

2. What strength should I be using?

   • .8 - 1 on base pony

   • 1 on merges

3. Pairs of images?

   • Left image is AutismMix, Right image is base Pony

4. Why not put this as a version on the existing model?

   • As far as I'm aware civit does not allow me to independently tag standard LoRA versions and LyCORIS versions on the same model page, so I've separated them to reflect this.

5. Zankuro was discovered to be in base pony, your efforts are for nothing and you are a fool.

   • 1. I already had the dataset before this information

   • 2. I think base pony looks like poop, and don't like the inbuilt version

   • 3. I think I've improved quite a bit and wanted to see if I could make this better

   • This doesn't perform like the inbuilt version, it leans more towards flat/saturated, if you don't like it use the tag instead

6. What does your filename structure mean?

   • It's split into 4 parts of [XX]-MAnon-[YYYY]-[ZZZZZZ]

     • [XX] - Artist initials

     • MAnon - My moniker

     • [YYYY] - SD version and Model version (Ex. XLV4.1 is on SDXL [Not 1.5] and it's version 4.1 of the model)

     • [ZZZZZZ] - The epochs, only useful to me

描述:

Version 7

Proclivity for shortstacks/big heads, remedy this with "short stack" in negatives or "mature [sex]" in the positives

plump

Will rarely ignore background colors, might just be a base pony problem though

On base pony it is much more saturated with thick lines, drop the strength if this becomes a problem

Artist signature was pruned, Yay! Hooray!

训练词语: