Loraマージ時のミスで予定外のLoraをマージしていたのでV2.1をリリースしました。V2.0は破棄してください。
V2.1 was released due to an error during the Lora merge that resulted in unscheduled Lora merges.Discard V2.0.
V2.1の詳細 / Description of V2.1 version
・私自身の好みの反映 / Reflection of my own preferences
・VAEの同梱 / Baked VAE
NSFWのテストはあまりしていないので、以前のバージョンより劣るかもしれません。
好みで使い分けてください。
相変わらず、目や手などの描写に問題が出ることがあるので、ADetailerなどで補正してください。
サンプル画像の大半は、ADetailerのEyes detection (Adetailer)で補正しています。
背景にこだわりすぎると、人が消えたり小さくなったりします。
Not much NSFW testing has been done, so it may be inferior to previous versions
Use according to your preference.
As always, problems with depiction of eyes, hands, etc. may occur, and should be corrected by ADetailer or other means.
Most of the sample images are corrected with ADetailer's Eyes detection (Adetailer).
If you focus too much on the background, people will disappear or become smaller.
Description of V1.0b version
Change Lora merge ratio
The purpose is to reduce the impact on the picture pattern and to reduce the failure rate of eyes, hands, etc.
Use ADetailer if necessary as the eyes tend to collapse
Use face_yolov8n,s
Check "Use separate noise multiplier" to reduce discomfort.
このバージョンについての説明
・Loraのマージ比率の変更
絵柄への影響度を下げて、目、手などの破綻率の低減が目的です
目が崩れやすいので、必要に応じてADetailerを利用してください
face_yolov8n,sを利用
Use separate noise multiplierにチェックをいれると違和感が減ります
Model Description
This model is a Pony model.
There may be more broken hands and feet because I gave priority to my favorite patterns.
For reference only
Steps: 35~45, Sampler: Restart, CFG scale: 6.5~7.5
ADatailer is recommended for full body
Model Merge Recipe
I used SuperMerger
ChacolPoLoMixXL_V1.0 (Not open to public)
ponypang_XL_test x (1-alpha) + Saint_LoriponyV3 x alpha (0.0,1.0,0.842423804012346,0.71508487654321,0.615234375,0.540123456790123,0.487003279320988,0.453125,0.435739776234568,0.432098765432099,0.5,0.523847415123457,0.544945987654321,0.560546875,0.567901234567901,0.564260223765432,0.546875,0.512996720679013,0.459876543209876,0.384765625,0)
ChacolEbaraMixXL_V1.0
ebara_pony_1 x (1-alpha) + ChacolPoLoMixXL_V1.0.fp16 x alpha (0.0,0.0,0.00427756931309475,0.0170370868554659,0.0380602337443566,0.0669872981077808,0.103323329854383,0.146446609406726,0.19561928549564,0.25,0.5,0.565263096110026,0.62940952255126,0.691341716182545,0.75,0.804380714504361,0.853553390593274,0.896676670145618,0.933012701892219,0.961939766255643,0)
ChacolEbaraMixXL_V2.0
ChacolEbaraMixXL_V1.0.fp16 x (1-alpha) + escapeXLPony_v01LightningBeta x alpha (0.2,1.0,0.984809027777778,0.940972222222222,0.87109375,0.777777777777778,0.663628472222222,0.53125,0.383246527777778,0.222222222222222,0.5,0.687065972222222,0.871527777777778,0.94921875,0.777777777777778,0.616753472222222,0.46875,0.336371527777777,0.222222222222222,0.12890625,0)
ChacolEbaraMixXL_V1.0a,b,2.0
Lora Merge: I learned My Lora(Not open to public)
use model
ponypang_XL_test
Saint_LoriponyV3 (Now undisclosed)
ebara_pony_1
About License
If you utilize a model merged with ChacolEbaraMixXL, you are encouraged to share the merge recipe.
check the license of each models