v8

肌理更新：強化了以下標籤的學習：

Texture Update: 以下標籤在訓練中被強化：

realistic, photorealistic, flat color,

shiny skin, matte skin, shiny hair,

請注意，在 Danbooru 數據集中有多個標籤用於描述「照片」或「類照片風格」。我在訓練集中將這些圖片統一標記為「photorealistic」。但使用 Danbooru 訓練集訓練的 SDXL 模型大多無法很好地繪製寫實圖像，因此「photorealistic」建議僅在較低權重下使用來改變畫面肌理。「realistic」標籤則可在較高權重下正常運作。

Please note that Danbooru dataset contains multiple tags to describe "photo" or "photo-like styles". I’ve tagged all such images as “photorealistic” in dataset.

However, most SDXL models trained on the Danbooru dataset do not render realistic images well. “photorealistic” is only recommended at low weight, where it can help adjust texture rather than create realism images. The “realistic” tag can work properly at higher weight.

快速上手 | Quick Start

這是什麼？ | What is this?

• Pony: People's Works （ppw）是一個實驗性微調模型系列，數據集約85%來源於 CivitAI 上用戶發布的 AI 生成圖片。早期 ppw數據集最初建立於pony v6生成的圖片基礎上，因此本系列模型生成的圖片帶有部分 Pony Diffusion 特徵。

• 本系列模型使用標準 Danbooru 標籤，主要擅長生成中、近景風格化人像。其主要功能是在不使用畫師關鍵詞及較少品質提示詞的情況下，保持穩定的圖像質量，節省提示詞中的token 空間。
  此模型非風格 LoRA，不同提示詞和生成條件下可能有細微風格差異。

• Pony: People's Works (ppw) is a experimental fine-tuned model series, approximately 85% of the dataset comes from AI-generated images published by users on CivitAI. Since the earlier ppw dataset was built on images generated by Pony V6, the outputs of this series also carry some characteristics of Pony Diffusion.

• This series uses standard Danbooru tags and is mainly optimized for generating stylized portraits at medium and close range. The primary effect of this model series is to allow the basemodel to achieve relatively stable image quality, without artist keywords or long quality tags, freeing up token space for prompts.
  These models are not style LoRAs. There may be subtle stylistic variations depending on different prompts and generating conditions.

版本信息 | Version Info.

• 本頁面發布的是 ppw 的高維 LoCon版本，也是本項目的主頁面。

• LoCon 版本的 ppw 可以靈活搭配各類功能 LoRA 和底模，且效果強度具更高可控性。高維版本的 LoCon 擁有更強泛化能力和細節表現，但會佔用更多儲存空間及計算資源。

• 主要用於線上生成服務和高性能電腦用戶的本地生成。

• This page features the high dim LoCon version models of ppw, which also serves as the main page of this project.

• The LoCon versions of ppw can be flexibly combined with various functional LoRAs and checkpoints, offering greater controllability over effect weight. High dimension versions provide stronger generalization and more detailed rendering, but it requires more storage space and computational resources.

• They are mainly intended for online generation services and local use by users with high-performance PCs.

精簡版 LoCon | Lightweight LoCon ver.

基礎模型版 | Checkpoint versions (Illustrious)

基礎模型版 | Checkpoint versions (NoobAI)

更多 | more

使用方法 | Usage

正向提示詞 (positive):

masterpiece, best quality, very aesthetic

負向提示詞 (negative):

low quality, displeasing

更新記錄 | Change log

v7

v7 版本對數據集結構做了較大幅度調整，並採用了不同的訓練參數和策略，因此可能導致 v7 不如以前的版本穩定。

The v7 version has undergone significant structural adjustments to the dataset, and utilizes different training parameters and strategies. As a result, v7 may be less stable than the previous versions.

v-pred 模型在 CivitAI 在線生成器上的表現與 TensorArt 的在線生成表現完全不同，相同參數無法復現結果，我也不清楚原因。

The v-pred model's performance on the CivitAI online generator is completely different from online generation on TensorArt. The results are entirely unreproducible with a same parameters. I have no idea why...

TensorArt 版本 CivitAI 使用相同參數版本 和 CivitAI 高權重版本

v7 版本簡介：

這是基於前作數據集發展的圖像質量 LoCon，約 90%-95% 圖像數據來自 CivitAI 上發布的圖片。

它允許模型在不使用畫師標籤及較少品質提示詞的條件下達成相對穩定的圖像質量，節省更多 token 空間，並可修復部分模型固有的生成缺陷（不包括手部）。

由於數據集選擇因素，生成圖片帶有 Pony 的質感；但因未指向特定畫師、風格或繪畫技法，不同提示詞與模型條件下可能會有細微風格差異。

This is a generation quality LoCon developed based on the dataset from the previous work. About 90%-95% of the image data comes from CivitAI.

It allows models to achieve relatively stable image quality without artist tags or using long quality prompts, freeing up more token space. Additionally, it can fix some inherent generation flaws of the model. (except for hands)

Due to the dataset selection, the generated images exhibit a Pony-like style. However, since it does not reference any specific artist, style, or painting technique, there may be subtle stylistic variations depending on different prompts and checkpoint conditions.

數據集來源及許可證 | Dataset Source & License

• 數據集內每張圖片皆由作者人工篩選、分類及標註編輯，其中數百張圖片經過手工修正。

• 本模型為免費、開源模型，使用者可於私人設備自行部署。作者不會從模型銷售中獲利。作者不限制本系列模型用於商業生成服務或商業圖像生成，但請注意配合使用的 Checkpoint 與其他 LoRA 授權限制。

• 約 90%-95% 資料為 AI 生成，但約 250+ 張圖片來自公共媒體、新聞及出版物作概念補充，未來版本將逐步更換相關素材。有商用需求者請留意潛在風險。

  此數據集未包含任何個別畫師作品，也未標註畫師資訊（但不排除 AI 標註錯誤）。

• 此外，本模型禁止用於閉源商業用途、模型轉售，以及融合至閉源商業模型。對於開源融合模型用於生成服務則無限制，但建議標明融合模型來源。

• Every image in the dataset has been manually selected, categorized, and annotated by the author. Additionally, hundreds of the images have been manually edited and corrected.

• This model is free and open-source model, allowing users to deploy it on their personal devices. The author does not receive any compensation from selling the model. The author does not impose restrictions on using this model for commercial image generation services or generating images for commercial purposes. However, please be mindful of the license restrictions of the Checkpoint and other LoRAs used alongside this model.

• Approximately 90%-95% of the dataset consists of AI-generated images. However, around 250+ images have been collected from public media, news outlets, and publications to supplement concepts. Future versions will gradually replace these materials. Users with commercial intentions should be aware of the potential risks.

  This dataset does not include training data from any individual artist, nor does it contain explicit artist attributions (though AI mistagging cannot be entirely ruled out).

• Additionally, this model is not permitted for use in closed-source commercial applications, model resales, or merged into closed-source commercial models. There are no restrictions on open-source merged models being used for image generation services, but it is recommended to credit the sources of any merged models.