☕ Support My Work: If you enjoy my local tools and models and want to support me with real cents rather than just platform Buzz, you can do so here: <a target="_blank" rel="ugc" href="https://ko-fi.com/monsieurbaka">https://ko-fi.com/monsieurbaka</a>. Your support directly speeds up new software updates and model releases!<h3 id="core-trigger-tags-and-prompting">Core Trigger Tags &amp; Prompting</h3>To trigger the aesthetic, use:<ul><li><code>score_9, masterpiece, @tsm2330 style</code> or simply <code>@tsm2330 style</code></li></ul><h3 id="automation-and-workflows-(the-promptforge-ecosystem)">Automation &amp; Workflows (The PromptForge Ecosystem)</h3>To save you from manually copying and pasting style tags or tweaking configuration JSONs, I am releasing the official PromptForge library integration alongside this model:<ul><li><a rel="ugc" href="https://huggingface.co/datasets/AbdiZerga/tsmStylePromptForge/tree/main">[HUGGINGFACE LINK]</a> Official PromptForge Library Pack addon for this style (<code>.txt</code>, <code>.jpg</code>, <code>.json</code>)</li></ul>For those who value their time and want to streamline multi-character prompting alongside complex style LoRA management directly in ComfyUI:PromptForge is a standalone software designed to automate prompting workflows and seamlessly handle complex LoRA dependencies during image generation. The brand-new major release is faster, more functional, and completely overhauled for max workflow efficiency. Read the full changelog, guide, and download instructions in the official article:👉 <a target="_blank" rel="ugc" href="https://civarchive.com/articles/32428/promptforge-2-automatization-strikes-back">https://civarchive.com/articles/32428/promptforge-2-automatization-strikes-back</a><h2 id="tsm2330-art-style">tsm2330 Art Style</h2><h3 id="showcase-setup-and-model-weight">Showcase Setup &amp; Model Weight</h3><ul><li>Recommended LoRA Weight: <code>1.0</code></li><li>Showcase Setup: The entire showcase gallery on this page was generated at a solid weight of 1.0 for this Style LoRA (and <code>1.0</code> for character LoRAs whenever paired).</li></ul>We aren't hiding behind low weights here - this page is a raw flex to demonstrate that the model has deeply baked the artist's unique lighting dynamics, volumetric depth, and material rendering directly into its weights without collapsing the underlying geometry.<h3 id="about-the-artist-and-style-breakdown">About the Artist &amp; Style Breakdown</h3>This model is a dedicated style LoRA trained on the public artwork dataset of the talented neural artist tsm2330. In my opinion, tsm2330 has achieved the ultimate sweet spot in anime-style rendering: a perfect balance between sharp, aggressive rim lighting, soft shadows, and vibrant subsurface scattering (SSS) for skin materials.<h3 id="dataset-architecture-and-captioning-(shoutout-to-grok)">Dataset Architecture &amp; Captioning (Shoutout to Grok)</h3><ul><li>Dataset Size: 250 surgically clean images, all manually curated, hand-filtered, and carefully pre-processed.</li><li>The Captioning MVP: Credit where credit is due - a huge chunk of the detailed captions was generated with massive help from Grok. When it comes to unhinged, uncensored, and detailed description of spicy/NSFW visual content, Grok remains the only capable non-human assistant on the market willing to write captions without throwing a censorship fit.</li><li>NSFW Bias: Due to the specific nature of tsm2330’s portfolio, a significant portion of the dataset consists of NSFW material. As a result, the model possesses a strong natural bias toward realistic skin rendering and detailed anatomy. Keep this in mind when crafting your prompts!</li><li>Zero Concept Bleeding: Despite the high stylistic dominance, all recurring characters inside the training set were isolated on a semantic token level. Calling a specific lighting setup or pose tag will never force an unwanted character's features or hair traits into your generation.</li></ul>

Files with SHA256: e2ddab4916df9ee5baa0b72d6bedfef53d86d8cc47db66fb7aaf78c86b565db4