<h2 id="nick-and-charlie-or-lora-heartstopper-or-base:-anima">🍃 Nick &amp; Charlie | LoRA — Heartstopper | Base: ANIMA</h2><p><strong>Captures the iconic hand-drawn art style of Alice Oseman's <em>Heartstopper</em> webcomic and brings Nick Nelson and Charlie Spring to life in ANIMA.</strong></p><hr /><h2 id="before-anything-else">🙏 Before anything else</h2><p>This LoRA took a lot of work — picking the images, building the dataset and tags, training, and testing the checkpoints. If you have a moment, drop a <strong>Like</strong> and especially <strong>post your results in the comments</strong> (with the prompt/settings you used) — it helps me a lot to know what to improve in the next versions. 🔥</p><blockquote><p>⚠️ The model was trained <strong>without any NSFW images</strong> and <strong>without any photos of the live-action actors</strong>. It is strictly based on the original graphic novel / webcomic art style. Use it respectfully.</p></blockquote><p>💬 Check out the image descriptions and prompt examples on the posted preview images — they show how the prompts are structured!</p><p>💬 Leave your generations in the comments, and tag the LoRA so I can see how it's looking — it would help a lot.</p><p>📦 Trained with <strong>92</strong> carefully selected images in total from the original webcomic art.</p><hr /><h2 id="about-this-model">📊 About this model</h2><p>The main goal of this LoRA is to capture the iconic hand-drawn art style of Alice Oseman's <em>Heartstopper</em> webcomic and bring Nick Nelson and Charlie Spring to life in ANIMA.</p><p>While the LoRA is not 100% perfect, it works extremely well!</p><p></p><p>💬 Just to remind you, the comics have simple backgrounds and pastel tones, emphasize this in the prompt, otherwise the characters won't 'match' the rest of the work</p><h3 id="nick-nelson">🌀 Nick Nelson</h3><p>Nick generates consistently with high quality across most poses and outfits. His facial features, hair shape, and athletic frame lock in very reliably.</p><h3 id="charlie-spring">🌀 Charlie Spring</h3><p>Charlie's output quality depends significantly on the generation resolution. Since the original comic source drawings vary widely in scale and detail, generating Charlie effectively may require a few resolution tweaks and test runs until you find the sweet spot for your specific scene.</p><h3 id="using-the-art-style-with-other-characters">🎨 Using the Art Style with Other Characters</h3><p>If you want to extract the <strong>Heartstopper art style</strong> without generating Nick or Charlie:</p><ul><li><p>Use the trigger tags <code>heartstopper</code> + <code>webcomic</code> in your prompt.</p></li><li><p>Combine this with other character LoRAs to apply Heartstopper's unique hand-drawn pastel aesthetic to any character!</p></li></ul><hr /><h2 id="generation-tips">ℹ️ Generation tips</h2><ul><li><p><strong>Resolution Flexibility:</strong> There is no single "standard resolution" for this LoRA. Because the source art varies greatly, trying different resolutions (portrait, square, or wide 2:3 / 3:4) depending on scene context makes a huge difference in image quality.</p></li><li><p><strong>Fixed Seed Testing:</strong> When experimenting with resolutions or prompt tweaks, try using a fixed seed to observe how resolution directly impacts detail and character rendering.</p></li><li><p><strong>Recommended starting point:</strong> On ANIMA base with 12 steps + acceleration LoRA, the output keeps a clean hand-drawn webcomic feel that fits Heartstopper perfectly.</p></li><li><p>Check the descriptions of the posted preview images — they demonstrate how to structure tags and natural language descriptions (or multi-panel layouts) for best results.</p></li></ul><hr /><h2 id="generation-settings-used-for-preview-images">⚙️ Generation settings used for preview images</h2><p>All preview images were generated using:</p><ul><li><p>Base model: Anima 1.0</p></li><li><p>Acceleration LoRA: strength 1.0</p></li><li><p>CLIP: anima default</p></li><li><p>Steps: 12</p></li><li><p>CFG: 1.0</p></li><li><p>Sampler: <code>er_sde - beta</code></p></li><li><p>Image size: there’s no fixed resolution for this LoRA; it works at resolutions above 512x and in any format, although the performance is pretty hit or miss. I recommend testing various resolutions for your specific prompt.</p></li></ul><hr /><h2 id="one-more-thing">👍 One more thing</h2><p>I'm just getting started with this — if you have a moment, drop a <strong>Like</strong> and share your generations with the settings you used. It really helps me improve the next versions. Have fun! 🍃</p>

Files with SHA256: 0d9a7113fd65cc0c87951dbcc7ea9e389609d887d2f5e868a83165a24f1455b2