Please consider joining my <a target="_blank" rel="ugc" href="https://patreon.com/Zavy376">Patreon </a>so I can keep most of my work available for everyone - I'll be releasing early access models there for a cheaper monthly fee than buying them here individually, and also will be providing exclusive models and more! For business inquiries, commercial licensing, custom models, and consultations, please get in touch at <a target="_blank" rel="ugc" href="mailto:hello@zavyai.com">hello@zavyai.com</a> or <a target="_blank" rel="ugc" href="mailto:sales@zavyai.com">sales@zavyai.com</a>. You can also contact me here through CivitAI DM or join my <a target="_blank" rel="ugc" href="https://discord.gg/jZZA4Dzmta">Discord</a>.<hr /><h1 id="introduction-aad4bba1g">Introduction</h1><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/69d26f21-af4b-4df3-9a96-32a1b41329b8/width=525/69d26f21-af4b-4df3-9a96-32a1b41329b8.jpeg" />Many anime/comic models available here tend to have a harsh and highly detailed style with thin outlines, which may not be to everyone's preference. Therefore, I have specifically focused on addressing these three aspects in this model to reach a more authentic looking anime/comic style. The output it generates consists of softer and flatter images with thicker outlines. This particular style makes it relatively easy to edit areas that may not meet your liking with basic knowledge of image editing tools. Consequently, it becomes easier to shape an image according to your desired vision using img2img.I kindly request that you share your creations both here and on my <a target="_blank" rel="ugc" href="https://discord.com/invite/jZZA4Dzmta">Discord </a>server, as I would greatly appreciate the opportunity to see them.<h1 id="pros-and-cons-rjpi2hk51">Pros and cons</h1>Pros<ul><li>Excels in anatomically correct fingers and toes.</li><li>Produces impressive results without relying on terms like "masterpiece" or "best quality" prompts. This allows for a more efficient use of token space in the prompts, enabling clearer and more effective instructions to guide the image creation process according to your specific vision.</li><li>Produces images with minimal artifacting and random smudges. The resulting outputs are cleaner and exhibit fewer undesired visual distortions.</li><li>The images generated by this model have a noticeably soft appearance. The output tends to have a gentle and smooth quality, which can contribute to a more soothing and aesthetically pleasing visual experience.</li><li>The flatter image style produced by this model makes it easier to edit as there is less colour complexity in the gradients. The reduced presence of gradients simplifies the editing process, allowing for more straightforward adjustments to achieve the desired outcome.</li><li>The emphasis on thick outlines enhances the visual style, giving the artwork a distinct and recognizable appearance commonly seen in traditional anime and comic illustrations.</li><li>This model is highly receptive to various levels of detail, allowing you to adjust the complexity (with detail enhancing LORAs especially) of the image according to your preferences.</li><li>This model excels in producing exceptional results when generating grayscale manga-style images.</li></ul>Cons<ul><li>In certain cases, the image style can be significantly influenced by certain LORAs, resulting in a harsher appearance that deviates from the intended softness of this model. To address this, techniques such as block weighting or utilizing img2img can be employed to restore the desired softness and bring the image closer to the original intent of the model.</li><li>The model occasionally generates jagged-looking pupils, which may require manual editing to achieve a more accurate and visually pleasing appearance.</li></ul><h1 id="roadmap-9b3he0u0t">Roadmap</h1><ol><li><s>In-painting model release.</s> Done.</li><li><s>Update the model to minimise some of its drawbacks and incorporate a more current blend of models.</s> Done.</li><li><s>Model inspired by anime and comic styles.</s> Done.</li><li>SDXL models.</li><li>Pioneering uncharted LORA subjects (withholding specifics to prevent preemption)</li></ol><h1 id="tips-p0ph3iday">Tips</h1><ul><li>To better understand the preferences of the model, individuals are encouraged to utilise the provided prompts as a foundation and then customise, modify, or expand upon them according to their desired objectives.</li><li>Avoid using face restoration techniques as they are often unnecessary and can lead to inferior results compared to the original model.</li><li>Consider utilising Dynamic Thresholding as a method to control CFG. This technique can help you get improved results.</li><li>If you find that the details in your work are lacking, consider using adetailer (or similar LORA) if you’re unable to fix it with prompt alone. Adetailer or similar tools can enhance and enrich the level of detail, but in turn it will lose some of the softness intended by this model.</li><li>For even better LORA results, make use of the LORA block weighting. By incorporating that, you can further enhance the outcome and achieve superior results in the LORA process.</li><li>Avoid using underscores in your prompt, unless part of the trigger word for a LORA.</li><li>To maintain optimal results and avoid excessive duplication of subjects, limit the generated image size to a maximum of 768x768 pixels or 768x1024 pixels (or vice versa). If you require higher resolutions, it is recommended to utilise the Hires fix, followed by the img2img upscale technique, with particular emphasis on the controlnet tile upscale method. This approach will help you achieve superior results when aiming for higher resolution outputs.</li></ul><h1 id="prompts-and-ti-bs3ojurrn">Prompts and TI</h1>No specific positive or negative prompts are specifically recommended for this model. It functions well as a base without relying on any particular type of prompt. You have the freedom to mix and match prompts according to your preferences, increasing or decreasing their influence in the direction you desire. This flexibility allows you to experiment and customise the output to your liking, shaping the image generation process to align with your creative vision.For negative prompts, consider employing Textual Inversion techniques. The following are recommended: <a target="_blank" rel="ugc" href="https://civarchive.com/models/7808/easynegative">easynegative</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/models/55700?modelVersionId=60095">bad_prompt</a>, and <a target="_blank" rel="ugc" href="https://civarchive.com/models/16993/badhandv4-animeillustdiffusion">badhandv4</a>. However, there are other TI available that potentially yield better results. Feel free to experiment with different TI and share your findings!<h1 id="recommended-settings-6qkbb4hb9">Recommended settings</h1><ul><li>Vae-ft-mse-840000-ema-pruned.vae. (Other more anime oriented ones like blessed2.vae work well too).</li><li>Euler A.</li><li>Steps 30~40 (can go lower with DPM++ 2M Karras, but I’ve found Euler A to make a more pleasing image).</li><li>Hires upscaler: 4x_foolhardy_Remacri or 4x-UltraSharp depending on image.</li><li>Hires upscale: Whatever maximum your GPU is capable of, but preferably between 1.5x~2x.</li><li>CFG scale 4-8 (unless you use Dynamic Thresholding).</li><li>Clip skip 2.</li></ul><h1 id="model-recipe-gri9nukaw">Model recipe</h1>The recipe for the model is quite complex, as it involves utilising different weight settings for each individual input and output layer in each model used in the mixture. Consequently, the following instructions may not yield results that closely resemble this specific model. However, the fundamental recipe is as follows:<pre><code>Model A:
duchaitenLofi
Model B:
realcartoon3d
Model C:
kidsmix_v10

Model A:
ZavyComics_a1
Model B:
childrenStories_v13D
Model C:
flat2DAnimerge_v30</code></pre>I would like to express my gratitude to the creators of the models incorporated in this mixture, as well as the creators of the models that came before them.<h1 id="social-media-mtpa2lef2">Social media</h1>You are welcome to join my recently created <a target="_blank" rel="ugc" href="https://discord.com/invite/jZZA4Dzmta">Discord </a>server (it’s not very active yet, I’ll ramp up activity there with SDXL), where we can engage in discussions, share our experiences in AI, and showcase the things we’ve made with AI. You are encouraged to join and ask any questions or seek additional tips and tricks related to my models or AI in general. Your participation would be greatly appreciated. Furthermore, I will be releasing early testing versions of my models on Discord. If you are interested in being among the first to try them out, I invite you to join the Discord server.<a target="_blank" rel="ugc" href="https://discord.gg/jZZA4Dzmta">https://discord.gg/jZZA4Dzmta</a>

Files with SHA256: bc4ba4c7ab388e8b7987f1f1b47c5b38ffe85f2adabe1d7de51ec72416183ea4