<h3 id="warning:-version-1.0-and-all-other-variations-of-version-1.0-are-experimental."><span style="color:rgb(250, 82, 82)"><strong>Warning: </strong></span><span style="color:rgb(250, 176, 5)"><strong>Version 1.0 and all other variations of version 1.0 are experimental.</strong></span></h3><h3 id="if-you-notice-low-quality-please-report-it."><span style="color:rgb(250, 176, 5)"><strong>If you notice low quality, please report it.</strong></span></h3><p></p><hr /><p></p><h2 id="about-version-1.0-crystal"><strong>About Version </strong><span style="color:rgb(250, 82, 82)"><strong>1.0-Crystal</strong></span></h2><p></p><h3 id="version-1.0-crystal-is-the-most-complete-and-best-variation-of-version-1.0."><strong>Version 1.0-Crystal is the </strong><span style="color:rgb(64, 192, 87)"><strong>most complete </strong></span><strong>and best variation of version 1.0.</strong></h3><h3 id="it-features-the-highest-quality-and-excellent-prompt-following-capability."><strong>It features the highest quality and excellent prompt following capability.</strong></h3><h3 id="this-version-is-available-in-two-formats:-fp16-and-fp32."><strong>This version is available in two formats:</strong><span style="color:rgb(64, 192, 87)"><strong> fp16 and fp32.</strong></span></h3><h3></h3><h3 id="for-higher-quality-and-better-details-i-highly-recommend-using-the-fp32-version."><span style="color:rgb(250, 176, 5)"><strong>For higher quality and better details, I highly recommend using the fp32 version.</strong></span></h3><h3></h3><h3 id="unfortunately-version-1.0-orange-suffered-from-data-bias-issues."><span style="color:rgb(34, 139, 230)"><strong>Unfortunately, version 1.0-Orange suffered from data bias issues.</strong></span></h3><h3></h3><h3 id="good-support-of-lora."><span style="color:rgb(253, 126, 20)"><strong>Good support of LoRA.</strong></span></h3><h3></h3><h3 id="version-1.0-crystal-includes-numerous-significant-improvements-particularly-in-the-unet-and-text-encoder-components."><strong>Version 1.0-Crystal includes numerous significant improvements, particularly in the </strong><span style="color:rgb(130, 201, 30)"><strong>UNet and text encoder components.</strong></span></h3><h3></h3><h3 id="recommended-settings-for-version-1.0-crystal:"><strong>Recommended settings for version 1.0-Crystal:</strong></h3><ul><li><p><strong>Sampling Method: LCM</strong></p></li><li><p><strong>Steps: More than 8 steps (recommended: 10 steps)</strong></p></li><li><p><strong>CFG Scale: 1.0–1.5 (1.5 is preferred)</strong></p></li><li><p><strong>CLIP Skip: 1</strong></p></li></ul><h3></h3><h3 id="hires-fix:-if-you-have-access-to-hires-fix-please-use-it.-otherwise-make-sure-to-refine-the-image-generated-by-txt2img-using-img2img.-for-refining-(img2img)-use-a-denoising-strength-between-0.4-and-0.6."><strong>Hires Fix:</strong><span style="color:rgb(130, 201, 30)"><strong> If you have access to Hires Fix, please use it. Otherwise, make sure to refine the image generated by txt2img using img2img. For refining (img2img), use a Denoising Strength between 0.4 and 0.6.</strong></span></h3><h3 id="please-report-any-low-quality-images-you-encounter-during-image-generation."><span style="color:rgb(250, 82, 82)"><strong>Please report any low-quality images you encounter during image generation.</strong></span></h3><h3></h3><h3 id="thank-you-ysnrfd"><span style="color:rgb(21, 170, 191)"><strong>Thank you, YSNRFD</strong></span></h3><h3></h3><hr /><p></p><h3 id="about-version-1.0-orange"><strong>About Version </strong><span style="color:rgb(250, 82, 82)"><strong>1.0-Orange</strong></span></h3><p><br /><strong>This Version have</strong><span style="color:rgb(18, 184, 134)"><strong> better VAE</strong></span><strong> (baked)<br /><br /></strong><span style="color:rgb(21, 170, 191)"><strong>Good support of LoRA (enhanced)</strong></span><strong><br /><br /></strong><span style="color:rgb(21, 170, 191)"><strong>CLIP model enhanced</strong></span><strong><br /><br /></strong><span style="color:rgb(64, 192, 87)"><strong>better UNet pruning</strong></span><strong><br /><br />Settings:<br /><br />Sampling method:</strong><span style="color:rgb(64, 192, 87)"><strong> LCM</strong></span><strong><br />CFG scale: </strong><span style="color:rgb(64, 192, 87)"><strong>1.0-1.5</strong></span><strong><br />Steps: </strong><span style="color:rgb(64, 192, 87)"><strong>6&lt;X</strong></span><br /><strong>CLIP skip:</strong><span style="color:rgb(255, 255, 255)"><strong> </strong></span><span style="color:rgb(64, 192, 87)"><strong>1.0</strong></span><strong><br /><br /></strong><span style="color:rgb(230, 73, 128)"><strong>Mid-Range Anime Style Checkpoint</strong></span></p><p></p><hr /><p></p><h3 id="about-version-1.0"><strong>About Version </strong><span style="color:rgb(250, 82, 82)"><strong>1.0</strong></span></h3><p></p><h3 id="an-mid-range-anime-checkpointgood-support-of-lorasampling-method:-lcmsteps:-8lessxcfg-scale:-1.0">An Mid-Range Anime checkpoint<br /><br /><strong>Good support of LoRA</strong><br /><br />Sampling method: <span style="color:rgb(64, 192, 87)">LCM</span><br />Steps: <span style="color:rgb(64, 192, 87)">8&lt;X</span><br />CFG scale: <span style="color:rgb(64, 192, 87)">1.0</span></h3><p></p>

Wonder Dream LCM

<p><strong>This Version have better VAE (baked)<br /><br />Good support of LoRA (enhanced)<br /><br />CLIP model enhanced<br /><br />better UNet pruning<br /><br />Settings:<br /><br />Sampling method: LCM<br />CFG scale: 1.0-1.5<br />Steps: 6&lt;X<br /><br />Mid-Range Anime Style Checkpoint</strong></p>

Files with SHA256: 843dcdd1c70c316f53fc9ba3161df40d9c9391bf0743ccf491866abf6888452b