<p>Special thanks to <span class="MentionNode-module-scss-module__nz4R1q__mention" data-type="mention" data-id="mention:7465689" data-label="KH38MT">@KH38MT</span> for providing gpt-image-2 access, which made the dataset collection much quicker.<br />Feel free to check out his profile for the loras he published too!</p><h1 id="quick-version-guide">⚡ Quick Version Guide</h1><p><span style="color:rgb(250, 82, 82)"><strong>v1</strong></span></p><p>Default girl = mature type.</p><p>Exquisite details and beautiful backgrounds.</p><p>👉 Pick this if you like the mature lady vibe.<br /></p><p><span style="color:rgb(64, 192, 87)"><strong>v2-petite</strong></span></p><p>Default girl = petite / small body type.</p><p>Cleaner image with less "noise" (fewer details, simpler backgrounds).</p><p>👉 Pick this if you want petite girls and prefer a cleaner look.<br /></p><p><span style="color:rgb(76, 110, 245)"><strong>v2.1 ✨</strong></span></p><p>Default girl = petite / small body type.</p><p>Details and background quality on par with v1.</p><p>👉 Pick this if you want petite girls BUT also want the exquisite details from v1.<br /></p><p><span style="color:rgb(121, 80, 242)"><strong>v3-fusion</strong></span></p><p>Default girl = neutral (not locked to petite or mature).</p><p>Detail level sits between v1 and v2.</p><p>👉 Pick this if you want a versatile all-round version and can accept slightly less detail than v1.</p><p></p><h1 id="basic-info">📌 Basic Info</h1><p>All the training data comes from images I generated using a style I really loved that I rolled from gpt-image-2.</p><p><span style="color:rgb(34, 139, 230)">Train model</span>: Anima_base_v1.0</p><p><span style="color:rgb(34, 139, 230)">Trigger word</span>: @gpt2</p><p><span style="color:rgb(34, 139, 230)">Weight</span>: 0.8 ~ 1.5</p><p><span style="color:rgb(34, 139, 230)">CFG</span>: 4.5</p><h3 id="sampler-+-scheduler"><span style="color:rgb(34, 139, 230)">Sampler + Scheduler</span></h3><p>Best: exp_heun_2_x0_sde + beta (best quality but very slow)</p><p>Fine: dpmpp_3m_sde + sgm_uniform</p><p>---</p><h1 id="version-notes">📝 Version Notes</h1><h2 id="v2.1-(latest)"><span style="color:rgb(76, 110, 245)">v2.1 (Latest)</span></h2><p>This is released after v3. It's called v2.1 because it's a direct upgrade to v2, not a new direction.</p><p>I took v1's dataset captions, changed the subject from mature woman to petite girl, and trained a new lora.</p><p>So basically you get v2's petite body type with v1 level details and backgrounds.</p><p>If you liked v2 but wished it had more details, this is the one.</p><h2 id="v3-fusion"><span style="color:rgb(121, 80, 242)">v3-fusion</span></h2><p>So... v2 did make the characters petite, but it came with a cost.</p><p>It lost a lot of details. The compositions felt less refined, and the backgrounds became super plain unless you prompt really hard.</p><p>To fix this, I combined the v1 and v2 datasets together to train v3.</p><p><strong>Pros</strong> 👍:</p><p>Brings back the "exquisite" details and beautiful effortless backgrounds from v1.</p><p>Still keeps the default "1girl" output in a smaller, youthful "maiden" style (won't revert back to v1's mature look).</p><p>Basically v3 is the best of both worlds. You get the details from v1 with the body proportions from v2.</p><h2 id="v2"><span style="color:rgb(64, 192, 87)">v2</span></h2><p>v1 leans too much towards mature characters and can't really do "petite" well.</p><p>To fix this, I generated 16 new petite images using gpt-image-2 to train v2.</p><p>(Even though I'm not really into petite women myself lol)</p><p>The v2 dataset is completely different from v1, but the art style stays exactly the same.</p><p>BTW if you want to use "The Tag That Must Not Be Named" 🥑, v2 works WAY better than v1.</p><h2 id="v1"><span style="color:rgb(250, 82, 82)">v1</span></h2><p>Pros 👍: You get beautiful backgrounds without specifying details in your prompt, and no extra detailer needed.<br /></p><h1 id="v2.1">V2.1</h1><h1><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/491fca73-871d-400e-a89e-f08a5a4b9c97/original=true/491fca73-871d-400e-a89e-f08a5a4b9c97.jpeg" /></h1><h1 id="v3:">v3:</h1><h1><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/17c8f347-615a-47a6-9b3e-1682be4da881/original=true/17c8f347-615a-47a6-9b3e-1682be4da881.jpeg" /></h1><h1 id="v2:">v2:</h1><h1 id="v1:"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/d428d8b8-657d-4412-8adf-0fa803dce8d8/original=true/d428d8b8-657d-4412-8adf-0fa803dce8d8.jpeg" />v1:<br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7000b5f9-bcd8-4176-820b-4b084d93509a/original=true/7000b5f9-bcd8-4176-820b-4b084d93509a.jpeg" /></h1><p></p>

Ps GPT-image-2 style

Files with SHA256: 2eaa0e52810e8955ad3f85653ecd2a17d7746ca257bc58234cfd831ae48e5033