💢 Cuteness is justice!! 💢Hey everybody, it's been a while. I'm back with Sparkla this time, certainly a kindred spirit that understands that Maou-chan is the cutest in the world.I'm testing a different approach to training this time (bunch of new settings, different tagging, etc etc). So expect a few more LoRAs in the near future and some updates to some of my older ones.15/06/2026: Added Anima B1 version.Give it a try! Have fun!<hr /><h3 id="recommended-weight:-0.8-~-1.0">Recommended weight: 0.8 ~ 1.0</h3><hr /><h3 id="default-base-appearance">Default - Base Appearance</h3><pre><code>stell4sp4rkl3, purple hair, yellow eyes, hair over one eye, under-rim glasses, hair flaps, long hair, lantern hair ornament, purple star, large breasts,</code></pre><h3 id="default-character-outfit">Default - Character Outfit</h3><pre><code>green beret, white shirt, three-quarter sleeves, chest harness, green capelet, green bowtie, leather belt, green skirt, medium skirt, side slit, brown pantyhose, leather gloves, brown mary janes, footwear bow,</code></pre>Additional tags / Accesories / Weapons<pre><code>wheelchair, steampunk,</code></pre>There's literally a single angle for the Wheelchair in the Dataset (yes, the one from her game sprite). So, expect the wheelchair be super stiff and difficult to generate.<hr />

Sparkla / スパクル (Stella Sora / ステラソラ) [Illustrious/Anima]

Testing a different approach to training this time:<ul><li>Ditched raw Kohya for LoraEasyTrainingScripts.</li><li>Changed network from LoRA to LoCON (Lycoris) with much lower ranks and max allowable batch size (around 50MB size for output files now).</li><li>Switch to CAME + REX annealing (no repeats, no warmup, but slight weight decay) for optimizer and scheduler.</li><li>Reduced datasets to &lt;=30 images (only 15 images in this case) with full tagging.</li></ul>So far, the results have been pretty good (have been testing them for the last month). I have like 4 different characters already trained with this setup now, posting them later this week hopefully.

Files with SHA256: b0f0dc5d77d61f0dae0fe51c9e279028a6991909e9b2119b451382782fe8c1e3