Recommendations (v2.0 - Anima)
Trigger: sage \(valorant\) (escape the brackets in ComfyUI). Weight: 1.0 (0.8–1.2); use 0.8 for outfit changes.
Settings: 30 steps, CFG 4, er_sde / simple, 896×1152 or 1024×1024. Add signature, watermark, text, artist name to the negative. More tips are in the version notes.
XL version of LORA, trained on model generation results Valorant Sage / CIXF and random art that was lying around on my disk.
Recomendations:
Lora weight: 0.5 (0.4-0.7)
Tokens: You can add Realistic or Illustration tokens to control the style.
Try adding a bodysuit to get a more covered suit
Description
Anima Base v1.0 version, trained with the Civitai trainer on 33 images.
Trigger: sage \(valorant\). Escape the brackets in ComfyUI, otherwise they are read as weight syntax.
Weight: 1.0 (0.8–1.2). Use 0.8 for outfit changes.
The trigger alone gives her in-game look: long black high ponytail, teal earrings, the white top with teal trim, the teal sash, black leggings and boots. Base Anima already knows her when the brackets are escaped, but the outfit drifts (a teal top, gold hair ornaments); the LoRA keeps it consistent and adds the art style of the dataset (see the weight comparison).
Tips:
Works at 896×1152 and 1024×1024.
Outfit changes: use weight 0.8 and describe the clothes fully (e.g.
casual clothes, white hoodie); the teal collar of the uniform tends to show through.The Halloween witch look from the dataset works at 0.8:
witch hat, skull ornament, black off-shoulder dress, teal trim, black choker, elbow gloves.Proportions follow the dataset;
slimdoes not change them.
Tested with: 30 steps, CFG 4, er_sde / simple.
Negative: worst quality, low quality, score_1, score_2, score_3, blurry, jpeg artifacts, sepia, signature, watermark, text, artist name (add nsfw, nude, topless for SFW). The dataset has artist signatures; the signature tags keep the corners clean.
DiT-only LoRA (no text encoder weights), dim 32 / alpha 32, 1500 steps.








