Recommendations (v3.0 - Anima)
Trigger: fade \(valorant\) (escape the brackets in ComfyUI). Weight: 1.0 (at 0.8 the character fades out; 1.2 still works).
For her in-game outfit add dark blue hooded tactical jacket; in other outfits and scenes add black and white ombre hair. Settings: 30 steps, CFG 4, er_sde / simple. 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 Fade Valorant
Recomendations:
Lora weight: 0.5 (0.4-0.7)
Tokens (only v1): You can add Realistic or Illustration tokens to control the style.
You can also try adding tokens: shoulders tattoo, heterochromia, blue eyes, scar on nose, orange eyes, jacket
Description
Anima Base v1.0 version, trained locally (ai-toolkit) on a 33-image dataset.
Trigger: fade \(valorant\). Escape the brackets in ComfyUI, otherwise they are read as weight syntax.
Weight: 1.0. At 0.8 and below the character fades out (see the weight comparison); 1.2 still works but leans toward the fan-art look with the jacket open.
Without the LoRA Anima draws a different agent. With it you get the black-to-white ombre hair, the dark stripe across her face, heterochromia, the nose piercing, dark lipstick and the dark blue hooded jacket.
Useful tags (from the training captions):
dark blue hooded tactical jacket— her in-game outfitblack cropped top, black cargo pants, tattooed forearm— the fan-art lookblack and white ombre hair— keeps the hair right in other outfits and scenes (without it one test came out blonde)heterochromia, blue and red eyes, nose piercing, dark lipstick— close-up details
Tested with: 30 steps, CFG 4, er_sde / simple, 1024×1024 and 896×1152.
Negative: worst quality, low quality, score_1, score_2, score_3, blurry, jpeg artifacts, sepia, signature, watermark, text, artist name (add nsfw, nude for SFW).
DiT-only LoRA (no text encoder weights), rank 16 / alpha 16, 2000 steps.



