Blossum — Character LoRA for Anima
An independently trained Anima LoRA of Blossum, the sakura tree spirit associated with @blossumVT.
Pink hair, pointed ears, flowering branches, and cherry blossoms throughout the design. The references include several hairstyles and outfits, with the same floral identity running through them.
WHY I TRAINED THIS
VTuber promoting fake Glaze and Nightshade alternatives to gaslight people into "protecting" their art by deliberately making it ugly.
You can dislike AI, but don't sell snake oil to people who are actively afraid of it.
So I did the funny and poisoned her dataset to show it does nothing.
A nightshaded LoRA
CHARACTER CREDIT
Character: Blossum / @blossumVT
Official character information and design credits:
My contribution is this LoRA training. No affiliation or endorsement is implied.
THE DESIGN
The main visual features represented in the references are:
• Long pink hair, often arranged into twin tails
• Pink eyes, with flower-shaped pupils in detailed depictions
• Pointed ears
• Branches extending from the head, decorated with cherry blossoms
• Flower ornaments throughout the hair
• Hanging ornaments, including a floral wind chime in some designs
• Pink-and-white clothing with floral details
The head branches are part of her sakura-spirit design. Describe them as flowering branches if the model turns them into plain horns.
She has multiple designs, so not every accessory or outfit needs to appear together.
TRIGGER WORD
blossnimaStart with the trigger, then describe the scene and outfit.
Example:
blossnima, solo, pink kimono, wide sleeves, standing beneath cherry blossoms, looking at viewer, soft daylightOptional supporting tags:
pink hair, long hair, twintails, pink eyes, flower-shaped pupils, pointy ears, flowering branches, cherry blossoms, hair flowersThese are descriptive tags, not additional trigger words. Use them to reinforce features that need help.
AVAILABLE VERSIONS
• 3000 Steps
The earlier checkpoint from this training run.
• 3600 Steps
The same run after another 600 training steps.
Both use blossnima. Compare them with the same prompt, seed, base model, and LoRA strength. More steps don't automatically mean a better result.
STRENGTH
Try 1.0 as an initial setting, then adjust for your checkpoint and prompt.
Lower it if the LoRA overwhelms the style or makes outfit changes difficult. Raise it if Blossum's features aren't coming through clearly.
This is a starting point rather than a tested recommendation for every setup.
OUTFITS & DETAILS
The references include floral kimono-inspired clothing, dresses, and alternate outfits. Describe whichever look you want instead of relying on the trigger to select a particular costume.
The branches, hair ornaments, and flower-shaped pupils are small or intricate features. Check them closely when they matter to the image, especially in full-body compositions.
EXAMPLE PROMPTS
Simple character check:
blossnima, solo, upper body, looking at viewer, gentle smile, pink hair, flowering branches, white backgroundCherry blossom scene:
blossnima, solo, full body, pink kimono, wide sleeves, standing beneath a cherry blossom tree, falling petals, soft daylightCasual outfit:
blossnima, solo, cream cardigan, pink skirt, sitting by a cafe window, holding a teacup, warm afternoon lightNight portrait:
blossnima, solo, portrait, looking back, starry sky, soft moonlight, cherry blossoms in hairCOMPATIBILITY
Character LoRA for Anima. Results depend on the checkpoint, prompt, strength, and other LoRAs in your setup.
Training details are listed under each version.
Description
The 3600-step Blossum checkpoint for Anima.
This is a later checkpoint from the same training run as the 3000-step release, saved after another 600 steps.
QUICK START
Trigger: blossnima
Initial strength to try: 1.0
Example:
blossnima, solo, cream cardigan, pink skirt, sitting by a cafe window, holding a teacup, warm afternoon light
Adjust the strength according to the result. If the LoRA dominates the style or restricts the outfit, lower the weight or compare with the 3000-step checkpoint.
VERSION DETAILS
• Base model family: Anima
• Training steps: 3600
• Network rank: 6
• Format: SafeTensors
• Training software: AI Toolkit Experimental 0.13.23
• File: blossum_anima_000003600.safetensors
More training doesn't guarantee a better result for every prompt. Compare both checkpoints using the same prompt, seed, base model, and strength.
The image-protection experiment behind this training is described in the main model description. This checkpoint alone is not a controlled test of Glaze or Nightshade.
CREDIT
Character: Blossum / @blossumVT
Official character information and design credits:
Independently trained LoRA. No affiliation or endorsement is implied.




