Recommendations (v1.0 - Anima)
Prompt: phonefocus, smartphone, pov, hand holding phone, phone in foreground. Weight: 0.8 (0.6–1.2).
Settings: 30 steps, CFG 4, er_sde / simple, 896×1152 or 1024×1024. Add question mark, speech bubble to the negative. More tips are in the version notes.
A hand from outside the frame holds a smartphone up to the character, POV shot. She reacts: surprise, blush, or a hypnotized stare with spiral eyes. Works with named and original characters.
Description
Anima Base v1.0 version, trained with the Civitai trainer on 36 images.
Prompt: phonefocus, smartphone, pov, hand holding phone, phone in foreground. The trigger alone does not work: the phone turns into a random object. The four helper tags hold the concept, write all of them.
Weight: 0.8 (0.6–1.2). At 1.0 a collage background of blue panels shows up more often. Base Anima draws the girl inside the phone screen instead (weight 0 in the comparison).
Tips:
Hypnosis:
spiral eyes, hypnotized, blank stareorempty eyes, hypnotizedinstead of an emotion. Spiral eyes appear most of the time.By default the phone is at chest level near the frame edge;
close-up, portraitbrings it closer to the face.The hand is often dark-skinned, as in the dataset. Describe it if you need another one.
Named characters work (Genshin Impact in the previews).
Add
question mark, speech bubbleto the negative against the "?" marks from the dataset. They still show up sometimes.
Limitations: no selfies (the character holding the phone herself), and the screen always faces the character, never the viewer.
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, question mark, speech bubble (add nsfw, nude, topless for SFW).
DiT-only LoRA (no text encoder weights), dim 32 / alpha 32, 1500 steps.






