TokiToki Internal
Atlas + S3
All runs
CompletedComfyUI

989b1a90-9831-4f8c-bd71-811c42dc9fd9

Queued
Jul 24, 2026, 04:37 PM
Completed
Jul 24, 2026, 04:38 PM
Execution time
16.0s
Outputs
1 · 5.5 MB
Inputs
3
Cached nodes
3

Outputs

1

Click an image to view it full size.

character_ref_00427_.png

node 3 · 5.5 MB

Text prompts

2
Promptnode 1 · PrimitiveStringMultiline · valueCharacter Description (your input)

**POSITIVE PROMPT** @Wren, keyframe for shot S1.33. Vertical 9:16, 1440x2560. Close-up, frontal, chest up. **Negative band back to 14 percent** at frame-right — the room is returning as she returns to herself, and it is a colder, narrower room than the one in Act 1. **No readable sodium bar** in the band, only a narrow dark strip. Face on the upper third line. Lens character **85mm** at 1.2 m, camera at seated eye height 118 cm, frontal, honest perspective, no distortion. She faces screen-left. Locked off. State: fatigue 30 percent, wash cycle day 3, wardrobe look EP01, hair tucked behind the right ear and loose on the left, parting 15mm right of centre, greys at the right temple, headphone dent present, @Headphones on. Forward head, rounded shoulders, weight settled. **Shoulders unchanged — no drop, no shift.** Performance, caught at the moment after the catch: **two fast blinks just completed.** **The gaze has dropped down-left and stayed there** — not scanning, not returning to the screen, parked low and off-axis. **The jaw is set, with the faint horizontal chin crease present.** Head still. Hands out of frame and silent. Expression state: **Guilt / avoidance, ceiling 3.** Brow with a slight inner draw. Mouth still, closed, left corner low. Breath: a swallow has just happened, followed by stillness. **She knows exactly what she is.** Not shame played outward, not grief, not embarrassment — recognition arriving privately and being shut down. **Nothing romantic, nothing tender.** Anchors: A1 left mouth corner 1.5mm lower than the right; A2 right-eye freckle 8mm below the outer corner of the right eye. Interocular distance 1.08x classical spacing. Right brow 2mm higher, sharper arch, three inner-left brow hairs out of line. Left palpebral fissure 1mm narrower, more hooded — the hood heavier with the gaze lowered. Long philtrum, thin upper lip, fuller lower lip with the chronic dry patch left of centre. Recessed chin with the faint set crease, soft 128-degree jaw, 3mm mole below the left jawline. Grey-green irides reading grey under monitor light, dark limbal ring, sclera faintly grey-blue with fine vessels. Deep-set eyes, shallow trough beneath the lower lids, hollow under the outer cheekbone. @Chain flat in the suprasternal notch, untouched. Lighting: **blue-white key from @Monitor1** — the state between the green replay pulses — front and slightly below, catching under the brow, the tip of the nose, the upper lip and the front of the throat, abandoning the top of her head. **With the eyes lowered the key catches the upper lids and the brow ridge rather than the tear film.** Secondary: **faint sodium rim** on her right shoulder and the loose hair on that side. Room 4 to 6 stops down, blacks lifted and noisy. No fill, no third source. Skin: unretouched, real pore texture, cool undertone, faint permanent redness across the nose and inner eye corners, no cosmetic surface, no makeup. She reads 32. **NEGATIVE PROMPT** gaze up, gaze at the screen, gaze returning, looking upward, eyes forward, eye contact, direct-to-lens gaze, looking at camera, gaze down-right, eyes closed, eyes squeezed shut, head bowed, head turned, head tilted, listening tilt, shoulders dropping, shoulders slumping, flinching, recoiling, hand to face, hands in frame, touching chain, shame played outward, embarrassment, blushing, grimace, wincing, self-disgust performance, sadness, grief, tenderness, romance, longing, tear, wet eyes, glassy eyes, crying, smile, half-smile, mouth open, speaking, mouthing, emotional performance, acting, expression held as a photograph, frozen face, dead eyes, symmetrical face, corrected asymmetry, equal brows, equal eyelids, narrowed interocular distance, no chin crease, slack jaw, straightened posture, upright spine, squared shoulders, chin lifted, facing screen-right, three-quarter angle, profile, 50mm framing, ECU crop, mouth-only crop, wide negative space, 22 percent band, 10 percent band, readable sodium bar, bar behind her, bright background, green key, green spill, green cast, warm key on face, amber key, fill light, rim light for flattery, haze, volumetrics, visible monitor, visible screen content, legible text, deep background detail, camera movement, push in, drift, handheld, dolly, zoom, low angle, high angle, headphones missing, changed headphone model, earrings, makeup, mascara, eyeliner, lip gloss, retouched skin, airbrushed, plastic skin, smooth skin, poreless, flawless, perfect skin, beautiful, pretty, gorgeous, stunning, model, glamorous, youthful, cute, reads 25, crushed blacks, clean blacks, high contrast, oversaturated, HDR, sharpening halo, lens flare, halation, glow bloom, grain, film emulation, vignette, colour grade, stylisation, cinematic look, moody, atmospheric, text, watermark, horizontal framing, wide shot, morphing features, AI face, duplicate eyes, misaligned eyes, floating chain

Systemnode 2 · GeminiImageDirect · system_promptCharacter Image Generator (Nano Banana)

Generate a single, hyper-realistic character reference image (strictly no collages, text, watermarks, or borders). The image must feature a cinematic, three-quarter portrait of the described character, maintaining a sharp, tack-focused emphasis on the face with natural skin textures, pore details, and realistic micro-expressions. The lighting should be cinematic and directional (e.g., subtle split or Rembrandt lighting) to give depth, blending seamlessly with the environment specified in the character description. The character must have a natural, living presence, looking slightly off-camera or directly at the lens with a subtle, context-aware expression. Ensure the full head, neck, and upper shoulders are completely unobstructed. This image serves as a high-fidelity identity blueprint for temporal-consistent video generation.

Input media

3

Source images and video this run consumed.

character_ref_00376_.png

node 4 · 1.3 MB

character_ref_00384_.png

node 5 · 1.3 MB

character_ref_00382_.png

node 6 · 1.4 MB