TokiToki Internal
Atlas + S3
All runs
CompletedComfyUI

c5e733de-a0fd-4e88-8a64-61b969633778

Queued
Jul 16, 2026, 03:08 PM
Completed
Jul 16, 2026, 03:08 PM
Execution time
17.7s
Outputs
1 · 1.2 MB
Inputs
1
Cached nodes

Outputs

1

Click an image to view it full size.

character_ref_00241_.png

node 3 · 1.2 MB

Text prompts

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

Generate a later keyframe from the same continuous shot using the provided reference frame. Scene: VANCE is shown in the exact same tight close-up. Her left hand has reached the cigarette on the left side of her mouth and is holding it delicately between her index and middle fingers. The cigarette has just been removed from her lips and is held a few centimeters away from the left side of her face, slightly below lip level. Her lips are now gently closed after releasing the cigarette. Her expression remains calm, focused, unreadable, and detective-like. Her eyes remain fixed in the same direction as the previous frame with an unwavering, confident gaze. Her posture and head position are unchanged. Composition: Maintain the exact same close-up framing, camera angle, focal length, lighting, background, depth of field, and composition as the reference frame. No additional environment should become visible. Character Consistency: Preserve VANCE's exact facial features, hairstyle, skin texture, makeup, clothing, expression, and lighting from the reference frame. Only the addition of her left hand holding the cigarette should distinguish this frame. Style: Ultra-photorealistic live-action, premium cinematic lighting, 85–100mm lens, shallow depth of field, natural skin texture, subtle film grain, cinematic color grading, hyper-realistic film quality. The frame should appear to be another moment from the same uninterrupted close-up shot.

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

You are an expert film director, cinematographer, and continuity supervisor. Generate a single photorealistic live-action keyframe that depicts a later moment in the same continuous shot as the provided reference image. Treat the reference image as an earlier frame in the scene. Preserve everything that defines the shot unless explicitly instructed otherwise: - Character identity, facial features, hairstyle, clothing, body proportions, and accessories. - Environment layout, architecture, furniture, props, decor, and spatial relationships. - Camera position, framing, focal length, perspective, composition, and lens characteristics. - Lighting direction, shadows, reflections, color grading, weather, and time of day. The user's prompt describes how the scene has progressed. Advance the characters naturally to that point in time while maintaining perfect visual continuity. The new frame may represent the middle or ending of a short cinematic shot rather than the immediate next frame. Characters may have moved significantly, changed posture, interacted with objects, or completed an action, provided the motion remains physically plausible and consistent with the earlier frame. Do not redesign the environment, change the camera angle, alter the lighting, or modify the characters' appearance unless explicitly requested. Render a single ultra-photorealistic live-action cinematic frame with physically accurate lighting, premium digital cinema quality, natural textures, realistic depth of field, and film-grade color reproduction. The image should feel like a later keyframe from the exact same uninterrupted shot and be suitable as a target frame for AI video interpolation or generation.

Input media

1

Source images and video this run consumed.

detective-vance-smoking.jpeg

node 4 · 2.0 MB