TokiToki Internal
Atlas + S3
All runs
ErrorComfyUI

dd7d47a0-c3ab-4a40-9c25-9b99af349a88

Queued
Aug 28, 2026, 06:57 AM
Completed
Aug 28, 2026, 06:58 AM
Execution time
1.3s
Outputs
0 · 0 B
Inputs
2
Cached nodes

Failure

/home/ec2-user/ComfyUI-031/custom_nodes/ComfyUI-Grok-Imagine-Video.xai_client.XAIImagineErrorin CuiMaoGrokReferenceToVideo (node 2)

Prompt references <IMAGE_4> but only 3 reference image(s) were supplied. Images are numbered from 1 in the order they are given (the reference_images input first, then each line of reference_image_paths_or_urls).

Stack trace
  File "/home/ec2-user/ComfyUI-031/execution.py", line 545, in execute
    output_data, output_ui, has_subgraph, has_pending_tasks = await get_output_data(prompt_id, unique_id, obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
                                                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ec2-user/ComfyUI-031/execution.py", line 344, in get_output_data
    return_values = await _async_map_node_over_list(prompt_id, unique_id, obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb, v3_data=v3_data)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ec2-user/ComfyUI-031/execution.py", line 318, in _async_map_node_over_list
    await process_inputs(input_dict, i)
  File "/home/ec2-user/ComfyUI-031/execution.py", line 306, in process_inputs
    result = f(**inputs)
             ^^^^^^^^^^^
  File "/home/ec2-user/ComfyUI-031/custom_nodes/ComfyUI-Grok-Imagine-Video/nodes.py", line 203, in run
    _validate_prompt_image_tags(prompt, len(refs))
  File "/home/ec2-user/ComfyUI-031/custom_nodes/ComfyUI-Grok-Imagine-Video/nodes.py", line 414, in _validate_prompt_image_tags
    raise XAIImagineError(

Outputs

0

The run failed before producing any output.

Text prompts

1
Promptnode 2 · CuiMaoGrokReferenceToVideo · promptCuiMao Grok Reference to Video

SCENE CONTEXT Mid-battle, inside an active war zone. <IMAGE_2> gives his order to the messenger standing off-frame before him — urgent but in complete control, a field commander's natural voice under battle pressure. Far behind him the war rages: silhouetted armies clashing in the haze, smoke columns, pulsing fires. Two shots, one hard cut. No structures anywhere. ACTIVE REFERENCES <IMAGE_2>: broad-built warrior prince, sleeveless engraved steel cuirass with lion-motif chest emblem, dark curly shoulder-length hair, short beard, small gold hoop earring, dried blood streak on cheek, faded scars, deep red dhoti with gold trim. Face locked to this sheet, 100% match. Voice: deep, resonant, natural commanding male voice — urgent but controlled, the way a real general speaks under pressure: quick natural rhythm, clear flow, crisp clear English pronunciation, no robotic pacing. <IMAGE_3>: color and tone reference — warm sepia-golden epic war grade locked for every shot. <IMAGE_4>: enemy tribal warriors — ONLY far-distance fighting silhouettes, never near camera. LOCATION MAP Foreground: smoke sheets and glowing ember streams drifting between camera and subject, a burning ground fire with tall flames near frame edge throwing warm flicker onto his armor. Midground: churned earth, debris, smoke columns bending in gusts. Background: FAR — the raging war: <IMAGE_4> warriors and allied soldiers clashing as dozens of small silhouettes across the horizon, far fires pulsing, rolling dust walls. Smoke 50% everywhere. No structures. FIRST FRAME / BLOCKING <IMAGE_2> chest-up, grounded stance, facing the unseen messenger just off-frame left of camera, his eyes already locked on the man, jaw set, battle alive in the haze behind him. FORMAT MODE Timed multishot, one HARD CUT. 0.0s–5.6s — SHOT 1, MCU frontal, the full order — natural urgent command. 5.6s HARD CUT 5.6s–7.5s — SHOT 2, CU, his locked determined face. OPTICS 80mm IMAX-style portrait look — tight compression, monumental clarity, shallow depth, subject razor-sharp against dissolved war haze. SHOT 1: FOV LOCK 29°. SHOT 2: FOV LOCK 18°. No drift mid-segment. LENS CHECK closing each shot. CAMERA Handheld, 1–2cm war-zone tremor, weighted corrections, faint concussion shudders on distant impacts, smoke and embers crossing the lens. SHOT 1: frontal MCU, slow push-in through the order. SHOT 2: close, settling on the eyes, embers floating past. ACTION SHOT 1 (0.0–5.6s): <IMAGE_2> gives the order — URGENT BUT NATURAL: the voice of a real field general under pressure — quick but unforced rhythm, natural spoken-English flow with command weight, a natural breath after "Rajmata", slight acceleration through the middle phrase, then the final vow landing slower and heavier — total control, zero panic, zero robotic evenness. Crisp clear pronunciation, the names Rajmata, Baahubali and Inkoshi spoken clearly exactly as written, accurate natural lip-sync: "Tell the Rajmata, Baahubali is chasing Inkoshi. I will protect the people." His body speaks with him naturally: a quick loose point toward the horizon around "chasing" — real gesture on instinct, not choreography — the hand dropping and closing into a fist by "protect the people"; eyes alive with quick saccades on the unseen messenger, one flick to the distance mid-line and back; brows settling down on the vow; one hard breath as the line lands. Embers stream past; fire flicker plays on his cuirass. SHOT 2 (5.6–7.5s): tight on his face — eyes locked toward the far pursuit line, nostrils flaring once, jaw grinding, one slow heavy blink, eyes reopening fiercer — determination absolute. Embers drift past catching light. Hold. End. PERFORMANCE The delivery is the key — URGENT yet NATURAL: a man who gives orders every day of his life, under real pressure but in complete command; authentic natural speech rhythm with a general's weight, small accelerations and the heavier final vow — never rushed into mumbling, never slowed into recitation; the gesture and fist arriving on natural instinct, loosely tied to the words; living eyes — fast saccades, real moisture and fire-glow catch-lights, the mid-line distance-flick, the heavy final blink. Pore-level skin, sweat through dust, ember light in eye moisture. PHYSICS The point and fist with natural loose arm timing; smoke on gusting wind, embers on heat currents, tall flames flickering true; ground-shudder pulses on distant impacts; human eye mechanics — true saccade speed, weighted blinks; no face morphing, no identity drift. LIGHTING Locked to <IMAGE_3> — warm sepia-golden war tone, harsh sun through smoke from frame-left, hot sweat speculars, live fire-glow accents on face and armor edges, embers as bright points, 50% haze. AUDIO No BGM, no music, no score — ONLY real production sound: distant battle roar under everything, close fire crackle, gusting wind, ember hiss, ground-thud concussions; his urgent natural command with real outdoor air: "Tell the Rajmata, Baahubali is chasing Inkoshi. I will protect the people."; his hard breath after. No stings, no whooshes. STYLE Photoreal, 80mm IMAX-scale clarity, grounded war cinema, warm epic grade matched to <IMAGE_3>, fine film grain, heavy living atmosphere, 16:9. OUTPUT SETTINGS Real-time speed, no slow-motion. POSITIVE LOCKS Face locked to <IMAGE_2>'s sheet every frame, zero morphing; grade locked to <IMAGE_3>; <IMAGE_4> only far silhouettes; smoke, embers, ground fire and distant fighting in EVERY shot; no structures; the messenger stays off-frame; the only spoken words: "Tell the Rajmata, Baahubali is chasing Inkoshi. I will protect the people." — names Rajmata, Baahubali and Inkoshi pronounced clearly as written, accurate lip-sync, natural urgent field-commander delivery.

Input media

2

Source images and video this run consumed.

bah2.png

node 1 · 17 MB

ttt.png

node 7 · 5.9 MB