Eve library

ISIS v2 Cinematic Shot Sequence Control-Conditioned Workflow Graph

Guide

Revision 1

On this page

Generated: 2026-08-02T19:12:18.411Z Checklist: MXA-0472 (ISIS2-00670) Asset Family: cinematic-shot-sequence Phase: Phase 4 | Milestone: M3 Launch (2026-07-27 to 2026-10-30)

Objective#

Define a control-conditioned workflow graph for cinematic-shot-sequence that supports pose/depth/segmentation/normal/canny/lineart guidance where applicable while preserving deterministic framing readability and temporal continuity defaults.

Deterministic Defaults#

Setting Value
checkpointModel sd_xl_base_1.0.safetensors
promptTemplateId cinematic-shot-sequence-control-conditioned-v1
styleProfileId cinematic-shot-sequence-style-anchor-v1
colorScript cool cyan moonlight with warm tungsten practical accents
lensProfile cinematic-35mm-dolly
seed 2026113003
width 1920
height 1080
steps 30
cfgScale 6
samplerName dpmpp_2m_sde
scheduler karras
denoise 1
poseControlModel control_v11p_sd15_openpose.safetensors
depthControlModel control_v11f1p_sd15_depth.safetensors
segmentationControlModel control_v11p_sd15_seg.safetensors
normalControlModel control_v11p_sd15_normalbae.safetensors
cannyControlModel control_v11p_sd15_canny.safetensors
lineartControlModel control_v11p_sd15s2_lineart_anime.safetensors
poseStrength 0.62
depthStrength 0.74
segmentationStrength 0.66
normalStrength 0.72
cannyStrength 0.54
lineartStrength 0.58
batchSize 1
durationSeconds 4
frameRate 24
outputFormat mp4
outputChannels 3

Input Contract#

Input Type Required Default
pose_map image no n/a
depth_map image no n/a
segmentation_map image no n/a
normal_map image no n/a
edge_map image no n/a
lineart_map image no n/a
use_pose_control boolean yes false
use_depth_control boolean yes true
use_segmentation_control boolean yes true
use_normal_control boolean yes true
use_canny_control boolean yes true
use_lineart_control boolean yes true
pose_strength float yes 0.62
depth_strength float yes 0.74
segmentation_strength float yes 0.66
normal_strength float yes 0.72
canny_strength float yes 0.54
lineart_strength float yes 0.58
sequence_id string yes SEQ01_SH030
scene_description string yes Hero pivots toward the arriving train while foreground commuters blur into graphic silhouettes.
shot_goal string yes Preserve subject trajectory, continuity axis, and transition timing for editorial handoff.
camera_setup enum yes tracking
camera_motion enum yes dolly-in
composition_rule enum yes leading-lines
subject_blocking string yes Primary subject foreground-right, commuter cluster mid-left, strong perspective rails converging to vanishing point.
environment_context string yes Night station platform with practical sodium lights, reflective ground, and atmospheric haze.
style_preset enum yes cinematic-live-action
continuity_notes string yes Preserve subject eyeline, camera axis, and lighting continuity with adjacent sequence shots.
prompt_template_id string yes cinematic-shot-sequence-control-conditioned-v1
style_profile_id string yes cinematic-shot-sequence-style-anchor-v1
color_script string yes cool cyan moonlight with warm tungsten practical accents
lens_profile string yes cinematic-35mm-dolly
negative_prompt string no blurry, low readability, malformed anatomy, incoherent staging, continuity break, temporal jitter, warped perspective, duplicate subjects, text, watermark, compression artifacts
duration_seconds float yes 4
frame_rate integer yes 24
seed integer yes 2026113003
width integer yes 1920
height integer yes 1080
steps integer yes 30
cfg_scale float yes 6

Graph Nodes#

Node ID Class Role Output Count
n1 LoadImage depth-control-input 2
n2 LoadImage segmentation-control-input 2
n5 LoadImage normal-control-input 2
n3 LoadImage canny-control-input 2
n4 LoadImage lineart-control-input 2
n23 LoadImage pose-control-input 2
n6 CheckpointLoaderSimple model-load 3
n7 ControlNetLoader depth-control-load 1
n8 ControlNetLoader segmentation-control-load 1
n9 ControlNetLoader normal-control-load 1
n10 ControlNetLoader canny-control-load 1
n11 ControlNetLoader lineart-control-load 1
n24 ControlNetLoader pose-control-load 1
n12 CLIPTextEncode positive-conditioning 1
n13 CLIPTextEncode negative-conditioning 1
n25 ControlNetApply pose-control-apply 1
n14 ControlNetApply depth-control-apply 1
n15 ControlNetApply segmentation-control-apply 1
n16 ControlNetApply normal-control-apply 1
n17 ControlNetApply canny-control-apply 1
n18 ControlNetApply lineart-control-apply 1
n19 EmptyLatentImage latent-init 1
n20 KSampler latent-sample 1
n21 VAEDecode image-decode 1
n22 SaveVideo artifact-output 0

Graph Edges#

Edge ID From To
e1 n6:0 n20.model
e2 n6:1 n12.clip
e3 n6:1 n13.clip
e4 n6:2 n21.vae
e4a n12:0 n25.conditioning
e4b n24:0 n25.control_net
e4c n23:0 n25.image
e5 n25:0 n14.conditioning
e6 n7:0 n14.control_net
e7 n1:0 n14.image
e8 n14:0 n15.conditioning
e9 n8:0 n15.control_net
e10 n2:0 n15.image
e11 n15:0 n16.conditioning
e12 n9:0 n16.control_net
e13 n5:0 n16.image
e14 n16:0 n17.conditioning
e15 n10:0 n17.control_net
e16 n3:0 n17.image
e17 n17:0 n18.conditioning
e18 n11:0 n18.control_net
e19 n4:0 n18.image
e20 n18:0 n20.positive
e21 n13:0 n20.negative
e22 n19:0 n20.latent_image
e23 n20:0 n21.samples
e24 n21:0 n22.frames

Topological Node Order#

n1 -> n10 -> n11 -> n19 -> n2 -> n23 -> n24 -> n3 -> n4 -> n5 -> n6 -> n12 -> n13 -> n25 -> n7 -> n14 -> n8 -> n15 -> n9 -> n16 -> n17 -> n18 -> n20 -> n21 -> n22

Research Alignment#

Workflow Path Control Nodes
AnimateDiff with ControlNet libs/isis/workflows/src/workflows/video/animatediff-controlnet.json 2
AI Style Transfer libs/isis/workflows/src/workflows/utility/style-transfer.json 2
Image to Video Animator libs/isis/workflows/src/workflows/video/image-to-video.json 0

Acceptance Criteria#

  • Graph is acyclic and topologically sortable.
  • Pose/depth/segmentation/normal/canny/lineart control paths are explicitly represented through ControlNet loaders and apply nodes.
  • At least one control mode can be enabled per run without breaking deterministic sampler/model defaults.
  • Pose control is explicitly supported as optional where applicable.
  • Output contract aligns to MXA-0469 required output classes.
  • Controlled cinematic composition and temporal continuity remain first-class required inputs.

Traceability#

  • Checklist item: MXA-0472
  • Ticket key: ISIS2-00670
  • Source references:
    • docs/domains/isis/remediation-v2/ISIS_REMEDIATION_V2_TICKETS.json
    • docs/domains/isis/remediation-v2/ISIS_V2_CINEMATIC_SHOT_SEQUENCE_REQUIREMENTS_BRIEF.json
    • docs/domains/isis/remediation-v2/ISIS_V2_CINEMATIC_SHOT_SEQUENCE_BASELINE_WORKFLOW_GRAPH.json
    • docs/domains/isis/remediation-v2/ISIS_V2_CINEMATIC_SHOT_SEQUENCE_IMAGE_CONDITIONED_WORKFLOW_GRAPH.json
    • libs/isis/workflows/src/workflows/video/animatediff-controlnet.json
    • libs/isis/workflows/src/workflows/utility/style-transfer.json
    • libs/isis/workflows/src/workflows/video/image-to-video.json
    • docs/domains/isis/remediation-v2/ISIS_V2_CONTROL_TAXONOMY.json
    • docs/domains/isis/remediation-v2/ISIS_V2_WORKFLOW_AUTHORING_GUIDE.json
    • docs/domains/isis/remediation-v2/ISIS_V2_OUTPUT_PACKAGE_TAXONOMY.json
    • ISIS_TODOS_2.md