Skip to content

retarget-fx-shot.md

Latest commit

 

History

History
132 lines (103 loc) · 5.53 KB

retarget-fx-shot.md

File metadata and controls

132 lines (103 loc) · 5.53 KB
Raw

Skill: retarget-fx-shot (Nuke)

Retarget an existing FX rig from one shot to another. Duplicates a named node group (and its BackdropNode), renames shot-coded nodes, then remaps all file references from the source shot's sequences to their analogues in the target shot.

Invoke

"Retarget the [backdrop/group name] network from [SOURCE_SHOT] to [TARGET_SHOT]."

Or supply the three inputs explicitly:

Input Description Example
SOURCE_NODES List of node names + backdrop label to copy Read1, Grade2, Merge3, PARK003 Rig
SOURCE_DIR Root directory of the source shot's sequences /path/to/raw_videos/PARK003/
TARGET_DIR Root directory of the target shot's sequences /path/to/raw_videos/CWG006/

Phases

1. Copy

  • Verify Nuke connection (ping or get_script_info).
  • Read the position of every source node via get_node_info (knobs xpos, ypos).
  • Find the BackdropNode enclosing the group: it has knobs xpos, ypos, bdwidth, bdheight, and a label matching the group name.
  • Determine bounding box of the group: max_xpos across all nodes + node width (~80).
  • Compute offset_x = bounding box width + a gap of at least 150 units (Nuke's coordinate scale is larger than Houdini's).
  • Duplicate via execute_python: 
    nuke.selectAll()
nuke.invertSelection()
for name in source_node_names:
    nuke.toNode(name).setSelected(True)
nuke.nodeCopy('%clipboard%')
nuke.nodePaste('%clipboard%')
  • Move each new node's xpos by + offset_x.
  • Create a new BackdropNode at the same offset: copy bdwidth, bdheight, label.

2. Rename

  • Scan new node names for the source shot code, case-insensitive.
  • Also catch Nuke's auto-suffix artefacts: Nuke appends a number on paste (e.g. Read_PARK003 → Read_PARK0031). Detect these by the shot code stem.
  • Propose renames: replace source shot code with target shot code.
  • Also update the BackdropNode label knob if it contains the source shot code.
  • Show the rename list and get user confirmation before applying.
  • Apply: node['name'].setValue(new_name) via execute_python or modify_node.

3. Scan

  • Scan all new nodes for file knob values containing the source shot directory.
  • Also check inside Group nodes — recurse via nuke.toNode(group_name).nodes() and scan each child's knobs.
  • Other knobs that may contain file paths: vfield_file, lut, script (on Group nodes), any knob of type File_Knob.
  • Collect: node name | knob name | current value.

4. Match

  • List the source shot directory — collect sequence folder names.
  • List the target shot directory — collect sequence folder names.
  • Match sequences semantically, in this priority order:
    1. Shot plate — keywords: pl01_ref, ref, HD_ref, plate
    2. Clean plate — keywords: cp01, cleanPlate, clean_plate
    3. Roto — keywords: roto
    4. Render passes — match by pass name: Normal, Depth, Alpha, BaseColor, Roughness, Metallic, Specular, Source
  • Use process of elimination: once a source sequence is matched, remove it from the candidate pool for subsequent matches.
  • Flag format changes (.exr ↔ .png) — note these explicitly.
  • Flag unmatched source sequences — leave their parameters unchanged.

Frame expression translation table:

Nuke uses printf-style or hash-style frame tokens, not Houdini's $F syntax.

Source pattern Detected by New pattern
%04d printf 4-digit %04d (usually unchanged)
%06d printf 6-digit %06d (usually unchanged)
#### 4 hashes #### (usually unchanged)
%04d with frame offset expression node upstream Adjust offset value, keep format

Determine which pattern applies by inspecting the first filename in the target sequence folder: count the digits, check for leading zeros, check whether the frame number matches %d directly or requires an offset expression.

If the source used a Nuke frame offset (e.g. Read node frame_mode: offset, frame: -1000) and the target sequence starts at a different frame number, update the frame knob accordingly — do not change the file path pattern.

5. Confirm

  • Present the full mapping table before touching any parameters:
    • Old path → New path
    • Expression/offset change (if any)
    • Format change (if any)
    • Unmatched sequences (listed separately)
  • Do not proceed until the user approves.

6. Remap

  • For each approved mapping, update the file knob via modify_node or execute_python: nuke.toNode(name)['file'].setValue(new_path).
  • If the frame padding changed, also update the format or frame knobs as needed.
  • Apply to every node/knob pair found in the scan.

7. Save

  • Call save_script after all remaps are applied.
  • Report a final summary: N nodes copied, M nodes renamed, P parameters remapped, Q sequences unmatched.

Notes

  • Do not remap unmatched-sequence parameters — leave them pointing at the source shot. The user handles those manually.
  • In Nuke, the BackdropNode label is the primary human-readable identifier for a rig group — always update it to reflect the target shot name.
  • Nuke Read nodes have a colorspace knob — if the source and target shots use different colorspace tags (e.g. different camera LUTs), flag this for the user but do not change it automatically.
  • If source nodes use a TimeOffset or Retime node with frame expressions, those may need adjustment when the target shot has a different frame range — flag but do not auto-fix.