Batch Gartner

A local render queue manager for Foundry Nuke.

Queue Nuke scripts, send Write nodes straight from Nuke, and render them one after another (or several at once) on your own workstation, with automatic resume, crash recovery and a live frame preview. No render farm needed.

---

Why

Rendering from the Nuke GUI blocks your session, and queuing several comps overnight by hand is fragile. Batch Gartner renders in the background with Nuke's command line, so you can keep working. It keeps track of every frame and deals with the usual problems:

• You keep editing a script while it waits in the queue → it renders the version you queued (script snapshots).
• A render crashes or you stop it halfway → it resumes from the frames already on disk.
• An output file is briefly locked (a viewer, a cloud-sync app) → the frame is retried instead of failing the whole job.
• You re-render into a folder that already has frames → it asks before overwriting.

---

Features

Queue

• Add .nk scripts with + ADD, drag & drop, a whole folder, or straight from Nuke
• One job per Write node, with priority, dependencies ("render after job X"), scheduled start and chunking
• Move jobs up and down (Ctrl+Up / Ctrl+Down), suspend, lock, duplicate, multi-edit
• Lock protects approved renders from edits, removal and re-rendering
• Undo for queue edits, and you can save and load queues

Rendering

• Renders with Nuke's command line (nuke -i -x / -t), so the GUI stays free
• Script snapshots: every job renders the script as it was when queued. If you saved changes since, START asks whether to use the queued or the latest version.
• Automatic resume: failed or stopped jobs render only the missing frames
• Overwrite protection: Render Missing Only / Overwrite All / Cancel
• Half-written frames are deleted after STOP, skip or timeout
• Test frame and First / Middle / Last renders, plus Render Missing Frames
• Per-frame retry for locked files, per-task retries and timeouts
• Render Power (process priority) and Render While I Work mode
• Resource protection: waits for free RAM or CPU before starting a job, up to a maximum wait you set
• Simultaneous renders (optional, for big machines): up to 4 Nuke processes at once, each with an equal share of threads (-m) and cache (-c)
• Pause after current job, Skip, Stop, and shutdown or sleep when the queue finishes (only after a queue that really completed)

Monitoring

• Live progress, current frame, ETA and finish time, and render duration
• Built-in preview of finished frames with a scrub slider (EXR, TIFF, PNG, JPG…). It never locks a file Nuke is writing.
• Sequence check for missing and zero-byte frames
• System monitor: CPU, RAM, GPU/VRAM (NVIDIA), and the Nuke process's memory
• Tasks, Log, Details (job, output, performance, errors) and History tabs
• Clear error types (Write error / file locked, missing file, out of memory, license…)
• Desktop notifications

Nuke integration

• Nuke › Batch Gartner menu:
  • Send Selected Write(s) (Ctrl+Alt+B)
  • Send All Writes
  • Send Selected Write(s) – Current Frame
  • Open Batch Gartner
• Starts Batch Gartner automatically if it isn't running, without freezing Nuke
• Create Read in Nuke: right-click a finished job to drop a Read node for the render next to its Write node, in the Nuke session that has that script open

Reliability

• Crash recovery: restores the real render state after an unexpected quit
• Settings are saved atomically, and only one instance runs at a time
• Startup errors are shown in a window and written to a log file

---

Requirements

Nuke	Nuke / NukeX with command-line rendering (developed with Nuke 15.x)
Python	A separate Python 3.9+ installation (not Nuke's own Python)
PySide6	python -m pip install PySide6
psutil (recommended)	CPU/RAM monitor, Render Power, resource protection: python -m pip install psutil
OpenEXR + numpy (optional)	EXR preview: python -m pip install OpenEXR numpy

Developed and tested mainly on Windows. macOS and Linux are supported by the code (Nuke detection, file opening), but they are less tested.

---

Installation

1. Download or clone this repository and keep the folder somewhere permanent.

   git clone https://github.com/PedroGartner/BatchGartner.git
   cd BatchGartner/BatchGartner

2. Install the Python packages with the Python you will use to run Batch Gartner:

   python -m pip install PySide6 psutil
   # optional, for EXR preview:
   python -m pip install OpenEXR numpy

3. Install the Nuke menu (run once, and again after every update):

   python install_batch_gartner_integration.py

   This copies BatchGartner_NukeIntegration.py into ~/.nuke, adds the menu to ~/.nuke/menu.py, and remembers where BatchGartner.py is.

4. Restart Nuke. You'll find the Batch Gartner menu in Nuke's menu bar.

5. Run Batch Gartner, from Nuke (Batch Gartner › Open Batch Gartner) or directly:

   python BatchGartner.py

6. In the toolbar, pick your Nuke version from the list, or click ... (Tools › Set Nuke Executable) to browse for it.

Manual install: if you'd rather not run the installer, copy BatchGartner_NukeIntegration.py into your .nuke folder (or anywhere on NUKE_PATH) and paste the contents of menu_snippet.py into your menu.py.

---

Quick start

1. In Nuke, select one or more Write nodes and press Ctrl+Alt+B. The script is saved and the jobs appear in Batch Gartner.
2. Choose a Render Power in the dialog (Normal is fine).
3. Press START.
4. Watch progress in the queue and the Preview, and read Nuke's output in the LOG tab.
5. When the job finishes, right-click it › Create Read in Nuke to bring the render back into your comp.

Useful workflows

You want to…	Do this
Check a look before a long render	Right-click › Render Test Frame… or More › Render First / Middle / Last
Continue a stopped or failed render	Select it › Retry › START (only missing frames are rendered)
Render the latest save instead of the queued snapshot	Queue › Refresh Script (use latest save)
Fill gaps in a sequence	Right-click › Render Missing Frames
Protect an approved render	Right-click › More › Lock / Unlock
Render overnight and switch off the PC	Post Render › Shutdown when queue finishes

---

How it works

• Rendering: each job runs as a separate Nuke command-line process. With snapshots on (the default), a small helper script opens the snapshot inside Nuke and sets root.name back to the original script path. That way [value root.name] expressions, relative paths and versioned output names behave exactly as in the GUI.
• Snapshots: a copy of the .nk is taken when a job is queued. Unused snapshots are cleaned up automatically at startup.
• Nuke → Batch Gartner: the Nuke menu sends jobs over a localhost socket (default port 54321).
• Batch Gartner → Nuke: each open Nuke GUI listens on the next free port (54322–54331) for Create Read in Nuke. It only runs in the GUI, never inside render processes.

All network traffic is local only (127.0.0.1).

---

Preferences

Tools › Preferences

Section	Settings
Appearance	Font, size, row height, status colours
Queue	Restore queue on launch, confirm removal, newer-version warning, default chunk size and retries
Rendering	Default Render Power, Render While I Work, script snapshots, auto-resume, overwrite confirmation
Performance	Monitor refresh, CPU/RAM protection, max wait for resources, simultaneous renders
Notifications	Desktop notifications on complete or fail
Post Render	Verify output, open output folder after queue
Interface	Remember window, panel and column layout
Advanced	Receiver port, recovery snapshots

---

Files and locations

Path	What
BatchGartner.py	The application
BatchGartner_NukeIntegration.py	The Nuke menu (installed into ~/.nuke)
install_batch_gartner_integration.py	Installer for the Nuke menu
menu_snippet.py	Manual alternative to the installer
CHANGELOG.md	Version history
~/.batch_gartner/settings.json	Settings and queue
~/.batch_gartner/logs/	Render logs per script, plus batch_gartner_errors.log
~/.batch_gartner/snapshots/	Script snapshots
~/.batch_gartner/recovery.json	Crash-recovery snapshot

On Windows ~ is C:\Users\<you>. Paste %USERPROFILE% into Explorer to get there.

---

Troubleshooting

Batch Gartner doesn't open. Run python BatchGartner.py from a terminal to see the error, or check ~/.batch_gartner/logs/batch_gartner_errors.log. If it's already running (possibly hidden), starting it again brings the existing window to the front. Otherwise, end the leftover python.exe in Task Manager.

"Could not find an external Python installation with PySide6" (from Nuke). Install PySide6 into a normal Python (python -m pip install PySide6), or set the environment variable BATCH_GARTNER_PYTHON to that python.exe.

A frame fails with "Cannot open". The output file is locked, usually by an image viewer showing the frame being written, or a sync app (MEGA, Dropbox, OneDrive). Batch Gartner retries the frame for about 18 s. Use the built-in preview instead of an external viewer while rendering.

Create Read in Nuke says no Nuke was found. Run install_batch_gartner_integration.py again and restart Nuke.

No EXR preview. Install OpenEXR and numpy into the Python that runs Batch Gartner.

The queue waits with "Waiting (resources)". Free RAM or CPU is below the job's protection threshold. Lower it under Render Power / Resources, or set a shorter Max wait in Preferences.

---

Known limitations

• The sequence check, resume and preview need a job with a single Write node and a plain output path. Paths built from TCL expressions (e.g. [value root.name]) are rendered correctly but can't be checked on disk.
• Snapshot renders go through Nuke's -t mode and render frame by frame.
• CPU % readings in the monitor are approximate.
• GPU monitoring requires NVIDIA (nvidia-smi).

---

Author

Pedro Gartner