new node generateVideo from sfmData

[servantftransperfect]

Generate a new node to which will generate a video given a sfmData.

[Copilot]

Pull request overview

Adds a new Meshroom “Video Utils” node intended to generate a video from an SfMData by ordering views via frameId.

Changes:

• Introduces a new GenerateVideo command-line node that reads an SfMData and creates a sequential image set via symlinks.
• Builds an ffmpeg command to encode those frames into an MP4 in the node cache.
• Adds a root .gitignore entry for __pycache__.

Reviewed changes

Copilot reviewed 1 out of 2 changed files in this pull request and generated 6 comments.

File	Description
meshroom/videoUtils/generateVideo.py	New node that loads SfMData views, creates sequential symlinks, and runs ffmpeg to produce a video.
.gitignore	Ignores Python __pycache__ artifacts.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Setting the root logger level inside a node (logging.getLogger().setLevel(...)) can have side effects on unrelated logging in the same process. Prefer configuring a module-specific logger (e.g. logging.getLogger(__name__)) or using Meshroom’s standard logging facilities for nodes if available.

[Copilot]

Parameter naming is inconsistent with the existing nodes in meshroom/videoUtils (framerate is used elsewhere). Consider renaming frameRate to framerate for consistency and to make scripting/graph reuse more predictable.

[Copilot]

The frameRate input is not used: the ffmpeg command hardcodes -framerate 24. Use the node parameter value for -framerate so the UI control actually affects the generated video. Also consider adding -y (as in other videoUtils nodes) to avoid failing on reruns when the output file already exists.

Suggested change

	self.commandLine = f"ffmpeg -framerate 24 -start_number 0 -i %012d{suffix} -vf \"scale=trunc(iw/2)*2:trunc(ih/2)*2\" -c:v libx264 -pix_fmt yuv420p {chunk.node.outputVideo.value}"
	self.commandLine = f"ffmpeg -y -framerate {chunk.node.frameRate.value} -start_number 0 -i %012d{suffix} -vf \"scale=trunc(iw/2)*2:trunc(ih/2)*2\" -c:v libx264 -pix_fmt yuv420p {chunk.node.outputVideo.value}"

[Copilot]

This loop only overwrites links from 0..N-1; if the node cache is reused and a later run has fewer frames than a previous run, stale symlinks beyond pos will remain and ffmpeg will silently include old frames. Clear previously generated links (or generate them in a dedicated subdirectory that you recreate/clean each run) before creating the new sequence.

[Copilot]

If the SfMData contains zero views, suffix stays empty and no links are created, but the node still builds an ffmpeg command that cannot work. Add an explicit check after loading/views extraction to fail fast with a clear error when no views/images are found.

Suggested change

	# fail fast if no views/images were found
	if not viewsPerFrameId:
	logging.error("No views/images found in the input SfMData.")
	raise RuntimeError()

[Copilot]

The error handling here drops important context: it logs a generic message and raises RuntimeError() without any message/path, which makes failures hard to debug. Include the input filepath in the log/exception and raise with a descriptive message (or rethrow a more specific exception) so the UI/logs show actionable info.

Suggested change

	logging.error("Cannot open input")
	raise RuntimeError()
	logging.error("Cannot open input sfmData file: %s", chunk.node.input.value)
	raise RuntimeError(f"Cannot open input sfmData file: {chunk.node.input.value}")