How to incorporate sequence names and shot names within naming and pattern browsing?

[leovfx]

Hi. I work in commercials. We name our shots 0010, 0020, 0030, etc and rely heavily on tokens for pattern browsing in and writing out to and from named folders.

We’re about to undertake a bit more long form and are thinking about changing our pipeline to incorporate sequence identifiers within the naming of clips and directories.

The idea is to have shots live on the server in sequence directories and within those shot directories, then have the media itself incorporate both sequence and shot name.

so that might look something like this as a file path:

vfx/shots/01/0010/comp/v001/01_0010_comp_v001.exr

I’ve noticed in the past that when you use the “sequence” token for pattern browsing, it’s treated as a custom token and the sequence name is filed away in the metadata and can’t really be recalled anywhere within the software.

This would primarily be a problem when trying to write out renders from batch into the appropriate sequence directory on the server.

I’m interested in how people who are working on longform regularly have solved this problem.

Any help appreciated. Thanks

[cnoellert]

Provided you publish out your conformed shots from a sequence you’ll have sequence information baked into your batch write nodes.

This is basically how we work in commercials. First conform the sequence of shots (or assemble them in a single sequence if just shots. Generate shot sequence. Publish the sequence—in our case renamed to the prefix of the shots—to our project server, transcoding material to exrs and including options for batch files and openclips.

All batch write nodes have the sequence info baked in the path and resulting published sequence is linked to those open clips and published media.

Or am I misunderstanding?

[leovfx]

@cnoellert

Publish the sequence—in our case renamed to the prefix of the shots

Thanks for your reply.

…i can’t see a sequence token in the sequence publish tokens list.

[cnoellert]

It’s defined by the sequence name you are publishing.

So if memory serves it’s the name token. Sorry not in front of the machine right now. It’s quite easy to test since the paths resolve in the export menu.

It’s is an inconsistency though. Plus you have segment name and shot name to confuse stuff further.

[cnoellert]

If it helps:

<?xml version="1.0"?>
<preset version="14">
	<type>sequence</type>
	<comment>EXR shot pub 230711

</comment>
	<sequence>
		<fileType>NONE</fileType>
		<namePattern></namePattern>
		<composition>&lt;name&gt;</composition>
		<includeVideo>True</includeVideo>
		<exportVideo>True</exportVideo>
		<videoMedia>
			<mediaFileType>image</mediaFileType>
			<commit>Original</commit>
			<flatten>NoChange</flatten>
			<exportHandles>True</exportHandles>
			<nbHandles>1000</nbHandles>
		</videoMedia>
		<includeAudio>True</includeAudio>
		<exportAudio>False</exportAudio>
		<audioMedia>
			<mediaFileType>audio</mediaFileType>
			<commit>Original</commit>
			<flatten>NoChange</flatten>
			<exportHandles>True</exportHandles>
			<nbHandles>10</nbHandles>
		</audioMedia>
	</sequence>
	<video>
		<fileType>OpenEXR</fileType>
		<codec>1185912</codec>
		<codecProfile></codecProfile>
		<namePattern>_04_shows/&lt;name&gt;/&lt;shot name&gt;/FLM/images/plates/&lt;segment name&gt;_rec709_&lt;version name&gt;/&lt;segment name&gt;_rec709_&lt;version name&gt;</namePattern>
		<compressionLevel>45</compressionLevel>
		<includeAlpha>False</includeAlpha>
		<overwriteWithVersions>False</overwriteWithVersions>
		<resize>
			<resizeType>fit</resizeType>
			<resizeFilter>lanczos</resizeFilter>
			<width>0</width>
			<height>0</height>
			<bitsPerChannel>16</bitsPerChannel>
			<numChannels>3</numChannels>
			<floatingPoint>True</floatingPoint>
			<bigEndian>False</bigEndian>
			<pixelRatio>1</pixelRatio>
			<scanFormat>P</scanFormat>
			<ResSelectorMode>3</ResSelectorMode>
			<ResAdaptativeMode>0</ResAdaptativeMode>
			<ResScalingMode>1</ResScalingMode>
			<ResScalingValue>50</ResScalingValue>
		</resize>
	</video>
	<name>
		<framePadding>4</framePadding>
		<startFrame>1001</startFrame>
		<frameIndex>3</frameIndex>
	</name>
	<createOpenClip>
		<namePattern>_04_shows/&lt;name&gt;/&lt;shot name&gt;/FLM/clip/sources/&lt;shot name&gt;_L&lt;track##&gt;</namePattern>
		<version>
			<index>0</index>
			<padding>2</padding>
			<name>v&lt;version&gt;</name>
		</version>
		<batchSetup>
			<namePattern>_04_shows/&lt;name&gt;/&lt;shot name&gt;/FLM/batch/&lt;shot name&gt;_&lt;version name&gt;</namePattern>
			<exportNamePattern>_04_shows/&lt;name&gt;/&lt;shot name&gt;/FLM/clip/&lt;shot name&gt;</exportNamePattern>
			<outputPathPattern>_04_shows/&lt;name&gt;/&lt;shot name&gt;/FLM/images/renders/&lt;shot name&gt;_FLM_&lt;version name&gt;/&lt;shot name&gt;_FLM_&lt;version name&gt;.</outputPathPattern>
			<batchGroup>False</batchGroup>
			<batchInputsType>1</batchInputsType>
			<batchGroupPath>/Desktop</batchGroupPath>
		</batchSetup>
	</createOpenClip>
	<reImport>
		<namePattern></namePattern>
	</reImport>
</preset>

[leovfx]

OK. Right. This can work so long as I’m the person publishing the edit and creating batches for every single shot at that moment.

I’m still not sure what other artists do in the scenario where they’re picking up a single shot. (Aside from having to wire a batch group containing the write node from the lead Flame)

Ideally there would be a way to tokenise the path in a write node so it can automatically point back to the correct sequence directory.

[cnoellert]

Yeah, but it is. The <name> token (which is actually your sequence name in this case) is actually hard coded into the write node of each batch written at publish. It’s not optimum in the way that a batch as a <shot_name> token but it’s still written into the published write nodes so the artist never has to think about it and you don’t have to do anything.

In terms of passing shots to another artist yes you could put all of the shots in a shared library and let them copy from there. But that’s the least smart way to do it. After you publish with open clips and batch setups, the open clips are tied to the batches. Sooooo and artist can:

1. Load an open clip and promote the open clip to a batch group.
   

2. Load an open clips and version it in the timeline.
   

3. Load the batch setup directly in an empty batch.
   

All three of the options are just using standard Flame publishing. The machine that I’m on recording these gifs has never touched this project before so all metadata is being loaded on the fly. All pathing remains intact no matter which machine the sequence was published on. This does not require NIM or Shotgrid or whatever to function.

[kirk]

This is a good solution unless you need to keep version info or any other data in your sequence names.

We have a feature request for an enhancement to tokens that would let you, say, remove characters from the return of a token, so a shot name could be parsed to create a sequence folder, but I’ve been told this is a far-reaching request.

[cnoellert]

@Kirk that’s why I would advise pubbing the shot_sequence but renaming to your shot prefix.

So if your sequence is Fruit_30sec_v01 and your shot pattern is like

fruit_010
fruit_020
fruit_030
fruit_040

Rather than exporting your sequence Fruit_30sec_v01 instead rename your connected shot sequence to fruit and pub that to shots folders.

Then you can flatten publish your versioned edits with link inside the fruit sequence folder. Shit’s super fast then and always attached to the latest OpenClips versions.

So on the filesystem level you’ll have

fruit/
 sequences/
  Fruit_30sec_v01/(linked flattened img sequence of v01)
  Fruit_30sec_v02/(linked flattened img sequence of v02)
 fruit_010/
 fruit_020/
 fruit_030/
 fruit_040/

Or am I misunderstanding the use case?

[kirk]

That is indeed a clever solution. Certainly worth investigating.

We don’t use shot sequences at the moment for all kinds of reasons; lots of edit changes, lots of sequences being updated all the time, multiple resolutions and colorspaces in the same sequence, in house tools that automate publishing, and the fact that we don’t generally finish conforms out of flame, but perhaps they’re worth another look. If only to get around this particular pain point.

[cnoellert]

I guess what I’m getting at is that a single shot is an atomic component… sequences are just playlists of those pieces. They get rev’d and more shots get added but they are still just higher level component of an ordered group of elements that get’s shuffled around.

Seems to me you leave the shots in place and order the revs of the playlist around them or?

[kirk]

The shots frequently change in duration, but we only change the names/numbers under very specific circumstances and it has to basically go in front of a tribunal. This has tended to confound a lot of the built in publishing tools, and it means, for example, that we put the name of the edit revision into the plate name that goes in the footage directory for each shot. So we’re already using the name token, just for something else.

[cnoellert]

But why not using the media versioning system in the publishing tools themselves rather than packing it into the name?

In the output media you can add a version number as part of the clip name…

Then in clip options you can affect the version number used in the pub. Defualts to 0 but you can consistently update it and it will jam that information into the source OpenClips as well.

Food for thought I guess.

[kirk]

Food for thought indeed. You are a generous man, Chris. I owe you a beverage or snack of your choice at some point.

[johnt]

@cnoellert with this type of workflow, how does it handle half a dozen edits that share shots?

Would you just make a c-mode sequence and have the shots live under that on the server?

[philm]

connected conform

[johnt]

Yes. C-mode kind of equals shots sequence. Just wanted to be precise.

[leovfx]

@kirk

We have a feature request for an enhancement to tokens that would let you, say, remove characters from the return of a token, so a shot name could be parsed to create a sequence folder, but I’ve been told this is a far-reaching request.

This would be a perfect solution - of course they’re not working on it.

[knhn]

@kirk

This is a good solution unless you need to keep version info or any other data in your sequence names.

We have a feature request for an enhancement to tokens that would let you, say, remove characters from the return of a token, so a shot name could be parsed to create a sequence folder, but I’ve been told this is a far-reaching request.

If I understand your request, if helpful, here is an example of making your own custom tokens in python. Its just a crude thing, to be professional, I think it would need a pyside UI to preview the final extrapolated names before actually publishing so you do not make a mess of things.

custom_export_tokens.py (3.0 KB)

[leovfx]

The request (at least my initial request) would be to be able to generate a write node based on the name of a clip in batch.

So if the clip name was 01_0010_plate_v001

the resulting write node path would be:

vfx/shots/01/0010/comp/v001/01_0010_comp_v001.exr

so it’s obtaining the sequence and name directories by breaking up the plate name.