Asset tracking across a game production from concepting through final model delivery

A game production runs parallel asset pipelines: characters, environments, vehicles, each one through multiple passes from concepting to final delivery. The reference for a single character flows through concepting, rigging passes, animation studies, and final texture approval, and it needs to be findable by anyone working on that asset at any stage—always at the exact level of detail they need today, not what was relevant last month.

A reference system built around folders and tags handles this because references stay where they live, tags layer on top without reorganizing, and a single image can earn as many tags as it needs. A canvas built at the start of each asset pass becomes the version history, and archiving old work does not mean deleting it—it means tagging it so the active library stays fast.

This is what a production day looks like.

8:00 am — Check overnight reviews and update armor references

The character rig passed modeling review yesterday but armor needs one more pass. The armor modeler sent reference feedback: the shoulder guards are too thin for a warrior class. Check what references they were working from.

Open the saved view from yesterday's modeling pass:

tag:character-warrior tag:armor rating:>=3

Forty-eight reference images pulled from across the archive. Most of them are too light—shoulder armor that works on a rogue does not read on a warrior silhouette. The modeler was right.

Pull up new reference material:

type:image tag:armor tag:heavy-metal rating:>=2 in:/Volumes/Reference-Archive

One hundred and twelve images. Start a new canvas—name it "Armor revision 2"—and begin selecting. Drag images that show broad shoulders, layered plating, and overlapping metal onto the board. Some images are full model sheets; others are detail close-ups of how plate armor overlaps at joints. All of them need to be on one canvas so the modeler sees the full range of visual direction at once.

9:30 am — Build the armor reference canvas and note material finish changes

Rate the best matches up to 5. Tag them armor-revision-2. As you drag each one onto the canvas, note which ones show layered construction, which show leather underlayers, and which show the aging and wear that reads on a character model at distance.

Build the canvas as ten plates: full warrior silhouettes at different scales, shoulder construction detail, sleeve joint detail, waist plate overlap, leg protection, weathered metal studies, leather underlayers, closure mechanisms, material contrast examples, and color study. The canvas is a JSON file saved in the project folder, so the art director and the modeler can review it without needing to import anything.

Not every image is perfect. Some show the silhouette you want but the material finish is wrong. Rate those up to 3, tag them armor-revision-2-material-revision, and save them. They're reference iterations within the reference set.

10:30 am — The character passes armor review; now environment references need attention

The modeler approves the armor direction at 11. Move to environments. The level designer has two active environments: a dungeon interior and an ancient temple courtyard. Both need updated references for the lighting pass.

Search environment references from your archive:

tag:environment in:/Volumes/Reference-Archive rating:>=3

Two thousand images across twelve years of reference collecting. Filter by location:

tag:environment tag:dungeon in:/Volumes/Reference-Archive rating:>=3

Four hundred sixty images. These are too broad. Narrow to dungeons with the lighting quality the director described—stone material with existing light in the cracks:

tag:dungeon tag:material-stone tag:light-study added:>2025-06-01 rating:>=2

Ninety images. Better. Twenty-eight of them have the exact kind of light bounce the team is after. Start a canvas for each environment pass.

11:45 am — Create material reference swatches for texture handoff

Before lunch, prep the texture handoff. A canvas that shows material swatches does not need full images—detail crops work better. Add a note to each texture canvas with the rating threshold the texture artist should use:

tag:armor-revision-2 tag:surface-metal rating:>=4

The texture artist searches this same line and knows that every image rated 4 or higher is one that made it past art direction. Images rated 3 are "promising but needs review." Images rated 2 are archive—tried and not wrong, but not current direction.

1:30 pm — Concepting meeting: new enemy type means new asset class

Concepting approved a new enemy type that didn't exist in pre-production planning. It needs its own asset pipeline. Create new tags:

• tag:enemy-type-brute
• tag:enemy-type-brute iteration-1

Search for reference that might work:

tag:humanoid tag:strength in:/Volumes/Reference-Archive rating:>=2

Five hundred and nine images from anatomy studies to fantasy character art. Isolate the ones that show bulk and mass:

tag:humanoid tag:mass tag:silhouette rating:>=3 added:>2025-09-01

Eighty-four images from the past year of reference collecting. Create a canvas and pull the top thirty by rating. This is the starting direction for the new asset. Tag every image you pull with enemy-type-brute iteration-1.

2:15 pm — Cross-reference armor for the new enemy type

The new brute enemy wears armor, but not the armor the player character wears. It should read as heavier, less refined. Search for armor references that suggest weight and bulk:

tag:armor tag:heavy-construction rating:>=2 -tag:refined

Thirty-two images. Add a second canvas for the brute armor iteration, with a different direction than the warrior armor from this morning.

Tag everything with tag:enemy-type-brute tag:armor tag:iteration-1. Do not copy any files; the brute armor canvas and the warrior armor canvas reference the same images from the archive but with different tags and different arrangements.

3:30 pm — Archive last week's concepting direction

Last week's concepting pass is no longer active. Search everything tagged with the old direction:

tag:concepting-pass-12

Ninety-four images across all asset types. Rate them all down by one point to mark them as tried but archived. Replace the tag:

• Remove tag:concepting-pass-12
• Add tag:archived:concepting-pass-12

The images do not move. The query line changes. An artist who runs their saved view—tag:concepting-pass-12—immediately gets zero results and knows that direction is closed.

4:00 pm — Set up saved views for each asset pipeline

Each department on the production needs a specific view. Save views for them:

For the modeler:

tag:character-warrior tag:armor-revision-2 rating:>=3

For the texture artist:

tag:armor-revision-2 tag:surface rating:>=4

For the animator:

tag:character-warrior tag:silhouette rating:>=4 -tag:concepting

For the VFX artist:

tag:character-warrior tag:cloth-movement tag:animation rating:>=2

Each view is one line and lives in the app. When you rename a tag across the project, every view automatically uses the new name. No membership lists to repair. No imports to redo.

4:45 pm — Team hand-off: import a new environment reference batch from research

The research team sent over three hundred high-resolution environment photographs from a location shoot. Download them to:

/Volumes/ProjectDrive/References/Environment/Location-shoot-2026-09-11/

The watcher sees the new folder immediately and adds all three hundred images to the index without any action from you. Search for them:

added:2026-09-11 type:image

Three hundred matches. Begin tagging them by location:

tag:location-temple-exterior
tag:light-study
tag:material-stone

Each image gets at least three tags. Do not wait for a bulk tagging pass later. Tag as you go. The moment the research folder is fully tagged, every artist on the production can search:

tag:location-temple-exterior rating:>=3

And pull the references they need without ever seeing the raw pile.

5:15 pm — Export canvases for tomorrow's direction meeting

Export three canvases to the shared project folder—armor revision 2, brute enemy type iteration 1, and environment lighting study. Each one is a JSON file that goes straight into the shared drive or version control.

Tomorrow's direction meeting will review them, and the director will either approve or request changes. If changes are needed, create a new canvas—"armor revision 3"—and iterate. The old canvases stay in the project folder and in the archive. Three weeks from now, if the director asks "what happened to that shoulder design we looked at," the answer is in revision 2's JSON file.

The scale of the reference library

On day one of production, references are thin. On day one hundred and eighty, there are forty thousand images across six external drives. The query line scales.

Old asset passes are tagged archived:character-warrior pass-1, archived:character-warrior pass-2. New passes inherit tags from earlier work—a rigging pass updates tag:character-warrior tag:silhouette, an animation pass adds tag:character-warrior tag:cloth-movement. Saved views stay the same; only the tag names change as the project evolves.

When a vendor sends reference material, tag it with the date and a vendor tag: vendor:artstation or vendor:location-shoot. Search vendor:* to count how many references came from outside sources. The ones that made it to final art get tag:used:character-warrior. The ones that didn't stay tagged and become precedent for the next character that needs the same reference type.

Iteration without duplication

A character might go through six silhouette iterations, three rigging passes, five armor variations, and two material directions. That is sixteen canvases, but it is not sixteen sets of images. It is one set of images, re-tagged and re-arranged each time.

Armor iteration 1 pulls from tag:armor-revision-1 rating:>=3. Armor iteration 2 pulls from the same image archive, but with different tags and a different canvas. Both canvases live in the project folder as JSON files. Both reference the same images. Neither one requires duplication or import.

When armor iteration 2 is approved, tag every image in that canvas with approved:armor. Those images become the foundation for the next asset that needs similar reference. A vendor image used in armor iteration 2 gets tag:used:armor-iteration-2, and when you build the final asset delivery documentation, search tag:used:* to prove which reference materials made it to the final game.

The archive that becomes the next project's foundation

At the end of production, archive the project. Move the References folder to an external drive or a locked folder in your main archive, and tag everything:

tag:archived:project-name tag:archived:2026

Three years later, a new production needs a brute enemy character. Search:

tag:enemy-type-brute rating:>=4 added:>2023-01-01

What you learned about brute silhouettes, armor construction, and material finish from the first project is the foundation for the next one. The canvases are version control. The tags are institutional knowledge. Nothing is deleted; it is all still there, still searchable, and still useful.

Why this approach scales across large teams

A reference system built on the folder structure and metadata—not a proprietary database—means every person on the team can open the workspace and see the exact same reference library the first person set up. No syncing tools. No import steps. No waiting for a central database to catch up. The watcher reconciles the filesystem with the index; the same query works for the concept artist, the modeler, the texture artist, and the animator.

This is particularly valuable when your team spans time zones. A Vancouver artist tags a character silhouette at the end of their day, files it into the reference folder, and a Seoul artist sees it in the next morning's search without any action from either person. A texture discovered on ArtStation by the Japanese artist is tagged and rated during their shift, and the concept artist in Los Angeles finds it during theirs with a single search line.

The folder structure also means nothing special is required to bring new team members up to speed. A fresh artist clones the project folder or syncs via whatever system the team already uses—git, Dropbox, Synology—and opens the app. The workspace is already there, full of tagged references, complete with saved views that the senior artists created. There is no "getting the database" or importing a library backup. The library and the workspace are the same thing.

Comparing this to DAMs and other systems

A DAM forces you to upload and index copies of your reference material into a centralized system. That means your references live in two places and have to stay in sync. A design that changes mid-project—a texture you hated last week but need this week—requires you to go back to the DAM, find it, re-rate it. A reference system that works on the filesystem means the reference lives in your project folder or your archive, and you search it where it lives. You are not duplicating and not syncing.

How to organize references by project covers the folder structure that scales across years of production. How to tag reference images consistently digs into the tagging system. How to keep references in sync across two Macs solves the sync problem for distributed teams. Building a moodboard is the technical step for packaging direction. What is a DAM digital asset manager answers the comparison question directly.

How to organize references by project covers the folder structure that scales across years of production. How to tag reference images consistently digs into the tagging system. How to keep references in sync across two Macs solves the sync problem for teams on different machines. Building a moodboard is the technical step for packaging direction. Or just download it and try this workflow on your next concepting pass—thirty days, every feature, no card.