A Monday scheduling meeting. Art reports the estimate: this character still needs five days. You cannot tell what those five days cover — five more days of concept art, or modeling and animation as well? You are not an artist, so you do not press, and you nod. Two weeks later the character is still not in the build. When you ask, the answer is that concept art was revised once in the middle.

Many PMs get stuck here. I am Hao, in my 10th year as a game PM, originally a game designer, and I now write Hao’s Game PM Notes (Hao的游戏PM笔记). I was a PM who did not understand art and was then pushed to manage the art line. This piece is how I think game art project management should be done, written for people in the same position.

One reassuring point first: you do not need to know how to draw in order to manage art. You need to treat art as a production process.

1. Split art into two pipelines

To a PM, art often looks like a black box. Once you take it apart, it becomes clear.

Game art assets fall roughly into two kinds: the character line and the scene line. The character line starts from concept and design, then goes through 3D modeling, texturing, and skeletal animation, and finally enters the engine. The scene line has similar stages, but more parallelism: several scenes can go to different people at the same time. The scheduling logic differs. The character line is mostly serial — the next stage cannot start until the previous one is done. The scene line can add people to buy time. UI and VFX have a process structure similar to the character line, so the same breakdown applies; I will not discuss them separately below.

The first thing I did when I took over the art line was to schedule these two lines separately. Mixed together, the character work got stuck, the scene people waited, and every weekly meeting was spent explaining another slip.

One more principle is a lifesaver: stage output should be something that can actually run, not a work-in-progress file. A signed-off concept piece does not count as output. A character that can be turned around in the engine does. Process files can mislead you. Something that runs will not.

2. Estimating time: do not ask “how many days are left”; ask “which step is it stuck on”

For a PM without an art background, asking for an estimate is the scariest part. Ask “how many days are left,” and the answer is always “two more days.” Those two days can last two weeks.

Break the asset into steps and ask about those. Do not ask whether a character is “done.” Ask whether concept has been approved, what stage modeling is in, who is doing the textures, and whether animation has been scheduled. Each step has a rough reference duration. Whichever step has overrun is where the problem is hiding. You do not need to do the work. You only need to ask questions against a step list.

Progress reports should change with this. Do not ask “is it finished?” Ask “which step is it in now, how many days has this step already taken, and who picks it up next?” After those three questions, real progress surfaces on its own. “This character still needs five days” becomes: how many days has concept already taken, and when will it be handed to modeling. Once the number is attached to a step, it is much harder to stay vague.

3. Outsourcing: most rework dies in the handoff, not in production

Art outsourcing is where schedule and quality risk concentrate most heavily. The failure pattern I have seen most often is this: the request is briefed verbally, nobody looks at the work in the middle, the file is opened the night before delivery, the style and the specs are wrong, it goes back for rework, and the version milestone is blown through.

The responsibility usually does not sit with the vendor. It sits in the handoff. There are four management actions:

Put the handoff on paper. Write the request, reference images, spec parameters, and naming rules into a handoff sheet, and start only after both sides have confirmed it. Saying “I want that kind of feel” out loud is an invitation to improvise.

Put the mid-point check at one-third, not at the end. If you wait until the work is finished before you look, it is already too late. A review at one-third is when a wrong direction is cheapest to correct.

Give the acceptance criteria in advance. The vendor should know, before work starts, what acceptance will look at. Adding requirements at the last minute is a main source of disputes.

Leave a channel for escalating problems. The most expensive case is a vendor that gets stuck, says nothing, and toughs it out until the last day. Say this clearly at the start: if you are blocked, report it at any time; an early report will not be treated as a fault.

4. Be the translator among design, engineering, and art

Half of the work of managing art sits at the boundaries.

Design speaks in “feel”: more ornate, more oppressive. Engineering speaks in “specs”: the polycount is over the limit, the bone count has to be cut. Art speaks in “look”: this light bake has to be done again.

The three sides are not speaking the same language. The PM translates “feel” into reference images and priorities, “specs” into change items that art can execute, and “look” into schedule numbers that can be put on the table in a meeting.

Requirement changes also explode at this boundary. Design revises a request, and three days of art work disappear. Changes need a record and a visible cost: which steps this change affects, and how many days it adds, should be put in the open. People become more careful once that is visible.

5. Watch three signals every week

Art risk does not need daily watching. Three signals a week are enough.

Stuck tasks. If the same step has sat past its reference duration, go and ask.

Change frequency. If one asset line is changed more than twice in a week, the upstream request has not been thought through. Block the source; do not only clean up downstream.

Outsourcing rework rate. If the same vendor is sent back for rework repeatedly, first check whether the handoff sheet exists only on paper, and only then consider changing vendors.

Behind these three signals is a simple tracking sheet. Half an hour a week going through it is far better than discovering a general delay at the end of the month.

6. A starter checklist

  • If you have just taken over the art line: split the character line and the scene line in the schedule, then list work-in-progress assets by step. In that one week you will already be ahead of most “nodding PMs.”
  • If you do not have an art background and feel unsure: first learn the terms. Once you can follow “high-poly,” “baking,” and “skinning,” meeting efficiency doubles, and it becomes easier to earn the art team’s trust.
  • If you already manage an engineering or design line and want to move up: volunteer to manage a small art asset line. Art management is a required question for a senior PM. The earlier you touch it, the earlier you accumulate the work.
  • If your team is about to start outsourcing: put the handoff sheet, the mid-point check, and the acceptance criteria in place first, and only then talk to vendors.

The entry path is plain: take the black box apart into a process, replace “feel” with steps, and replace verbal agreements with paper. The rest is practice.

Frequently asked questions

Can you manage an art project well if you know nothing about art?

Yes. A PM manages process, schedule, quality, and collaboration, none of which require an art foundation. What you need to add is two things: the production process for art assets, and the basic terms. Leave aesthetic judgment to the art lead. Use your judgment on whether the process is reasonable.

When art says it cannot be finished, how do you tell whether that is true or an efficiency problem?

Break it open by step. If every step is within its reference range and the total has still overrun, it is a scheduling problem: adjust the plan or add people. If one step is clearly overtime, look at the specific blockage: repeated revision, unfamiliar tools, or an unclear request. “Cannot be finished” carries no information by itself. It only does once you take it apart.

What do you do when outsourcing quality is unstable?

First check your own handoff sheet and acceptance criteria. If those two are unclear, the result will wobble no matter which vendor you switch to. If it is still unstable after the handoff is clear, look at fit. Some teams are strong at characters, some at scenes; assigning the wrong kind of work can also look like “unstable quality.”

How should an art PM and an art lead divide the work?

The art lead is responsible for art quality and style. The art PM is responsible for the schedule, the process, cross-functional collaboration, and outsourcing. In short, the lead watches whether it looks good; the PM watches whether it is on time and running smoothly. On projects where the two work well together, the art line rarely runs into serious trouble.


The public Chinese original is here: How to manage game art when you are not an artist.

More notes on game project management at https://pmnote.ai/, and in the WeChat mini program “Hao的游戏PM笔记”.