The Red-Pen Roundtable: How to Run a 90-Minute 3D Critique That Ends in Fixes

Every 3D artist has sat through a critique that went nowhere. The model looks sharp in the turntable, everyone nods, someone says it has great vibes, and you leave with no idea whether the thing will survive contact with an engine, a printer, or a client handoff. At 3D Model Club, we have seen plenty of gorgeous assets fall apart on details that renders hide: flipped normals, eight material slots, mystery names like Cube.003, and no LODs to speak of.

The fix is not a harsher room. It is a better format. The red-pen roundtable is a 90-minute peer review built around production readiness, not taste. It works online, and it works even better in a US makerspace with a projector and a long table. Here is how to run one without turning it into a roast or a compliment circle.

Before the Session: Make the Asset Auditable

Ask presenters to submit one asset 48 hours ahead. The package should include the source file, a game-ready export such as FBX or GLB, textures, a few screenshots from intended camera angles, and a short readme with the ship context. That readme matters more than people think. It should list the target platform, engine or printer, poly budget, unit scale, intended use, and deadline. Without that context, every critique becomes a debate about personal preference.

Cap the night at three assets. Four is possible if they are simple props. The host creates one shared scorecard per asset and shares it with the group. Online, that can be a browser doc or a whiteboard tool. In person, print half sheets and hand out pens. The red pen is not a gimmick. It forces reviewers to write down a fix instead of talking in circles.

The 90-Minute Agenda

• 0 to 5 minutes: Ground rules. Critique the asset, not the artist. Every note must be specific and tied to ship context. No make it pop unless the reviewer can say what to change.
• 5 to 10 minutes: The host reads the ship context aloud. Platform, camera distance, budget, deadline, constraints. Everyone hears the same brief.
• 10 to 70 minutes: Three 20-minute review blocks. Each block runs like this: 2 minutes for the presenter to walk through the asset, 6 minutes of silent scoring, 8 minutes of discussion, and 4 minutes for the presenter to write an action list. The presenter does not defend during discussion. They listen and take notes.
• 70 to 80 minutes: Cross-asset patterns. Are names consistent across the three assets? Are material slots reusable? Are UVs at similar texel density? This is where a small studio catches pipeline drift.
• 80 to 90 minutes: Commitments. Each presenter states the top three fixes they will make and when. The host records it. No fix is assigned to the group. The asset owner leaves with the list.

That agenda is tight on purpose. A 90-minute session with three assets gives each one real attention without turning into a five-hour marathon. The timer is not there to be cruel. It is there to keep the conversation from collapsing into one person's favorite edge loop.

The Six-Line Scorecard

Use three ratings: blocker, fix, and fine. A blocker means the asset cannot ship in its current state. A fix means it should be addressed before final delivery. Fine means acceptable for the stated use. Every blocker and every fix needs a one-sentence remedy. A score without a fix is just a mood.

1. Topology. Check edge flow around deformation areas, non-manifold edges, n-gons where they matter, and scale. A clean silhouette can still hide a mess at the elbow or the wheel well.
2. UVs. Look for flipped UVs, overlapping shells, bad seams, texel density, and packing efficiency. If the asset will be seen close up, texel density should match the camera distance in the readme.
3. Material slots. Count them, name them, and inspect PBR values and texture paths. Orphan slots and default names are pipeline tax. A prop with 12 slots when the brief says four is a blocker, not a style choice.
4. Naming. Objects, meshes, materials, textures, and files. Prefixes and version numbers should be consistent. Cube.003 is not a name. It is a warning sign.
5. LODs. Check silhouette at distance, poly counts, transition quality, and shadow behavior. If the asset has no LODs and the brief calls for them, that is a blocker. If LODs exist, make sure they do not pop like a cheap magic trick.
6. Handoff files. Units, applied transforms, axis orientation, export settings, texture folder structure, readme, and license. This is the least glamorous line on the card and the one that causes the most Friday-afternoon emergencies.

How to Keep It Kind and Useful

The presenter talks first, then listens. Reviewers ask questions before prescribing fixes. Use phrases like I would check the seam under the handle instead of you always do this. The host keeps time and makes sure quiet reviewers get a turn. Rotate the host every month so the format does not become one person's personality.

Online, use a shared scorecard and ask everyone to add timestamped notes in the file or doc. Record only with permission. In person, a makerspace in Minneapolis or a community college lab in Tucson can work well: reserve a room with a projector, bring power strips, and test the display before people arrive. Pizza is optional. A working projector is not.

What Happens After the Red Pen

The presenter has 48 hours to update the asset and post a short before-and-after note in the group channel. That closes the loop and teaches the room something. Keep a running fix log for recurring issues. After a few sessions, you will see patterns: everyone forgets texel density, or everyone names textures differently, or everyone skips the readme until handoff day.

A 90-minute critique will not replace a real QA pass. It will catch the stuff that renders hide and give artists a concrete list before the asset leaves their hands. That is the point of the red-pen roundtable. You are not there to admire the model. You are there to help it ship.