Thomas Colin 3D: Topology Studies in Blender

Thomas Colin 3D has been modelling for thirteen years and his whole channel points at one skill: Blender topology. His topology studies are not about modelling a specific object, they are about deciding where every edge goes before subdivision decides for you.

Who Thomas Colin 3D Is

A French channel created on 17 November 2022, with around 38.5k subscribers and more than two million views. The channel description gets straight to the point: a professional 3D modeller with thirteen years of experience, and if you want to learn about topology and subdivision modelling, as he puts it, you are in the right place.

The subscriber count is not the interesting part. The focus is: years on one subject, without chasing whatever tool is trending this month, build a library about the mesh itself.

A Channel with a Single Subject

The most watched videos are not the ones modelling trendy props: fixing bad topology sits around 143,000 views, the clean topology pro tips are past 73,000 and the complex shapes video is near 35,700. Fixing meshes, ordering meshes, understanding meshes. That is what the audience comes for.

He Also Builds His Own Tools

The channel has released its own modelling add-ons. One of them, Multi Grid Fill, ties directly into the Grid Fill operator used further down in this article. It says a lot about how much the subject matters to him: when the tool you need does not exist, you write it.

What a Topology Study Is

A topology study is an exercise where you pick a complex shape (a mechanical part, an object full of holes, a radial assembly) and solve its mesh, not its finish. The output of the exercise is the wireframe, not the render.

It works because it trains the decision: where a loop goes, where a pole is acceptable, how much density each area needs. You do not learn that by memorising the steps of one specific tutorial, you learn it by studying meshes and failing fast and cheap: copying a model versus learning to read a mesh.

How It Differs from a Modelling Tutorial

This article does not repeat the hard surface workflow with booleans and bevels, which already lives in hard surface modelling in Blender, nor the retopology of a sculpt, covered in retopology in Blender. Here the subject of study is the mesh itself: what happens to it when you subdivide or deform it.

Reading an Edge Flow: Loops and Poles

The Blender manual defines loops as sets of edges or faces forming continuous chains. Those that do not end at a pole are cyclic: they start and end at the same vertex and split the model in two parts. Almost all subdivision modelling and a good chunk of character animation is built on that idea, because a mesh with few vertices and clean loops looks natural when subdivided and deforms well.

Poles are something else: vertices connected to three, five or more edges. Vertices with one, two or four edges are not poles. A loop stops at a pole simply because there is no single obvious way to continue. The practical conclusion, and this is the lesson the studies teach, is that poles are not bad: they are unavoidable. The craft is hiding them where they neither show up nor deform badly.

Then there is density: loops sit closer where the shape deforms and looser across flat areas. The manual uses shoulders and knees as its example.

The Real Test: Subdivide

The Subdivision Surface manual is blunt: without control loops, a subdivided cube stops looking like a cube. It is the best argument for why topology matters and the fastest way to judge your own mesh: turn the subdivision up and check whether the shape is still the one you wanted.

The Channel's Four Topology Studies on the Platform

All four are curated on CGI Academy Hub under the Blender category, at beginner difficulty. Each one targets a different decision.

Flawless Edge Flow

Topology Study: How to Get a Flawless Edge Flow solves the transition between a curved area and a flat one with loops that follow the contour. Beyond the result, what it teaches is seeing which loops are redundant and where the flow ends without leaving a pole in sight.

Radial Array: Two Routes

Topology Study: 2 Radial Array Methods compares two ways of reaching the same place. With default tools, the Spin manual asks you to place the 3D cursor at the centre first, then spin around it with a 360 degree angle and Duplicate enabled. The detail almost nobody checks: the seam keeps duplicated vertices, so you select it, run Merge by Distance and confirm the vertex count matches the original profile. The other route is an Array modifier with an object offset, or a Simple Deform modifier in Bend mode with a full angle and limits.

The second route in the video is HardOps, a paid add-on that automates the radial cut and its cleanup. It is not magic: it does in one click what the default route does in five steps, with the same decisions underneath. The takeaway is that the centre of a radial array is always a pole, and the study teaches you how to spread density so it stops being a problem.

Cleaning Up After a Boolean

Two complementary videos: How To Clean Your Mesh After a Boolean and Boolean Cleanup in Blender is Easy. The angle is simple: a boolean leaves n-gons and triangles along the cut, and the study is about getting back to quads.

The tools live in the mesh cleanup section of the manual and there are fewer than you think: Merge by Distance, with its threshold and Sharp Edges to keep existing hard edges intact; Fill Holes to close whatever is still open; Grid Fill to generate a grid of quads from a roughly rectangular loop of boundary edges, best when each pair of opposite sides has the same vertex count; Edge Slide to reposition a loop without moving the silhouette; and Bridge Edge Loops to stitch two areas together.

What About N-gons?

A working rule: an n-gon on a flat face that you are never going to subdivide or deform can stay. An n-gon where subdivision comes into play produces creases and shading artefacts. That is where the mesh gets rebuilt.

How to Practise with These Studies

A routine that works: one shape per session, repeated in three variants (more density, less density, a different flow direction) and with a time limit that forces you to decide. Then review the mesh against criteria: do the loops end at poles you chose? Are there n-gons in areas that will be subdivided? Does density follow deformation?

For reviewing, the manual describes two helpers in its retopology section: the Retopology overlay in the 3D viewport, which shows the original mesh through the new one without the distraction of X-Ray, and Poly Build with Snapping to add faces and pin new vertices onto the original mesh.

Mistakes the Study Exposes

Triangles and n-gons inherited from the cut. Loops that die halfway. Poles placed somewhere visible. Density where the shape does not deform. Most problems in a hard surface mesh come down to one of those four.

How Long Before You Notice

No promises. Progress comes from repeating different studies on similar shapes, not from stacking up watched videos.

Conclusion

Topology is a trainable skill and this channel turns it into short exercises. All four studies live in the curated Thomas Colin 3D channel, alongside videos from every other modelling channel on the platform. If you want to keep building a route, the blog has the Blender Guru channel breakdown, Abe Leal 3D on game art and the learning CGI from scratch starting point.