In my work, retopologizing scanned assets is a non-negotiable step to make them usable in any real production. I’ve found that raw scan data, while high in detail, is a mess of triangles and inefficient geometry that will cripple your texturing, rigging, and rendering pipelines. My workflow is a systematic process of analysis, rebuilding, and validation that transforms these assets into clean, animation-ready meshes with proper edge flow and UVs. This guide is for 3D artists and technical directors who need reliable, production-ready models from the growing library of scanned marketplace content.
Key takeaways:
When you download a scanned asset from a marketplace, you’re getting a direct digital capture of a real-world object. What this means in practice is a mesh composed entirely of dense, irregular triangles. The polygon count is often astronomically high, with no consideration for edge loops, quads, or deformation needs. This topology is terrible for several reasons: it bakes poorly, creating artifacts in normal maps; it deforms horribly if you try to rig it; and it’s a performance nightmare for real-time applications. The UVs, if they exist, are usually a fragmented, non-optimized mess. In short, it’s data-rich but artistically and technically unusable as-is.
Before I start retopologizing, I define what "done" looks like. My goal is a mesh that serves the project, not just one that looks clean in the viewport. First, it must be primarily quad-based. Quads subdivide predictably, deform cleanly for animation, and are the standard for sculpting and displacement. Second, the edge flow must follow the form and anticipated deformation. For a character, this means loops around eyes, mouth, and joints. For a prop, it means edges that define hard surface contours. Third, the polygon count must be appropriate for the target medium—much lower than the scan but high enough to capture the intended silhouette and detail through baking. Finally, it must have clean, non-overlapping UVs laid out in preparation for texture baking.
I never jump straight into retopology. I start by thoroughly examining the high-poly scan. I identify the key forms, areas of mechanical detail, and regions that might need to deform. I ask: Where are the major silhouette edges? Where will seams go? What is the primary, secondary, and tertiary form? I often use a shader that highlights polygon density or curvature to understand the scan's detail. This planning stage is where I decide on my overall strategy: which parts might be suitable for an automated pass, and which complex areas I’ll need to handle manually. I also set my target polygon budget here.
This is the core of the process. I create a new, low-poly mesh over the surface of the high-poly scan. My method depends on the asset:
A practical tip: I always keep my new mesh slightly inside the high-poly scan. This "shrink-wrapping" helps later during the baking process and prevents the low-poly silhouette from clipping outside the high-poly details.
Once my new, clean topology is complete, it’s just a smooth shell. All the detail from the scan is missing. This is where baking comes in. I first ensure my new low-poly mesh has a good UV layout. Then, I use a baking tool to project the high-poly detail onto the low-poly mesh via texture maps—primarily a normal map, but often also an ambient occlusion and curvature map. The key here is cage or ray distance adjustment. I meticulously tweak these settings to avoid baking artifacts like skewing or missing details. A successful bake means my low-poly model, when rendered with the normal map applied, looks virtually identical to the multi-million-poly scan.
Baking isn't the last step. I now inspect the baked maps for errors and clean up the mesh. I check for:
I default to manual retopology for any asset where control is paramount. This includes hero characters for animation, where every edge loop must be placed to facilitate smooth facial expression and body movement. It also includes key hard-surface props with complex, intersecting forms where automated tools often create a tangled mess of triangles at boolean intersections. The manual process is slower, but it yields a perfectly predictable, optimized mesh that I know will behave correctly through the entire pipeline.
For certain asset types, automated retopology is a massive time-saver. I use it for environmental assets like rocks, cliffs, or ruined walls where specific edge flow is less critical, and the primary goal is simply a drastic polygon reduction with preserved detail. It’s also useful for generating a first draft on more complex objects, giving me a quad-based starting point that I can then refine manually. In my workflow, I sometimes use Tripo AI to generate a base mesh from a concept image, which often arrives with surprisingly clean, quad-dominant topology that serves as an excellent starting block for further refinement, bypassing the initial chaotic scan data altogether.
My most common and efficient method is a hybrid one. I’ll use an automated algorithm to generate a base retopology for the entire object. Then, I selectively rework critical areas by hand. For a scanned statue, I might let the tool handle the drapery of the robes but manually retopologize the face and hands. This approach gives me the speed benefit of automation while retaining full artistic and technical control where it matters most. The key is to view automation not as a final solution, but as a sophisticated brush for blocking in the bulk of the work.
Retopology cannot be done in a vacuum. From the moment I start placing edges, I’m thinking about UV seams. I try to place seams in less visible areas and in straight lines to minimize texture distortion. After the new mesh is built but before baking, I finalize the UV layout, ensuring all UV islands are efficiently packed and have consistent texel density. I then create a dedicated "cage" or "projection mesh" for baking—a slightly inflated version of my low-poly that fully envelops the high-poly scan. This setup is crucial for artifact-free normal and displacement maps.
If the asset will be rigged, my retopology decisions are in service of the skeleton. For a character, this means:
Before I consider an asset finished and hand it off, I run through this final checklist:
moving at the speed of creativity, achieving the depths of imagination.
Text & Image to 3D models
Free Credits Monthly
High-Fidelity Detail Preservation