Working notes on where AI is genuinely useful in a 3D pipeline and where it is still oversold, written for the person who will have to fix the output — so every piece states what the technique does not handle. Published by Mixar, a 3D editor built on Blender with an AI agent inside it, and written to be useful whether or not you ever install it.
An AI map generator makes base colour, roughness, metallic and normal maps. Which generate well, which must be baked from geometry, and why.
7 min readJudging image to 3D for game assets means polycount, UVs, texel density, pivots and engine import, not the render. The checklist that matters.
7 min readBlender AI plugins fall into five categories with different failure modes. What each does, and the version, privacy and safety questions to ask.
7 min readPhoto to 3D works far better with the right reference. The lighting, background, angle and lens choices that decide the result, and what to fix.
7 min readGenerated skyboxes are convincing backdrops and weak light sources. What equirectangular output needs, how to wire one into Blender, and the limits.
6 min readText to 3D in one step is convenient and hard to direct. Generating an image first, then converting it, gives back the control that matters.
5 min readTripo vs Hunyuan 3D compared on the controls they expose, the shape of their outputs and how much of the pipeline each covers, without invented benchmarks.
6 min readThe step-by-step automation guides live under the AI agent for Blender pillar, and the head-to-head pieces are in comparisons.