Blender is free and magnificent and has a learning curve that defeats most designers in the first afternoon. Spline's achievement is making 3D approachable without hiding what 3D actually is, so you still meet meshes, materials, lighting and cameras, presented in an interface that does not require a week of orientation. The tutorials are the strongest part of the offering, being well produced, sensibly paced, and built around making something rather than touring the interface. Someone with no 3D background can follow them and have a scene they are pleased with in an evening, and that is genuinely hard to achieve.
The interactivity and export story is what makes it more than a toy. Building a scene that responds to scroll, hover or click, and shipping it to a real website, is the use case most designers arriving here have in mind, and the material covers it properly. Running entirely in a browser removes the setup and hardware friction that stops people before they start, and collaboration works the way people now expect it to. The AI generation features are the least compelling part and receive more prominence than they earn.
Generating a scene or an object from a description sounds appealing and produces results you then have to substantially rework, and the reworking requires the 3D skills the feature implied you could skip. The genuinely valuable thing here is the teaching, and I would rather the material leaned into that than chased the generation trend. Where I would push hardest is performance. A 3D scene on a web page is not a picture, it downloads assets, compiles shaders, and runs continuously on the visitor's device.
The tutorials build increasingly ambitious scenes and rarely stop to discuss what they weigh, what they cost on a mid range phone, or what happens on a slow connection. A designer who follows the material to its natural conclusion ships something beautiful that makes their site measurably slower and their bounce rate worse, and finds out from an engineer later. Optimisation guidance exists and sits well behind the creative material when it should run alongside it. Accessibility never comes up at all, which is a notable gap for a tool producing interactive web content.
Motion sensitivity, keyboard navigation, what a screen reader encounters, whether the content has a non 3D equivalent. These are real obligations for anything shipped to the public web and the material is silent. Ambitious scenes also hit browser limits sooner than the tutorials imply, and knowing where that ceiling sits before you design past it would save people rework. Three point six.
Excellent teaching and a genuinely well judged entry point into a difficult discipline, marked down for chasing AI features it does not need and for teaching people to build web experiences without teaching them what those experiences cost the person loading them.