[image: 1737596065204-header.jpg]
Recently, I came across the new image-to-3D tool Shapen.com, and given our frequent use of similar tools at ARLo, I decided to give it a quick test. My aim was to evaluate whether Shapen offers better quality or features compared to existing solutions like Meshy.ai, a tool we’ve tested in the past.
TL;DR: Shapen allows users to upload up to three images on its free plan, and its output is notably cleaner than Meshy’s default results. However, it struggles with fidelity, introducing noticeable AI “interpretation” that deviates from the original input. At $240/year for the base subscription, the value proposition is questionable, but the tool shows potential for future development.
Testing Process
For my test, I used a LEGO model of the Orca from "Jaws" as the subject. I photographed it on a white table under neutral lighting, capturing eight images from various angles. To optimize input quality, I removed all backgrounds, centered the images, ensured consistent exposure, and maintained equal scaling across the photos.
[image: 1737592402719-01.jpg]
[image: 1737592402750-03.jpg]
[image: 1737592402779-07.jpg]
After creating a free account on Shapen.com, I uploaded three of the images (a limitation of the free plan) that best captured the front and side views of the model. The tool’s model creation process is straightforward, with minimal configuration options. Impressively, the model generation completed in under a minute.
Shapen supports only two export formats: GLB and STL (XR Devs might not be familiar with STL - it's generally used in 3D printing). While this is sufficient for many workflows, it’s worth noting that competitors typically offer a wider range of export options, including FBX, OBJ, and USDZ. If your XR workflow depends on one of these formats, you may need to add a conversion step (with all its potential pitfalls) to meet your specific requirements.
Results and Comparison
Shapen's output featured clean lines and well-applied textures, resulting in a polished standalone model. When compared to Meshy’s default output, Shapen produced noticeably cleaner results with fewer artifacts. However, Shapen lacks advanced features for adjusting detail, mesh type (e.g., quad vs. triangle), or mesh complexity—features Meshy includes in its paid plans.
It’s important to highlight that Shapen’s AI appears to prioritize creating a visually clean model, sometimes at the expense of fidelity to the original input. For instance, certain details of the LEGO model were “interpreted” by the AI, leading to a less accurate representation compared to Meshy’s more faithful but less polished results.
Here are the links to view the models:
Shapen: Click to view Shapen model
Meshy: Click to view Meshy model
Verdict
Shapen offers an impressive user experience, generating clean 3D models with minimal effort. However, its limited input fidelity and lack of advanced controls may be dealbreakers for professional users seeking high-accuracy outputs. Additionally, the $240/year subscription fee for the lowest tier makes it a questionable investment unless you frequently generate 3D models.
Compared to Meshy, Shapen stands out for its cleaner results, but Meshy’s configurability and adherence to input accuracy remain valuable. While neither tool seems ready for high-end production use, Shapen’s potential to disrupt this space is worth watching as the technology matures.
For now, if you’re working on casual projects or exploratory work, Shapen could be worth exploring. For production-grade 3D modeling, you may still need to rely on traditional modeling pipelines or stock model purchases.
