.webp)
Introduction
In this article, we will compare two 3D AI technologies - TripoSR and Meshy - for 3D model creation. We will analyze their strengths, weaknesses, and specific use cases, offering insights into how each tool performs under different conditions. Whether you are a professional 3D artist or a developer integrating 3D models into your projects, understanding the capabilities of these AI-driven tools will help you make an informed decision. Let's dive into the details to explore which AI tool stands out and why.
Practical tests
Both models will be tasked with generating a 3D object based on a given image. Their ability to interpret the visual data and accurately reconstruct the object in 3D will be tested, providing insight into their capabilities in visual-to-spatial transformation. This challenge will highlight their performance in understanding complex visual inputs and converting them into precise 3D models.
Prompt 1. Geometric shape:
Let's begin with a simple cylinder with a rough surface to test their capabilities.
TripoSR output
Meshy output
Correct Answer: ❌
There is an attempt to capture the roughness of the object, but the cylinder’s surface is cut off.
There is an attempt to capture the roughness of the object, but the cylinder’s surface is cut off.
Correct Answer: ✅
The object looks good, albeit without texture.
The object looks good, albeit without texture.
Prompt 2. Topological Complexity:
Next, we will assess the topological accuracy of the models by providing them with an image of a torus featuring a high number of segments.
TripoSR output
Meshy output
.webp)
Correct Answer: ❌
TripoSR produced a highly confusing topological mesh, and the figure itself is inaccurately represented, failing to capture the intended shape.
TripoSR produced a highly confusing topological mesh, and the figure itself is inaccurately represented, failing to capture the intended shape.
Correct Answer: ❌
Meshy produced a less tangled mesh, but the shape is still inaccurately rendered, falling short of the desired outcome.
Meshy produced a less tangled mesh, but the shape is still inaccurately rendered, falling short of the desired outcome.
Prompt 3. Mesh Density and Detail:
In this test, we'll evaluate how they manage mesh density and detail.
TripoSR output
Meshy output
.gif)
Correct Answer: ❌
Generated a complex and confusing topological mesh, but the overall shape is conveyed well despite the mesh's intricacy.
Generated a complex and confusing topological mesh, but the overall shape is conveyed well despite the mesh's intricacy.
Correct Answer: ✅
The grid is beautiful and neatly arranged, effectively capturing the shape of the object with precision.
The grid is beautiful and neatly arranged, effectively capturing the shape of the object with precision.
Prompt 4. Symmetry:
Let’s provide them with a symmetrical image and observe how well they handle it.
TripoSR output
Meshy output
Correct Answer: ✅
The shape is captured well, passing the test.
The shape is captured well, passing the test.
Correct Answer: ✅
Both models handle symmetrical objects well, accurately preserving balance and proportions in their output, with Meshy showing better colors.
Both models handle symmetrical objects well, accurately preserving balance and proportions in their output, with Meshy showing better colors.
Prompt 5. Layered structures:
Let’s send them a photo of the cake to evaluate how they handle the layered structure.
TripoSR output
Meshy output
.gif)
Correct Answer: ❌
If you overlook the top of the cake, the layering is captured quite well, with the distinct levels clearly represented.
If you overlook the top of the cake, the layering is captured quite well, with the distinct levels clearly represented.
Correct Answer: ✅
Both models handle symmetrical objects well, accurately preserving balance and proportions in their output, with Meshy showing better colors.
Both models handle symmetrical objects well, accurately preserving balance and proportions in their output, with Meshy showing better colors.
Prompt 5. Complex figures:
The final test will involve providing them with a complex figure of a fox.
TripoSR output
Meshy output
Correct Answer: ✅
The model turned out to be accurate, effectively capturing the details of the image.
The model turned out to be accurate, effectively capturing the details of the image.
Correct Answer: ❌
Both models handle symmetrical objects well, accurately preserving balance and proportions in their output, with Meshy showing better colors.
Both models handle symmetrical objects well, accurately preserving balance and proportions in their output, with Meshy showing better colors.
Pricing
The price of TripoSR is around 10 times cheaper than that of Meshy for 3D AI generation. The price for Meshy is taken for its minimal price plan - $20 montly subscription.
Try for yourself
You've seen what these models can do - now you can try TripoSR 1.4 for your use case. Plug the code below into Google Colab or any IDE, use your API Key, and get testing!
import requests
def main():
response = requests.post(
"https://api.aimlapi.com/v1/images/generations",
headers={
"Authorization": f"Bearer {aiml_api_key}",
"Content-Type": "application/json",
},
json={
"model": "triposr",
"image_url": ("https://upload.wikimedia.org/"
"wikipedia/commons/7/7a/Basketball.png"),
},
)
response.raise_for_status()
data = response.json()
url = data["model_mesh"]["url"]
file_name = data["model_mesh"]["file_name"]
mesh_response = requests.get(url, stream=True)
with open(file_name, "wb") as file:
for chunk in mesh_response.iter_content(chunk_size=8192):
file.write(chunk)
if __name__ == "__main__":
main()
Conclusion
TripoSR and Meshy both have strengths and weaknesses in generating 3D objects from images. TripoSR 1.4 struggles with basic shapes and accuracy, but is better at layered structures and symmetry. Meshy handles basic shapes well and renders layered figures, but lacks detail and precision in textures and organic forms. The best choice depends on your project's specific needs. As AI technology advances, both tools will likely improve, especially with TripoSR 2 right around the corner.
You can check our model lineup here - try any of them for yourself with our API Key.
%20(1).png)
Testimonials
.png){kind=small}
.png){kind=small}
