Input Guidelines
Asset Creation Guidelines
In order to get the most out of your Kaedim generations, we have some guidelines on the types of images we accept for processing, as well as information about images that are not suitable for model generation. We have different asset types to choose from, and you can find the guidelines for each type below.
General Checklist
✅ Single object - Please make sure your image(s) show a single object.
Requests of scenes involving multiple objects will be dismissed for Prototype assets.
If your character has a lot of armour and accessories, the armour and accessories should be passed as a separate request from the character itself. Learn more at Characters
When using multiple images, they must all be of the same object but from different angles. See our input image examples to learn more.
✅ Clear background - Your object needs to be clearly distinguishable from the background.
Learn more at Input Image Examples
✅ Photo/Art/Sketch - Your 2D input can be a photo of an object (within guidelines, see below), a piece of concept art, or a sketch (pencil, pen, or paper).
✅ Clear - The object should be clearly visible in the foreground of the image. Avoid blurry images.
Can only be processed through Game-Ready Pipeline:
Hero Assets
We suggest the Game-ready pipeline for complex characters that have many clothes/accessories. Learn more at Characters
Real animals
Realistic animals that have fur or need high-detail textures are currently only available with Game-ready assets.
Foliage
For Prototype Assets, a realistic full-grown tree will probably get dismissed. However, plant life that is less complex and/or stylised should work.
Feel free to experiment with various inputs. You might be surprised by our software's capability!
Invalid and Tricky Inputs
Below are some examples of image requests that may not be processed:
❌ Inappropriate imagery
This includes any nude or sexual content, offensive imagery, blood and gore, etc. Any images of this type will be automatically dismissed.
⚠️ Text
Since text is not inherently a 3D object, it may not be processed. However, if the image primarily features an object with some lettering on it, that should be acceptable.
⚠️ Multiple objects in one image
Our focus is on creating individual objects. If you submit an image with multiple objects, it likely won't be processed. If you have multiple objects, separate them and pass them as individual requests.
If you'd like to generate multiple objects within a scene, you can do so with Scene functionality (available for Enterprise).
Please refrain from uploading any inputs that contain personal information or identifiable details such as passports, ID cards, or addresses.
Last updated