VANCOUVER, BC, June 8, 2022 – 3DQue Systems Inc. announced today that its state-of-the-art QuinlyVision failure detection system for 3D printers is ready for live testing on 3DQue’s Discord.
Packed with cutting-edge features, QuinlyVision sets itself apart with AutoCorrect3D™, to detect, automatically respond, and suggest solutions to 14 common print failures. By combining AI failure recognition with object tracking, photogrammetry, and subpixel analysis, QuinlyVision detects issues in seconds, even in low-light conditions. Get peace of mind with customizable auto response options, a built-in troubleshooting guide, 24/7 unlimited supervision, and secure remote access.
“The excitement of bringing an innovation to life is tempered by fear that the print will fail. Existing technology focuses on detection, not correction. Our goal is to provide an elegant solution to simplify failure correction by having QuinlyVision do the legwork – pinpointing the problem, automatically taking action, and providing suggestions to prevent future failures.” Co-founder and CEO of 3DQue, Steph Sharp.Quick Diagnostics, Solutions at Your Fingertips
Getting consistently high-quality 3D prints requires a delicate interplay of design, calibration and slicing. Determining what went wrong, why, and how to fix it can be a time consuming process of trial and error, as over 90,000 members of the “fixmyprint” subreddit can attest.
By classifying failures and how they develop over time, the built-in Correction Wizard pinpoints root causes such as hardware, slicing, or model issues. It provides detailed troubleshooting information right on the dashboard.Autocorrect3D 14 Failures
Like text autocorrect, QuinlyVision’s powerful algorithm diagnoses the problem and suggests possible solutions, saving time and frustration. Autoresponse goes even further, allowing different actions (pause, notify, or ignore) for each failure type, allowing printers to self-correct the moment an issue is detected.
14 failure types:
1. Detach | 6. Zits | 11. Poor-First-Layer |
2. Nozzle-Blob | 7. Stringing | 12. Model |
3. Skirt | 8. Spaghetti | 13. Poor-Bridging |
4. Under-Extrusion | 9. No-Extrusion | 14. Layer-Shift |
5. Over-Extrusion | 10. Warping |
QuinlyVision will be available for live testing in 3DQue’s Community Discord starting this week.