Try-on

                        ✖
                    
                    Camera input
                    
                    Camera resolution
                    
                    Rotate camera input
                    
                    Mirror camera X-axis
                    
                    Mirror camera Y-axis
                    
                    Low quality mode
                    
                    ML quality
                    
                    Mirror mode
                    
Restore default settings
Send logs

v141

Share

✖

                    Demo: 
                    
                    ML_model
                    
                    Shoe
                    
                    Enable camera
                    
                    Enable ML
                    
                    OF enable ▼
                    
                        OF image size
                        
                        OF window size
                        
                        OF timeout
                        
                        OF Thread
                        
                    Vertical sync
                    
                    Enable smooth
                    
                    Preffer points close to last pos
                    
                    Ignore confidence for pred points
                    
                    Compute jitter points
                    
                    Compute missing points
                    
                    Ignore points out of cluster
                    
                    Ignore points at wrong foot
                    
                    Ignore output by reprojection
                    
                    Ignore output by size
                    
                    Hide one shoe if overlapping
                    
                    Hide both shoes if one
                    
                    Enable cropping
                    
                    Async mode
                    
                    Show ML frame
                    
                    Show points / labels
                    
                    Show occluder / shoes
                    
                    FPS cap
                    30

                    Experimental ▼
                    
                        Compute mode
                        
                        TF backend
                        
                        Use ransac
                        
                        Distortion
                        
                        Ignore points by reprojection
                        
                        Keep shoe visible for short time
                        
                        Process output tensor by GPU
                        
                        Confidence
                        0.15

                        Buffer length 
                        1