CADLasso

Easily import real-world geometry into Fusion 360

Installation

Download files: git clone https://github.com/njwidmann/CADLasso

Install libraries: pip install requirements.txt

Tested only on Windows 10 so far

Usage

Start Python App

python3 gui.py

Load Image File

Press "File" to select an image file.
You can choose to downsample (resize) the image (e.g entering 4 will downsample image to 1/4 of original size).
Press "Load" to load the image into CADLasso
After you press "Load", the original (not downsampled) image dimensions will appear next to the file location
You can adjust the downsampling factor or the image file location at any time and press "Load" again to reload. Note that you will lose any contour selections you have done so far.

Select Contours

The contour selection dialog allows you to chain together multiple contours (found using OpenCV) to "lasso" an object of interest.

Click on the image preview in the GUI to open the contour selection dialog
Pan around the image using W,A,S,D keys and zoom in and out using E,Q keys.
Select contour points by clicking with the left mouse button. After your first click, as you move your mouse, a green preview will latch onto the closest recognized contours ("magnet mode") or connect to your previous point with a straight line ("straight line mode"). See recommended practices below.
Press the C key to toggle between "magnet mode" and "straight line mode" for contour selection
Right click to undo the previous selection.
If you are trying to lasso a closed loop, make your last click very close to your first click.
Press ENTER key to save selections.
Press ESCAPE key to exit contour selection without saving changes.
You can click on the preview again to edit your selection at any time

Set Scale

Set the scale to convert from pixels to real-world coordinates. You can use a caliper or a ruler to measure a dimension that will be used for the conversion. The scale selection dialog allows you to select the dimension width in pixels

Click on the image preview in the GUI to open the scale selection dialog
Pan around the image using W,A,S,D keys and zoom in and out using E,Q keys. Zooming in will help you more accurately select the dimension end points
Select dimension end points by clicking with the left mouse button. After your first click, as you move your mouse, a green preview line will show you the dimension.
Right click to undo the previous point selection
Press ENTER key to save selection.
Press ESCAPE key to exit without saving changes.
The measurement length in pixels will show up below the image
Enter real-world length of the dimension in millimeters in the GUI
You can click on the preview again to edit your selection at any time

Add Tolerance

Adding tolerances can help to ensure that your real-world parts will fit inside of the parts you design. This is especially useful for 3D printing. The below example shows a tolerance of 0.3mm that I used to fit a lens into my 3D printed glasses frame. This tolerance shifted all of the points outward normally from the tangent of the contour by 0.3mm. Positive tolerances expand shapes and negative contours shrink them.

Save Point Data

Press the "Save" button to save the point data to a CSV file. The file will save to the "CADLasso/output/" folder with the filename that you specify in the entry box. You can save point data as many times as you'd like so it's sometimes a good idea to save multiple times with different tolerances so you don't need to go back and retrace later.

Import into Fusion 360

Add FUSION_Import_Points.py script to Fusion 360 as shown in the image below

Run the script.
Select your .csv file saved from CADLasso Python App
You will be prompted to select a sketch for the points/spline to be added to. Click on the sketch in the feature tree to select.
Edit import settings at the beginning of the FUSION_Import_Points.py file:
- Change SPLINE to True or False to set whether to connect (and close) points with a spline
- Change DOWNSAMPLE to an integer >= 1. If set to 30, then only import once every 30 points. Check the number of lines in your saved CSV file to see how many raw points there are. Having too many will take a long time to load (See Best Practices below)
- Change CONSTRAIN to True or False to set whether to add dimensions to all the points so that they are fully constrained. This takes considerably more time but is very useful if you plan on importing into a design with existing geometry because you can move all the points around together.
- Change SCALE to a number >= 0 to set scaling. For example: 1 = no scaling, 1.5 = 150% scale, 0.5 = 50% scale.

Recommended Practices

Use pictures from different directions/views

By taking multiple pictures with a top, front, and/or side view, you can better recreate 3D geometry. The following example shows how front and side views of a scraper tool were used to build a model in just 1 extrude and 1 extrude-cut operation.

Preprocess Images

It will make the process much easier later on if you start by cropping, rotating, and straightening image files before loading into CADLasso. Try to match the orientation you plan to use in your design.

Downsample

Downsampling your images (either in CADLasso or beforehand) will make selecting contours and importing into Fusion 360 much faster.

Sometimes with very high resolution images, a large contour will be split into many smaller contours which may require you to click a lot around the object perimeter. Downsampling can reduce the number of clicks significantly.
Downsampling (either in the CADLasso app or later when importing into Fusion) will also greatly reduce the number of imported points. The loading time in Fusion seems to increase exponentially as the number of imported points increases. Try to stick to less than 50 points around an object perimeter if possible (the spline in Fusion will help fill in the gaps), especially if including dimensions to fully constrain the points.

Zoom out while selecting contours

When possible, (depending on your required accuracy) zoom out while selecting contours. As you zoom in, the resolution of the image passed into the contour recognition algorithm will increase. You can often select longer contours quicker and easier by zooming out.

Measure in the plane of the sketch

When setting the scale, make sure your measurement is in line in the z-direction with the plane that you plan to sketch on. Otherwise your measurements could become distorted due to the perspective of the camera.

Don't get too close

Make sure that you are far enough away from your subject when taking your picture that the geometry is not affected by perspective. Larger objects will require you to get further away. When taking a picture further away you can improve the image by using a higher resolution camera (and cropping) or a long focal length lens.

Tolerance is different than scaling

Scaling will increase the size of a shape by a specific percent while adding tolerance will widen/shrink a shape by a specific amount in all directions. If you have a perfect circle, these will be the same. But for an object that is wider than it is tall, or vice versa, they are very different.

Next Steps

Improve contour algorithm.
- Better recognition
- Connect multiple contours in a single click
Add Fusion 360 GUI so you don't need to edit settings in the script
Incorporate the CADLasso python GUI into Fusion 360 as an add-on so that the entire process is much more seamless and intuitive. This has proven difficult so far because you cannot easily import libraries (like numpy and OpenCV when running scripts in Fusion).

njwidmann / cadlasso Goto Github PK

cadlasso's Introduction