How to Remove Background from Photo by Color
PhotoScissors uses a deep neural networks algorithm to remove background, usually this method provides good results and in some cases just needs a small manual fix. Also you can use grouping by color algorithm. This algorithm can help if your background and foreground have different colors and may fail if background and foreground are similar.
Step 1: Open the photo you want to edit
When you open the image, PhotoScissors automatically removes the background from your photo. PhotoScissor automatically cuts out the foreground object as you do and displays the result in the right window.
But at this image we want to keep only the hands and we need to clear predicted mask. Click Undo or select "Edit->Clear Selections" menu.
Step 2: Enable Group by Color feature
Unlike many photo editors, PhotoScissors provides a simple way to identify which areas on the picture you would like to keep, and which ones (the background) you prefer to remove. The in-app two-color marker is a tool to mark these parts of the photo.
Step 3: Remove background by color
The green marker denotes foreground areas; the red marker is for background. So, all you need to do is put few rough strokes with your mouse and mark foreground and background areas. The result is displayed immediately in the right window. Here is how it looks:
You do NOT have to carefully select the areas, just make sure that the red (remove) marker touches any colors that you want to become transparent and the green (keep) marker touches any colors that you want to keep.
Stay Within The Lines
Make sure you stay within the lines of the object you want to cut out. If you mark an area outside of the foreground with a green marker, a part of the background can be cut out as well. Similarly, don't mark red on the foreground.
It's s a simple case of painting the part (or parts) of the image you want to keep with the green brush and the parts you don't want to keep with the red brush. The software is quite "intelligent” in that it is able to recognise, without explicit user input, parts of the object that should be kept.Maria E.