Problem: The ImageTester currently converts each PDF page into a single image/test. If changes occur (content added/removed) after baselines are created, the diffs cascade across downstream pages where the change or changes were introduced. e.g., content from page 3 is pushed to page 4, and so on.
Enhancement: Add an argument to group by X (e.g. 10 pages) number of pages. For example, the 1st 10 pages are combined into a single checkpoint, 11-21 into checkpoint 2, and so on. There could be logic added to calculate the number of pages per group by measuring the height of each page (they should be the same in theory). When it reaches 15k pixels, that would be the group. Furthermore, users can use ignore displacements to filter out displaced diffs, making it easier to see what changed.
This, of course, won't fix the issue entirely, as large PDFs will have issues at a certain grouped checkpoint. But it is an improvement of how it works now, IMO. There might be other things that could be done, such as lowering the resolution, converting the images to thumbnails, and thus allowing more pages to be grouped. Just some thoughts.
Problem: The ImageTester currently converts each PDF page into a single image/test. If changes occur (content added/removed) after baselines are created, the diffs cascade across downstream pages where the change or changes were introduced. e.g., content from page 3 is pushed to page 4, and so on.
Enhancement: Add an argument to group by X (e.g. 10 pages) number of pages. For example, the 1st 10 pages are combined into a single checkpoint, 11-21 into checkpoint 2, and so on. There could be logic added to calculate the number of pages per group by measuring the height of each page (they should be the same in theory). When it reaches 15k pixels, that would be the group. Furthermore, users can use ignore displacements to filter out displaced diffs, making it easier to see what changed.
This, of course, won't fix the issue entirely, as large PDFs will have issues at a certain grouped checkpoint. But it is an improvement of how it works now, IMO. There might be other things that could be done, such as lowering the resolution, converting the images to thumbnails, and thus allowing more pages to be grouped. Just some thoughts.