A utility to analyze the algorithms other applications use to resize (resample, scale) images.
Fetching latest commit…
Cannot retrieve the latest commit at this time
ResampleScope Version 1.1 Copyright (C) 2011 Jason Summers This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see <http://www.gnu.org/licenses/>. ResampleScope is a utility to help analyze the image scaling algorithms used by other applications. You'll have to know what various resampling filters look like -- it will show you a picture of the filter, but it won't try to name it. How to build ------------ libgd (with PNG support) is required. Linux (etc.): Try running "make". Windows: Although it can probably be built as a Windows console application, no project files are included. I suggest using Cygwin. How to use ---------- First, run "rscope -gen" to generate the source images. It will suggest the sizes to resize the images to, but in some cases you can use other sizes. Then, load them into an application, resize them, and save the resized images in PNG format. (Take a screenshot if you're testing a web browser or other application that doesn't let you save resized images.) Then, run "rscope <resized-image.png> <output-image.png>" on the resized images. It will write an image file containing a picture of the resampling filter used. For a list of options, run rscope with no parameters. Notes ----- If the application does any gamma correction, you'll have to find a way to turn off that feature. If you don't, the resulting graphs will be very obviously warped vertically. ResampleScope only works with simple 1-dimensional ("separable") scaling algorithms. All of the common algorithms (Lanczos, Mitchell, any kind of "cubic" algorithm, etc.) are of this type. ResampleScope only works with "nonadaptive" algorithms. If, for example, an application runs a sharpening filter on the image after scaling it, ResampleScope won't produce meaningful results. By default, ResampleScope analyzes the algorithm that was used to scale the image in the horizontal direction. To analyze the vertical dimension instead, generate new pattern files by running "rscope -r -gen", and include the -r option when you analyze the images. Downscaling is harder to analyze than upscaling. ResampleScope works best if you reduce the size as little as possible (but by at least 2 pixels). But unfortunately, you can't trust that an application uses the same algorithm for all scale factors, so sometimes you have to try some larger factors. Also, some algorithms can only be distingished at larger scale factors. If, for the "dots" pattern (pd.png), you get a picture that is very broken up horizontally, you probably need to use the "-sf" or "-ff" option. This is needed when the application scales the image *features* at a factor that is not exactly the same as the scale factor of the image itself. The "dots" pattern is extremely sensitive to this, and it requires the scale factor to be known very precisely. When using the "line" pattern (pl.png), ResampleScope prints an "area". Normally, this should be very close to 1.0, but there are a number of (good and bad) reasons that it might not be. The most common reason is that the application scaled the image features at an unexpected scaling factor, so you should use "-sf" or "-ff", as described in the preceding paragraph.