The GraphicsMagick oss-fuzz build fully supports all of the libheif "Supported formats", including VVC (using vvenc/vvdec) and Uncompressed. To date, no oss-fuzz issues (build or fuzzing) have been encountered regarding VVC or Uncompressed, but this is likely due to a lack of such files in the provided corpus for heif. The VVC-related code has been building each day without fail.
I have no knowledge of VVC, but am impressed by the size and apparent rapid development of the code.
We need a way to develop a good HEIF corpus which includes small samples for all of the codecs. If GraphicsMagick had a HEIF writer, then I could make that happen, but it does not have a writer yet.
Assuming proper functioning of the instrumented (e.g. ASAN/UBSAN) oss-fuzz build, it seems possible to use the build product to create the associated corpus of good files during the oss-fuzz build. I am not aware of any projects doing such a thing, but why not?
Originally posted by @bobfriesenhahn in #1350
I agree that the fuzzing corpus should include (small) examples for all supported codecs / features.
Originally posted by @bobfriesenhahn in #1350
I agree that the fuzzing corpus should include (small) examples for all supported codecs / features.