Fix the issue of determining animated WebP

[nichbar]

The condition of determining animated WebP in BigImageViewer might be wrong.

Based on what google said in WebP Container Specification#Extended File Format, animation bit is in byte 21 instead of byte 17.

You may mistaken ChunkHeader(VP8X) as a 4-bytes data while it's 8 bytes instead.

[nichbar]

https://github.com/Piasy/BigImageViewer/issues/188 might be related .

[Piasy]

Hi @nichbar , thanks for your contribution!

But why do you think we should check header[20] instead of header[16]? In Extended File Format section of the doc, it says 12 bytes of WebP file header (12 bytes), then 4 bytes of ChunkHeader('VP8X'), then the following byte is bit mask byte, which should be header[16], right?

[nichbar]

@Piasy 新年好！

Sorry for the late response and undetailed explanation.

In Extended File Format section of the doc, it says 12 bytes of WebP file header (12 bytes), then 4 bytes of Chucker('VP8X')...

Here's the reason why we should check header[20] instead of header[16].

Based on what google said on https://chromium.googlesource.com/webm/libwebp/+/master/doc/webp-container-spec.txt#132, what follow Chucker('VP8X') is another 4 bytes of what called chunk_size.

So the bit mask byte should be located at header[20].

This stackoverflow answer share the same idea.

The way that fresco use to determain animated webp support this idea too.

[nichbar]

These webp files can be used for testing.

Static WebP

Animated WebP

[Piasy]

Hmm, interesting, seems the official doc is not very straight forward, anyway, thank you for the contribution!

[Piasy]

v1.7.1 is released, check it out :)