Animated Webp size

[superdav]

I ran across a site http://littlesvr.ca/apng/gif_apng_webp.html highlighting the compression performance of GIF vs APNG vs WebP and sadly WebP was larger than each of the GIFs used in the test. Is gif2webp failing to optimize these images correctly? I've tested other gifs which seemed to preform much better than the ones used here. Are the images used in this test just not handled well be webp or do most images compress this poorly?

[oX Triangle]

if u have very very tiny pics (similar Gif-Size ~ 4KB aso) webp mostly bigger then gif

[oX Triangle]

an other reason is if you use transparance - then maybe webp create an alpha-plane (lossless)

try to skip alpha - or use a alpha-dither

[Urvang Joshi]

Hi,

There are a few aspects to note here:

1. The website you mentioned seems to show hand-picked examples where WebP seems to compress worse. In general, on a large set (about 7000) of  images, I've seen WebP to compress much better. Feel free to test on your own set of images, and let me know what results you get.

2.  gif2webp (with default options) is designed to balance (i) WebP file size and (ii) decode time in browsers while pausing/seeking forward animations. This is done by adding key-frames (full independently decodable frames) at regular intervals.

If you wanted only focus of WebP file size, you can avoid this addition of key-frames by using the "-kmin 0" option (see: https://developers.google.com/speed/webp/docs/gif2webp#options).

3. Meanwhile, I'm working on adding a "min_size" option to gif2webp as well, which further improve compression by optimizing the dispose method, blend method and change rectangles for each frame. I'll send an update once that's available.

[Pascal Massimino]

Hi,

On Fri, Oct 3, 2014 at 10:22 PM, Urvang Joshi <urv...@google.com> wrote:

Hi,

There are a few aspects to note here:

1. The website you mentioned seems to show hand-picked examples where WebP seems to compress worse. In general, on a large set (about 7000) of  images, I've seen WebP to compress much better. Feel free to test on your own set of images, and let me know what results you get.

2.  gif2webp (with default options) is designed to balance (i) WebP file size and (ii) decode time in browsers while pausing/seeking forward animations. This is done by adding key-frames (full independently decodable frames) at regular intervals.

If you wanted only focus of WebP file size, you can avoid this addition of key-frames by using the "-kmin 0" option (see: https://developers.google.com/speed/webp/docs/gif2webp#options).

3. Meanwhile, I'm working on adding a "min_size" option to gif2webp as well, which further improve compression by optimizing the dispose method, blend method and change rectangles for each frame. I'll send an update once that's available.

Note also that:

  - these apng files are encoded using Zopfli (-z2), which is quite slow (but 7zip (-z1) gives good results by default)

  - gif2webp.exe (http://www.rw-designer.com/gif2webp) gives good results:

BladeRunner.webp        191304 bytes

Contact.webp            466714 bytes

GenevaDrive.webp        220120 bytes

SteamEngine.webp        38932  bytes 

world-cup-2014-42.webp  36916  bytes

On Thursday, June 19, 2014 1:52:51 PM UTC-7, superdav wrote:

I ran across a site http://littlesvr.ca/apng/gif_apng_webp.html highlighting the compression performance of GIF vs APNG vs WebP and sadly WebP was larger than each of the GIFs used in the test.

Is gif2webp failing to optimize these images correctly? I've tested other gifs which seemed to preform much better than the ones used here. Are the images used in this test just not handled well be webp or do most images compress this poorly?

--
You received this message because you are subscribed to the Google Groups "WebP Discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to webp-discuss...@webmproject.org.
To post to this group, send email to webp-d...@webmproject.org.
Visit this group at http://groups.google.com/a/webmproject.org/group/webp-discuss/.
For more options, visit https://groups.google.com/a/webmproject.org/d/optout.