Mozilla Firefox Flips On AVIF Image Decoding By Default

[Quackdoc]

It seems you are assuming that AVIF is doing some sort of motion estimation when encoding animations, which is according to my knowledge incorrect.
AVIF is mostly just the still keyframe image compression part extracted from the AV1 standard (no particular video tricks are included).
They are also explicitely saying that AVIF is not particulary efficient at video encoding and suggest people use AV1 for that purpose.

you are wrong, AVIF is simply just an mp4 container under the hood, and AVIF animations are literally av1 videos, you can test it with this sample photo https://files.catbox.moe/gwt05k.avif

Open it in chrome and it should loop, rename it to .mp4 and it will play like a video in VLC or MPV. You can use AVIFenc to make a AVIF loop, but I prefer to make an AV1 video and use mp4box to set correct flags.

https://github.com/AOMediaCodec/libavif/issues/112 see the second comment

[lyamc]

The same person that had been working on FLIF is also working on JpegXL.

I know, I don’t care. JPEG XL introduces generational losses

[pkese]

I know, I don’t care. JPEG XL introduces generational losses

Please Google for "JPEG XL generational losses" then elaborate your claim.

[MadeUpName]

I mean without re-encoding. I would be very surprised if they weren't scaled down.

Exactly. People aren't blasting off 20MB jpeg files from their cell phones on a regular basis. But even if you did, as soon as it hits the platform they re-encode it with all the life squeezed out of it. I cry every time I upload a picture to Linkedin or Twitter and see the vandalism they do do my images

That doesn't get us to cameras where professionals mostly shoot in RAW and then need to convert to a consumer friendly format in post.

[pkese]

you are wrong, AVIF is simply just an mp4 container under the hood, and AVIF animations are literally av1 videos, you can test it with this sample photo https://files.catbox.moe/gwt05k.avif

Open it in chrome and it should loop, rename it to .mp4 and it will play like a video in VLC or MPV. You can use AVIFenc to make a AVIF loop, but I prefer to make an AV1 video and use mp4box to set correct flags.

https://github.com/AOMediaCodec/libavif/issues/112 see the second comment

You're describing container format.
But in the end it is still just a sequence of encoded keyframe (or delta P frame) images (which is what Jpeg XL is capable of doing as well).
The fact that it is stored in MP4 file does not imply that there's any motion estimation estimation being done or encoded.

[Quackdoc]

You're describing container format.
But in the end it is still just a sequence of encoded keyframe (or delta P frame) images (which is what Jpeg XL is capable of doing as well).
The fact that it is stored in MP4 file does not imply that there's any motion estimation estimation being done or encoded.

Again, AVIF animations are just MP4's encoded in AV1, with an avis brand. it is capable of anything an AV1 video is, JpegXL is missing a lot of the features that AV1 encoders have. Hence why they are superior for animations, Unless JpegXL has a fully featured encoder behind it, it will not have the same potential that AV1 will have. basically am AVIF is just a looping video, you can do anything you want to it as long as its something you can do with any av1 encoder (rav1e, svtav1, aomenc)

[pkese]

Again, AVIF animations are just MP4's encoded in AV1, with an avis brand. it is capable of anything an AV1 video is, JpegXL is missing a lot of the features that AV1 encoders have. Hence why they are superior for animations, Unless JpegXL has a fully featured encoder behind it, it will not have the same potential that AV1 will have. basically am AVIF is just a looping video, you can do anything you want to it as long as its something you can do with any av1 encoder (rav1e, svtav1, aomenc)

You shoud read some more:

AVIF allows encapsulating AV1 intra-frame coded content, thus taking advantage of excellent compression gains achieved by AV1 over predecessors.

Intra-frame or I-frame is the keyframe.
Apart from 'I' frames, AV1 video codec also allows for P (predicted frames) and B (bidirectional predicted frames) which is the motion estimatio stuff you're talking about.

AVIF however does not support those frame types. It is a subset of AV1 - basically just the I frames with a few extras.

[Quackdoc]

You shoud read some more:


Intra-frame or I-frame is the keyframe.
Apart from 'I' frames, AV1 video codec also allows for P (predicted frames) and B (bidirectional predicted frames) which is the motion estimatio stuff you're talking about.

AVIF however does not support those frame types. It is a subset of AV1 - basically just the I frames with a few extras.

using avif brand does not work with b/p frames, but using avis should

[lyamc]

Please Google for "JPEG XL generational losses" then elaborate your claim.

https://youtu.be/ytI7Yn_ShXk

It demonstrates the generational loss that occurs from repeated modification to the source (such as how memes are currently shared)

[xorbe]

We have webm, why avif now? (Serious question) And my research led me to webp2 and jpeg xl ...