How to get image dimensions from an avif file?

Question

I'm looking for a (fast) way to retrieve image metadata from an avif-file in my java-application. This is currently done for many image formats by analyzing the raw header bytes of a file.

I can recognize a file as avif through the magic byte sequence at the beginning of the file: ftypavif. But I'm not able to find any information how to parse the (max) image width & height of contained images. For JPEG2000 for example I can search for SIZ marker 0xFF51 and parse dimensions afterwards, but even after spending a lot of time reading through specifications of avif, I'm not able to find something similar for it.

Is my approach incorrect and I can't parse the image dimensions without decoding the image(s)? Or is there any documentation (or even usable library) I missed to retrieve width & height from an avif-file?

Any hints appreciated!

Here is a library implementation, https://github.com/AOMediaCodec/libavif and here is the standard. https://aomediacodec.github.io/av1-avif/#av1-item-configuration-property — matt, Feb 16 '21 at 10:45
Thanks for the links, but as stated above I've read through both of it without finding the required Information. Manually including a C-library in my java-application just to get some meta-information doesn't look ideal to me. Any additional hints? — Jochen Go, Feb 17 '21 at 06:57
"but as stated above I've read through both of it " you didn't state that anywhere. The links are the documentation and a library with a reference implementation. I'm sure the information is in there, but I don't know enough about reading the specs to quickly delve into it. Can you link an example image in your post? — matt, Feb 17 '21 at 09:21

Dimitri Podborski · Answer 1 · 2022-09-23T22:22:24.437

6

Search for 'ispe' box (this is a mandatory box/item), skip 4 bytes after the type (FullBox version and flags) read 4 bytes for image width, read 4 bytes for image height. Profit ;)

edited Sep 23 '22 at 22:22

answered Mar 09 '21 at 21:31

Dimitri Podborski

3,714
3
19
31

Remember to link to the relevant spec if you're going to make claims about byte layouts. – Mike 'Pomax' Kamermans Sep 13 '22 at 16:00
@Mike'Pomax'Kamermans ISOBMFF + HEIF. In the first you can find the generic definitions for Box/FullBox in the second the syntax of `ispe` itself. – Dimitri Podborski Sep 23 '22 at 18:50
I'm sure, but those specs have links, adding links is always good =) – Mike 'Pomax' Kamermans Sep 23 '22 at 20:56

score 2 · Answer 2 · answered Feb 17 '21 at 09:48

Some of the AVIF files have ftypavis (contains animation) and I also saw AVIF with ftypmif1

If you want to parse AVIF alone, you need to understand ISOBMFF format and HEIF container first. There are various properties (like Image spatial Extents (ispe)), then various transformations (crop, rotation, mirror). Images can be stored in a form of image grid...

How to get image dimensions from an avif file?

2 Answers2