I know you all immediately wondered, better compression?. We’re already working on that. And parallel encoding/decoding, too! Just like this update, we want to make sure we do it right.
We expect the next PNG update (Fourth Edition) to be short. It will improve HDR & Standard Dynamic Range (SDR) interoperability. While we work on that, we’ll be researching compression updates for PNG Fifth Edition.
One thing I’d like to see from image formats and libraries is better support for very high resolution images. Like, images where you’re zooming into and out of a very large, high-resolution image and probably only looking at a small part of the image at any given point.
I was playing around with some high resolution images a bit back, and I was quite surprised to find how poor the situation is. Try viewing a very high resolution PNG in your favorite image-viewing program, and it’ll probably choke.
At least on Linux, it looks like the standard native image viewers don’t do a great job here, and as best I can tell, the norm is to use web-based viewers. These deal with poor image format support support for high resolutions by generating versions of the image at multiple pre-scaled levels and then slicing the image into tiles, saving each tile as a separate image, so that a web browser just pulls down a handful of appropriate tiles from a web server. Viewers and library APIs need to be able to work with the image without having to decode the whole image.
gliv used to do very smooth GPU-accelerated panning and zooming — I’d like to be able to do the same for very high-resolution images, decoding and loading visible data into video memory as required.
The only image format I could find that seemed to do reasonably well was pyramidal TIFF.
I would guess that better parallel encoding and decoding support is likely associated with solving this, since limiting the portion of the image that one needs to decode is probably necessary both for parallel decoding and for efficient high-resolution processing.
Again, that would be TIFF. TIFF images can be encoded either with each line compressed separately or with rectangular tiles compressed separately, and separately compressed blocks can be read and decompressed in parallel. I have some >100GiB TIFFs containing elevation maps for entire countries, and my very old laptop can happily zoom and pan around in them with virtually no delay.
One thing I’d like to see from image formats and libraries is better support for very high resolution images. Like, images where you’re zooming into and out of a very large, high-resolution image and probably only looking at a small part of the image at any given point.
I was playing around with some high resolution images a bit back, and I was quite surprised to find how poor the situation is. Try viewing a very high resolution PNG in your favorite image-viewing program, and it’ll probably choke.
At least on Linux, it looks like the standard native image viewers don’t do a great job here, and as best I can tell, the norm is to use web-based viewers. These deal with poor image format support support for high resolutions by generating versions of the image at multiple pre-scaled levels and then slicing the image into tiles, saving each tile as a separate image, so that a web browser just pulls down a handful of appropriate tiles from a web server. Viewers and library APIs need to be able to work with the image without having to decode the whole image.
glivused to do very smooth GPU-accelerated panning and zooming — I’d like to be able to do the same for very high-resolution images, decoding and loading visible data into video memory as required.The only image format I could find that seemed to do reasonably well was pyramidal TIFF.
I would guess that better parallel encoding and decoding support is likely associated with solving this, since limiting the portion of the image that one needs to decode is probably necessary both for parallel decoding and for efficient high-resolution processing.
There is a reason why TIFF is one of the most popular formats for raster geographic datasets :)
deleted by creator
Yeah, I have a couple over 800MB PNGs that I can only get Gimp to open properly. I need to look into pyramidal TIFFs.
deleted by creator
Again, that would be TIFF. TIFF images can be encoded either with each line compressed separately or with rectangular tiles compressed separately, and separately compressed blocks can be read and decompressed in parallel. I have some >100GiB TIFFs containing elevation maps for entire countries, and my very old laptop can happily zoom and pan around in them with virtually no delay.