- From: Marcos Caceres <w3c@marcosc.com>
- Date: Fri, 19 Oct 2012 11:39:04 +0100
- To: Matthew Wilcox <elvendil@gmail.com>
- Cc: Kornel <kornel@geekhood.net>, public-respimg@w3.org
Hi All, I want to thank everyone on input on this thread - I think we've all gained a lot from it, but we are reaching the point of diminishing returns (and we seem to be going around in circles a bit). An action item from this thread would be to reach out to both the WebP people and the JPEG-XR people and ask them how we can help (or if they see <picture> as helping). Matt, can I put you in charge of that? For everyone else, it would be preferable if we could refocus our energy on our group's current big goal/milestone: reviewing and providing feedback on the Use Cases and Requirements document before we forward it to the HTML Working Group and WHATWG: http://usecases.responsiveimages.org Kind regards, Marcos -- Marcos Caceres http://datadriven.com.au On Friday, 19 October 2012 at 11:34, Matthew Wilcox wrote: > On 19 October 2012 10:53, Kornel <kornel@geekhood.net (mailto:kornel@geekhood.net)> wrote: > > On 19 paź 2012, at 08:29, Matthew Wilcox <elvendil@gmail.com (mailto:elvendil@gmail.com)> wrote: > > > > > Though I would *love* alpha on lossy images. A number of designs from my work have to be turned back and edited because they're simply not practical with PNG alphas. > > > > PNG does support a form of lossy images with alpha — palette-quantised images can have varying degrees of transparency. > > I use this via ImgAlpha and ImgOptim on the Mac. It's only suitable > for limited colour pallet images and is not a solution appropriate for > photographic imagery with an alpha. That's the point. > > > The problem is that Photoshop doesn't support this, so many authors assume it's impossible. > > > > http://pngmini.com http://pngquant.org > > > > While it's not as good as WebP: > > > > http://pngmini.com/vs-webp/ > > > > IMHO for majority of cases it's "good enough". > > No it isn't. As noted; I've had to send designs back because they're > not practically achievable with the file formats we have to work with. > > > There's also potential to make 24-bit PNG smaller by removing information (e.g. by posterising and lossy application of PNG's filters). > > You're arguing about how to "optimise" an image for a file format that > is not the correct one for the job. > > > -- > > regards, Kornel >
Received on Friday, 19 October 2012 10:39:40 UTC