|
From: | Peter Hillman |
Subject: | Re: [Openexr-devel] EXR B44A Usage |
Date: | Wed, 5 Oct 2016 10:57:38 +1300 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.3.0 |
Hi Chad, I'm afraid I only have an unhelpful answer for you: it's best to test and work out what works best for you. I reckon the five important metrics in picking a compression scheme are:
Compatibility could be an issue for the DreamWorks schemes, DWAA
and DWAB, as it was introduced in OpenEXR-2.2.0 and may not be
supported in all your tools yet. The other metrics depend on the
kind of images you are creating, your software, and your system
setup. Compression schemes behave differently depending on the
image content and the rendering algorithm. Also, how you choose to
trade off time against storage space and image quality is really
up to you, so it's hard for anyone else to give a definitive
answer. Given that, I'd suggest running your own tests to decide. You might try all the schemes with exactly the same image sequence and compositing setup, measuring the total time it takes to render the images, the compositing time, and the file size. Beware that timing tends to be quite variable for all sorts of incidental reasons that may not be anything to do with the compression scheme. For lossy compression, you could check the final images look OK to you compared to a lossless compression approach (e.g. ZIP). Personally I'd avoid lossy compression on images that will be
further processed unless there's a clear win elsewhere that's
important to you. I'm guessing that "Lossy 16-bit float, in blocks
of 16 scan lines" is the "B44" scheme. "B44A" may well be a better
choice than B44, but it would be worth running tests of all
options anyway. Peter On 29/09/16 03:00, Chad Ashley wrote:
|
[Prev in Thread] | Current Thread | [Next in Thread] |