vladdanilov

11 ms·

61.
by vladdanilov 9y ago
I'm working on Optimage [1] for both lossless and visually lossless (lossy) optimizations. Right now it's available for Mac only. But I have a Linux version working internally as part of the upcoming Server plan. [1] http:/&
62.
by vladdanilov 9y ago
Conditions applied: 100% zoom and average viewing distance. The real challenge is consistency. Mozjpeg quality is manually tuned here to match the filesize of the rest.
63.
by vladdanilov 9y ago
SSIM being a global metric is not sufficient for many cases. Images with soft gradient tend to be overcompressed. MozJPEG is a good improvement. But its trellis cost model causes noticeable blurriness on fine details [1]. Compare with the o
64.
by vladdanilov 9y ago
Guetzli is great. Here's a better comparison (same filesize) done by one of the authors [1]. I'm working on a more practical tool [2] which produces comparable results [3]. [1] https://drive.google.com/drive/u
65.
by vladdanilov 9y ago
As someone interested in the field, this does not look much different from state-of-the-art video codecs, e.g. variable-size blocks, wavelet, predicator, arithmetic coding. Only that the predicator is trained on real data. But symbol dictio
66.
by vladdanilov 9y ago
There's no accent in the article that images should be optimized for the target screen dpi. It's wasteful to serve a 2x image for 1x screens and 1x when 2x is already present on the page. Almost all CDNs/services compress ima
67.
by vladdanilov 9y ago
It's not even a complete solution which is roots for a ninth-degree polynomial. Approximation with quadratics is what makes it feasible with some caveats [1]. [1] https://youtu.be/OmfliNQsk88?t=19m7s
68.
by vladdanilov 10y ago
Seems like most of the image optimizations are undone by WordPress/CDN, saved >60% on the xkcd-styled charts.
69.
by vladdanilov 10y ago
I'm still thinking on the best way to implement it. Right now you can mount an FTP/SFTP folder and process from there. But I might try to bypass the mounting part. Server version is also in the works.
70.
by vladdanilov 10y ago
It's block-based so definitely yes.
71.
by vladdanilov 10y ago
At the image upload rate nowadays, there's a place for a practical solution, and for any external company, that kind of cloud computing will cost a fortune.
72.
by vladdanilov 10y ago
I'm working on a similar thing ( http://getoptimage.com ). While Guetzli is still visually better and a bit smaller in file size, it's terribly slow and requires a lot of memory. But it's a great experiment. So much
73.
by vladdanilov 10y ago
Agree that APNG is a poor replacement for GIF being mostly used to share video scenes, and VP9, having all those properties, is a perfect one. The only problem is that VP9 is still lossy with RGB -> YUV -> DCT -> YUV -> RGB, sim
74.
by vladdanilov 10y ago
I'll let you know. There's a newsletter for major updates, or you can ping me via email (in profile).
75.
by vladdanilov 10y ago
After 3 months the visually lossless image optimization tool Optimage ( http://getoptimage.com ) I'm developing in my free time has started to bring ~200$/month. Hoping to resolve the last performance issues and launch a
76.
by vladdanilov 10y ago
I'm working on a visually lossless optimization tool [1] and could reduce the images a bit further from 20.8MB to 16.11MB (-22.6%). But you're right, hidpi images is the main cost factor, adding an srcset polyfill would be a good
77.
by vladdanilov 10y ago
It's great you listen. So I'll try. 1. Scaling down in linear colorspace is essential. One example is [1], where [2] is sRGB and [3] is linear. There are some canary images too [4]. 2. Plain bicubic filtering is not good anymore.
78.
by vladdanilov 10y ago
Agree. Serving the right size is as important.
79.
by vladdanilov 10y ago
The AI approach is interesting but there are just too many problems with it. MozJPEG is actually using something similar called trellis quantization or soft-thresholding. The idea is to further quantize DCT coefficients in noisy areas. Ther
80.
by vladdanilov 10y ago
Had to do it for the project website. The solution was to generate inline placeholder images and substitute original src attributes with them when DOM is loaded, then back on scroll and click events. The placeholder images are JS-generated
81.
by vladdanilov 10y ago
> The best way to counter negative effects of loading image assets is image compression: using tools such as Kornel Lesiński‘s ImageOptim, which utilizes great libraries like mozjpeg and pngquant, we can reduce image byte size without sa
82.
by vladdanilov 10y ago
The only way I could make it fairer is to add a magnifying area zooming to 100% but that would be too much work. Hoping that links to images is enough.
83.
by vladdanilov 10y ago
I discarded the idea of using a sliding divider because it cannot switch images fast enough to notice tiny compression artifacts, if any. But thanks for your feedback. I have a couple of ideas how to make it more clear.
84.
by vladdanilov 10y ago
Ahem… are you fully satisfied with your current solution?
85.
by vladdanilov 10y ago
Thanks. Literally a few. Contact me if you need specifics.
86.
by vladdanilov 10y ago
CLI is cross platform. I'm certainly planning a server license with an option to spin it on AWS. I have some performance updates planned making it 2-3 times faster. Probably worth waiting. Email is in the HN profile.
87.
by vladdanilov 10y ago
Sorry for copypasting, I get this question a lot. Optimage can do automatic lossy (visually lossless) optimizations on JPEGs and PNGs. It means you don't have to manually tune JPEG quality for each image and can get some PNGs quantized
88.
by vladdanilov 10y ago
Optimage – a visually lossless image optimization tool http://getoptimage.com Released this week. Came unnoticed on HN but somehow got featured on PH. 1200+ downloads of free version and a few sales. Not sure if that'll cha
89.
Show HN: Optimage – Visually Lossless Image Compression Tool(getoptimage.com)
4 points by vladdanilov 10y ago|0 comments
90.
Show HN: Optimage – Advanced Image Optimization Tool(getoptimage.com)
3 points by vladdanilov 10y ago|1 comments

More ›