Not good enough for Google PageSpeed, but developer is awesome

It depends greatly on the individual image, but it’s worth noting that PSI is not a “gold standard” on anything in regards to image compression. It’s lower quality than every plugin you’ve tested, and flawed for retesting: https://docs.ewww.io/article/5-pagespeed-says-my-images-need-more-work

The primary difference is that PageSpeed simply drops the quality across the whole image to level 85, whereas all of the plugins you’ve mentioned use algorithms to preserve as much quality as possible.

I’m also not sure your testing mechanism is sound, if EWWW IO and TinyPNG give different results. EWWW IO uses the same exact compression technology as TinyPNG when set to the highest level. I’d be most interested in a link to the test image that you’re using.

Thanks for the quick reply. I’m using EWWW with “lossy” jpg compression and default quality level (box is empty, so I assume wp default of 82%). Unless I’m crazy?

Worth mentioning that TinyPNG gives a different result between the WP plugin and their online optimizer (the online optimizer produces a larger file).

Here is the image: https://s3-us-west-2.amazonaws.com/elasticbeanstalk-us-west-2-868470985522/ShareX/2018/02/original.jpg

It is not an ideal photo for testing compression, but is from a live website that I am trying to optimize.

Hey, the best tests are with actual images from live websites ??
I did confirm your results, so that was a bit surprising. We use TinyPNG on the “max lossy” setting, and it normally is about 5-10% better than regular lossy (but obviously that’s an average and not a hard and fast rule).

TinyPNG generally preserves more quality, so that probably accounts for the difference here. Their algorithm is slower, which usually ends up giving more compression AND higher quality, but not on that image.

I’ve been doing some testing with the ImageOptim API recently, and so I tested your image on that just for fun, and found some pretty good results. On the medium setting, it gets 76kb, and on the lower setting, it comes in at 51.2kb. Unfortunately, there’s only one plugin that uses ImageOptim, and it’s not a very good one. Teaser, that might change soon…

Ah look at that, impressive filesize difference for this photo with imageoptim. That’s interesting. Not sure if that’s what Google PageSpeed is using to reduce the filesize so far without a noticeable loss in quality. There’s a slightly noticeable quality loss when using imageoptim on low setting. But whoa, 50kb! Neat.

After your informative responses and investigation beyond my own issue, I see you’re doing some great work. I don’t know what will come of ImageOptim but if that helps EWWW at some point, I’m glad I might have at least sparked the idea!

I don’t know any other developer who is knowledgeable about their competitors and is willing to compare them on even ground and even defend them. That shows an impressive attitude.

I decided to add the missing star back in for your plugin. Sure, PSI somehow makes smaller images at this point, but I now feel more confident with EWWW just based on your communication here ??

Thanks!

• This reply was modified 7 years, 1 month ago by radgh.

PageSpeed is using a much less sophisticated method (because it’s faster), at least that’s what their devs claim. It’s basically ImageMagick at quality 85, with a final pass of jpegtran to optimize the encoding tables.

FYI, normally, I would expect both Imagify and TinyPNG to kick the crumbs out of PageSpeed on average. I once ran a test between PageSpeed’s method and TinyPNG on 30,000 images that were processed via our API, and TinyPNG came out ahead by 8-10%.

My smaller (limited to only 1,000 images) test suite showed that Imagify typically beats the compression of TinyPNG by about 4-5%, but TinyPNG has 3x less quality loss.