Software comparing quality between encode and original file.

[Chengbin]

Is there a software that can compare quality between 2 video files?

I feel like wasting a week worth of time encoding the Blu ray version of Kung Fu Panda to 4.3GB and 8.6GB x264 files using the absolute highest settings. I would like to check how close the quality of the encode is to the original Bluray source.

Dark Shikari, I really want the RDRC patch. I feel like wasting some time having fun encoding a video like your BBB episode.

[jj666]

Your eyes are always the best guide, as to be honest, it's really depending a lot on the quality of the equipment you are using to play the movie back. If you are projecting onto a massive screen, obviously you're not going to want to touch the original file. If it's a small LCD screen, you can probably get away with the two sizes you mentioned.

There are plenty of similar threads already on the forum which can be found using the search function, just don't get drowned in the statistics people will continually post instead of using their eyes ;-)

-jj-

[Chengbin]

Unforunately I don't have good equipments, that's the problem. I can't tell a difference between a Blu ray source and a x264 encode at 1/5 of the size (The Blu ray Cars) on my 22'' monitor. That's why I'm looking for a software that'll give me some percentages.

What should I type as keyword when I search?

[LoRd_MuldeR]

Better forget about the idea that some software can judge the quality of an encode for you!

While metrics like PSNR and SSIM are useful and required to tune certain aspects of an encoder, they are completely useless to judge overall quality. Especially with a Psy-optimized encoder like x264.

So the only useful software to "compare quality between 2 video files" is a software that displays the frames of two encodes interleaved, so you can easily compare them visually...

[Chengbin]

Then how does the codec comparison papers produce their numbers?

[Sagekilla]

They typically use metrics like PSNR, SSIM, etc. There are a few other metrics out there, but they're more or less the same as PSNR and SSIM: They tell how similar an image is to the source, but won't tell you if it's high quality. Interesting thing is, x264 used to favor a soft image when using trellis 2 as it was PSNR optimal. Now, with psy-trellis it favors a sharper image, which tends to decrease PSNR. Given two clips with psy-trellis and trellis, would you think the psy-trellis one is "lower quality" because it's PSNR score is lower? (Note: In psy-trellis vs trellis, bitrate hasn't changed)

That's the problem. Don't place too much trust in metrics. Use your eyes and figure out what you like best.

[Guest]

"When is PSNR applicable?"

http://forum.doom9.org/showthread.php?t=101862

[Chengbin]

I don't want to create another thread for this yes or no question. Can you put a decimal value in CRF? like 18.2?

[kemuri-_9]

--crf <float> Quality-based VBR (nominal QP)

yes you can.