Test 1: Saving Private Ryan

This movie is known to be notoriously hard to encode and therefore ideally fits my needs. I chose 3 scenes here: The beach landing (frames 8430 - 22281), a still scene in the controller's office (43790 - 45189) that has a lot of film grain and a night scene (frames 94122 - 106098).

Scene 1: The beach

I don't think there's too much to be said about the beach scene. No steadycam, underwater shots, a lot of debris and confusion on the screen. In order to better see the difference between the different codecs, I broke down the scene into smaller bits which I'll now indicate by framenumbers: 8849 - 9242 (on the boat), 9242 - 9496 (the first salvo), 9496 - 9676 (view from the bunker), 9948 - 10520 (underwater), 10520 - 10858 (coming out of the water), 10858 - 11300 (getting to shore), 1130 - 11900 (getting to shore 2), 12148-12451 (view from the bunker 2), 13656 - 14252 (ringing in the ears), 14252 - 14900 (the boat).

While I don't have any shots from my next two review scenes (first salvo), I rated both ateme and x264 first because they did not exhibit any blocking, followed by XviD and finally DivX.

Then we go under water.

In this scene, I rated the two AVC codecs on the level with XviD and DivX following on the last place. As we come out of the water, x264 manages to surpass the ateme encoder for the first time by exhibiting less visual blocking as we come out of the water and go in again.

As we see the GIs approach the beach, the constant firing makes the scene look like it's raining. And as we've seen in Matrix3, no codec likes rain.

Before we get out of the water completely, I had a long hard look at all the metal bars in the water, and DivX has the most problems with those - the other codecs do all right.

During the ringing ears scene, you can see the most facial details from our two AVC contestants (with no particular preference for either one), followed by XviD.

Finally we have the close-up of Tom Hanks with the burning boat in the background.

At the end of the scene, it comes down to a tie between ateme and x264. XviD follows in third place and DivX comes in last.

Scene 2: In the office

I also divided the office scene into two review parts. 43790 - 44120, and 44120 until the end.

Scene 3: The Church

Our final scene takes place at night, and black has proven to be notoriously hard to encode in the past.

I've once again divided the scene into different subscenes: frames 95730 - 96070 (the letter), 98075 - 98474 (fallen comrades), 98600 - 98782 (shaking hands), 100580 - 101829 (walk across the church), 105029 - end.

In between the previous shot and the next I had another 4 review scenes. x264 and XviD came in first in the fallen comrades scene because the blocking in the ateme encoder was rather disturbing. In fact, that's a pattern I saw throughout the church. I even came across a situation (where we see Giovanni Ribsi speaking of how his mother came home and he pretended to be asleep), where I noticed significant changes in the block patter from one frame to another, which I found highly disturbing.

Finally, our soldiers moving out during the night:

This scene came as a surprise to me. Looking at the results of my 5 review subscenes, the tally comes up with x264 first, and XviD actually beating the Ateme encoder and being ranked second, leaving the 4th place to DivX.

Conclusion

Since the Matrix 3 review turned out to be rather tricky, I decided to consequently assign places to each review scene, and award points: 4 points for a first place, 3 for a second, 2 for a third and 1 for a fourth. You can already guess by the intermediate results, that x264 won this round hands down. It was an even fight with the ateme encoder until the night scene, but that's when the differences became apparent. While XviD trumped over Ateme in the night scene, in the end it still only made the third place in the end. I never rated the DivX results any higher than any other codecs, consequently it ended up on the fourth place.

Now you can proceed to the Steamboy or go directly to the conclusion.

This document was last updated on December 26, 2005