GT5: technical analysis of the demo

GT5: technical analysis of the demo
By the time you read these lines you will have already had a taste of GT on PS Network. You will have tested the gameplay, driving model and evaluated the aesthetic aspects of the Polyphony Digital title. But we, again thanks to our "geek" friends at Digital Foundry, can now offer you a technical overview of the new developments in the game. Before starting, we first present a very high definition analysis of some images and a succulent HD gameplay video, just to refresh your memory. Now let's see if images and videos will be able to pass our tests properly ...

There has been much discussion (also in our forum) on the fact that, in reality, the demo version released on the PS3 online channel is not a real demo. This claim could also be considered correct, in the sense that we are faced with only one circuit and a single timed test, just a crumb of all that the complete code will have to offer. However, the fact remains that in terms of graphics and gameplay the game is exactly what you have had the opportunity to try, even with the limitations of the case. You can then put your heart in peace and rely on the technical analysis presented in the videos below, as little or nothing should change in terms of performance in the final product.

The initial screen, just as mentioned above, informs us that this is a real GT5 demo, although the options are obviously limited. In terms of rendering specs, little has changed from GT Prologue's code. Set your console to 720p and you'll get native resolution with 4x multi-sampling anti-aliasing (MSAA). If instead you choose to set it to 1080p, the framebuffer will be at 1280x1080. In this way there is a 50% increase in resolution compared to 720p, with the consequent lowering of antialiasing to 2x. By now you know, the higher the resolution, the lower the performance of AA.

Very little has changed since GT5 Prologue. The upper images give an example of the 720p resolution, while the lower ones show the full 1080p resolution. Sooner or later we will get a contest on gender: find the differences. Regarding the contents, you already know that it is a single circuit, the historic Indianapolis, with only a couple of cars. Well, actually just one, the Nissan 370Z, presented for the occasion with different specifications. The riding options are very limited (gearbox, trajectory and traction), so that we can offer everyone the same playability and the same degree of challenge. Specifically, this is a simple time-trial, which however offers you the opportunity to understand what awaits you in terms of simulation and driving feeling.

In its simplicity, the demo shows the ideal terrain to put to test the capabilities of the graphics engine. Only two cars (one of these is the ghost) and a track, nothing that can tire the circuits of your PS3 too much. The tests performed by Digital Foundry are essentially based on two factors: the first is the view inside and outside the cockpit, the second is related to 720p and 1080p resolutions. In order to seriously test the graphics engine, the guys from DF tried wherever possible to keep close to the ghost car. Here is the test performed at 720p.

About 5% of the frames are "ripped", in a rate that sometimes dropped to 52FPS. Interestingly, the greatest impact on frame-rate can be seen in the moments when both the trackside fences and the ghost car are present on the screen simultaneously. However, the video clearly shows that this is a sporadic phenomenon, and by keeping an eye on the number of frames in the upper right you can well see that it is a rather stable 60FPS. Too bad for the view of the public, which certainly perplexes ...

Let's now move on to the resolution of 1080p, once again evaluating the performance based on the view inside or outside the cockpit.

The results are more or less the same with the usual drop to 52fps and an increase in the percentage of 'ripped' frames, which at 1080p rises to 12%. Overall, we find more or less the same performance seen at the time with GT Prologue. It remains to be seen if the latest code filings will be able to remedy these negligible gaps, but for the sake of truth we cannot fail to emphasize that already from the demo present at gamescom, the results have more or less remained unchanged. To be clear: barring any last minute surprises, the final game should look like we showed you in our HD videos.

We want to point out now Polyphony's interesting take on ghost car racing . Rather than presenting the car via a transparency effect, interlaced resolution is employed to reduce bandwidth and fill-rate. Keeping 60FPS steady is certainly hard work - Forza 3 is arguably the best example of the game breaking down alpha textures to an absolute minimum, but GT5 also uses similar tricks in order to keep the framerate stable. Both titles use the technique called "alpha to coverage", thanks to which remarkable outdoor performance can be achieved by presenting transparent polygons. The use of AA multi-sampling partially mitigates the effects of this technique.

As expected from a Polyphny Digital title, the aesthetic quality of the game is excellent. The writer is literally devouring Forza 3 and, in my humble opinion, this GT5 has a slightly higher degree of cleanliness of the cars. Forza 3, for its part, alternates memorable and flawless landscapes with an aliasing that is sometimes certainly more marked, especially in close-up shots at the beginning of the race. However, remember that in the Turn10 title, all cars are susceptible to damage. Horrendous, at least in the Indianapolis GT 5 track, is the reproduction of trees and foliage, which appear only as textures suspended in midair, brushstrokes among other things in a very bad way.

The race for the podium between the two contenders is still long and at the moment it is not possible to give definitive judgments, therefore we just have to wait to get our hands on the complete game to finally be able to offer you a detailed and concrete technical analysis on the entire work done by the Japanese team .

Powered by Blogger.