[XBLA] Trials HD
- Thread starter DuckThor Evil
- Start date
Silent_Buddha
Legend
Very impressive and well deserved for an excellant XBLA title with little to no advertising.
Regards,
SB
Regards,
SB
Dungeonscaper
Newcomer
I hope they make a ps3 version, I played the hell out of the 360 version but it's dead now
I started playing this the other day again because my son (2yo) is totally into motorcycles and mopeds, so I thought he'd enjoy watching this. Yesterday I tried a level I silvered a while ago twice and got through it the second time with a really good time, putting me way above my friends, at about 50.000th in the global rankings. I figure that's probably out of a million plus so not bad?
Last year most Summer of Arcade games were bigger brands and had bigger marketing campaigns than Trials HD did (Marvel vs Capcom, Turtles, Shadow Complex). Epic's Shadow Complex campaign was biggest ever in the XBLA game scale. Shadow Complex started very strong and broke all the previous XBLA sales records. But it was a real surprise to all of us that Trials HD started to beat Shadow Complex in weekly sales charts in just a few months after the launch of both games, and now (year after) we have sold over twice the amount of copies. The continued sales are most likely caused by the many awards we have won and the visibility caused by those awards in different publications and around the net (Microsoft's XBLA best game of 2009 and best innovation of 2009 were the most important awards). The continued title update and DLC support has also brought the existing players back to the game many times. You could say that players are our most vital marketing tool
In technical/gameplay side of things, I'll say that the user content creation & sharing in one of the major things that has kept the game fresh for so long time. The track & ranking leaderboards naturally being the other most important thing (along with the real time ingame friend comparison meter).
Currently we have around 1.2 million players in the Trials HD leaderboard. According to GamerBytes (FADE LLC) sales analysis game is currently in third place of all time most sold XBLA games (counted by revenue). Battlefield is in second place (only slightly ahead), and Castle Crashers is holding the crown (like always).
We just received the top 100 user tracks (several thousands were submitted to the competition), and the quality looks very promising. The new DLC's going to be a blast!
I can't really talk about our future titles yet. However I can spill some beans about our future technology as always.Now give us the 4 player version
In Trials HD most of our object textures were limited to 512x512 because we just couldn't fit larger ones in the memory (all our materials have two texture layers, as we have per pixel specular, glossiness, ambient, etc). With our new technology we use mostly 2048x2048 textures (16 times more pixels) in our objects. Our memory footprint is still much smaller than in Trials HD thanks to our virtual texturing system. You can zoom really close to the surfaces and things do not get blurry. Basically the detail could be as high as the artist wanted, as the CPU/GPU processing time and memory usage stays constant no matter how detailed the surfaces are. The highest detail pixels are only loaded when needed. Sadly the storage still sets some limits (HDD space is limited).
In the end the Trials HD industrial halls looked a bit too repeating for our taste. Our new engine has a really sophisticated decaling system that allows us to have really unique detail everywhere (both textures and geometry, since our decals can also have displacement maps). The decals and lighting are not baked in the HDD virtual texture blocks like id software did in Rage. Instead the decals are baked to the virtual texture pages during runtime (this is much faster than applying decals on top of everything every frame, since the pages can be reused for hundreds of times). This saves a lot of HDD space, and allows us to add decals and modify lighting in real time (both features are very important for user created content). We are still running 60 fps (vsynch locked) and have a fully dynamic lighting (deferred light prepass this time instead of full deferred shading). Virtual texturing + fully dynamic lighting is entirely doable. id software choose to use baked lighting in Rage, but that's definitely not a limitation of the technology. Virtual texturing makes light baking cheaper to implement, so I understand perfectly the choice id-software made in Rage. Baked global illumination lighting definitely looks better and is faster (runtime) than a simpler looking fully dynamic lighting solution... If the game is mostly static environment, and there is no day/night cycle, then baking lights is the perfect choice (for both visual quality and for frame rate).
Last edited by a moderator:
Great update, sebbbi. I can't wait to see it! Summer of arcade 2011?
Why did you choose to go with a light prepass vs fully deferred lighting? The virtual texturing sounds like a really nice image quality boost. Definitely looking forward to seeing it. Trials HD was already a fantastic visual game.
Why did you choose to go with a light prepass vs fully deferred lighting? The virtual texturing sounds like a really nice image quality boost. Definitely looking forward to seeing it. Trials HD was already a fantastic visual game.
Most likely we will have no loading screens in our next game. And if we need to load something before the level can be started, we will somehow mask it behind the level startup screen.
Around 75% of our content (megabytes) in Trials HD was texture data. With the virtual texturing system all the texture data is streamed. Our virtual texture system also streams displacement maps, so the geometry data size will be reduced nicely. But we also stream the geometry data on demand and naturally the music is streamed also. The background data streaming will allow us to have (near) zero loading times just like in Trials HD, but it also allows us to basically create unlimited amount of new content (new DLCs for example). In Trials HD everything was loaded to the main memory at startup, and that really limited the expansion possibilities of the game.
I can't really discuss about our future/unannounced projects. The new technology will likely we used in several projects down the road.
Light prepass requires one less g-buffer texture fetch in the lighting shader. That's basically the main reason of the change. In Trials HD we we split lights to screen space tiles (similar system used in Uncharted). That system didn't suffer that much from the extra g-buffer texture fetch, since the lighting passes were reduced dramatically. The new tech uses a light geometry based approach (like Trials 2 SE for PC did). Light geometry based system is better in most cases since it culls light areas more precisely to the pixels affected (using stencil masking and depth comparisons). For example lights areas that are behind an obstacle are completely masked out and lights that are float nearby do not need to be calculated for the background pixels. We now also combine overlapping light areas together to reduce the bandwidth.
It was fun to watch Carmack showing their virtual texturing implementation running on iPhone at 60 fps. I have heard comments claiming that id software had to use baked lighting in Rage because the virtual texturing uses so much processing time. The iPhone port really proved that the virtual texture system is really efficient. I don't know exactly id software's virtual texture system's inner workings (all the optimizations they choose to use), but in our system I have noticed that the performance required for the indirection texture pixel lookups is really minimal. Indirection texture lookup itself is really cache and bandwidth friendly (it's sampled with +7 mip lod bias in our case, so huge amount of pixels sample the same texel). The performance gain that we didn't expect comes from the fact that hidden surfaces are not loaded to the virtual texture cache, and thus have no proper texture data. In this case the system uses lower resolution version of the data (basically smaller mip levels that contain the same texture area). Lower resolution data is naturally more cache and memory bandwidth friendly to sample. This saves performance and has no negative effect to visual quality (since the hidden surfaces are never visible).
Oh, man, there's so much side-scrolling greatness running through my mind right now.
Thanks for all the tech details.
Ya know, I played the demo for this last year at it didn't really gel with me for some reason (sorry). Since I've got a bunch of points incoming in the next week or so I thought I'd give it another go and loved it. Surprised it hasn't come down to 800 points in the intervening year or so but WTH, another 400 ain't gonna kill me.
I finally downloaded and played the demo last night. This game has that wonderful "just one more try" appeal to it. I actually beat a few of our regulars times on my friends list, but couldn't come close to RobertR1.
I will definitely be buying the full copy as soon as I get more MS points.
Great job sebbbi. I look forward to your next project.
I will definitely be buying the full copy as soon as I get more MS points.
Great job sebbbi. I look forward to your next project.
(offtopic) It's MotoHeroz WiiWare game. Well ok I don't know if Sebbbi is involved.
Similar threads
- Deleted member 86764
- Games