Dan's crazy, don't believe him. How can it take them 6 months to make a single car when in fact they have what, 400 staff? There's some cars that were remodeled and new cars added, yes, but not all. He's crazy...
Yes. I mean, logically, let's break down what GT6 currently:Random question: do you guys think that all of the tracks should have weather and time change by GT7? Just want to hear your thoughts on it.
Not exactly. They created another studio outside of Tokyo. The plan seems to eventually get the whole business there. They are expanding, but if you look at the type of people they have been hiring, it is not all modellers:
@Imari. To tell you the truth, that's impossible. 900 premiums in three years? You must be crazy too.
True.
👍
and fiddle with them all the way to the end (after months we feel our hero needs a change of cloths...). There is also a very healthy reuse of skeletons and core dimensions to replicate other characters including many NPCs (sharing lower LOD models), saving time. And though a car or more "static" object follows similar process, its clear PD take this to an extreme. For instance if you super impose a real photo over a PD model you'll notice some remarkable details like exact gaps between the tire and wheel wells, antenna angles, door handles, etc. and even looks more accurate than laser scans! (Yes. Laser scanning isn't perfect) and this is not limited to premiums! Take a Turn 10 model and do the same and you'll feel the opposite... Use a Pagani for example.
WOW! And you said you wrote all of that on a phone? Incredible! O_O I really enjoyed reading your post @Lawndart. It was very informative, thanks a lot. 👍There certainly is a lot of thrashing here. First off, Turn 10 do a lot of outsourcing. A year ago I know that T10 core team was quite small if you exclude outsource partners and playground games... So I won't elaborate as I know outsourcing very well. I think PD are all (mostly) in house.
Its hard to know for sure but I do remember the 6 month statement Kaz made awoke back (GT 4 perhaps?) and I would guess this is the most extreme case as it was mentioned to reenforce their quality, not as much to excuse their timing (perhaps both). And PD's cars are expecting in detail, more realistic than any game and in order to do that consistently (games have very inconsistent art due to different artists) means they are modeling a whole lot more than what we see. I wouldn't be surprised if all the suspension, frame, cross members, etc. are also modeled. Though some detail is likely stripped out when it gets "optimized" for the game.
GT 6 will require a 30 gig footprint that is a forced install so they are already downloading much of the high res assets, to address another point made earlier in the thread. And because blu-ray drives are very slow compared to a DVD drive (limiting streaming capabilities).
Things have changed in recent years but with the wide use of normal mapping for high res textures processes have remained similar. But in the past we had to do very high res models, then numerous lower res models for LOD. There are tools that do the lower les models procedurally now as well as improved rendering "tricks". The point being it does take about the same time to model a 500k poly model as it did 5-6 years ago to make a 5-10k model. But there is a detail limit based on resolution of TV's back then so anything over 5k was lost in real time environments. No longer an issue.
That being said, I'm reasonably certain PD aren't spending 6 months per car... Maybe a few but the ROI is not there to do every car that way and that's just premiums.
CAD data doesn't really save you time either, it just gives you a good foundation but all cad data needs to be redone as it's horribly inefficient. I also build unique real race cars and we use cad (great for engineers building real world parts) and I'm contemplating doing a crowd sourced build (thanks for the opening for the shameless plug!) but when I have an artist make me a render of my cad car he will still need to build it mostly from scratch so it can be animated for videos.
Though I can't answer exactly as I'm not PD I'll give you guys an example for context and I know I'm not the only developer in these forums as I can tell by some people's replies in other threads (we know who we are), so they may jump in also.
Building a character asset for example is a 4-6 week process to model all LODs (level of detail), texture all LOD, rig it for animation, all facial bones (mostly procedural these days), etc. this doesn't include feedback and revisions which can easily double this number... Sometimes you might even add 2-4 more weeks if its a player character (not NPC) but many developers will claim they are never done with main characters
So the best conclusion I can draw is PD do "more" than most and take their sweet time to get it right (technical term sic) and this does show in the final product, if you are only measuring quality of assets.
On a more personal note, I believe this is the leading reason they struggle with damage. They build their models so accurate that they didn't plan ahead to account for damage beyond a very simple "morph" of textures. But they did do this for the rally cars (oddly absent in GT 6) and likely found it took twice as long because of their demanding quality benchmark and simply pulled the plug? Now with tesilation they could pull this off a bit easier so perhaps GT 7 will have more damage. But it will have to be a procedural system, and it's not easy.
In conclusion (since this is already long and I'm typing on a phone):
I hope this was informative, I've taken the liberty to add some of my own thoughts to the larger debates as I felt is was appropriate in the context of discussion.
- PD don't spend 6 month on ever car, that point was for marketing.
- They have spent 6 months on certain cars.
- 6 man months per car? Probably more actually but that's entirely different math, implying multiple people as animator a and character builders are different.
- PD have been doing models near premium quality in development since GT 3 but the source model data is to inefficient to use in real time and they are likely still looking to convert it over (fix standards) as they won't shrink their car numbers and they have potential to add hundreds more from other GT games.
- GT 5 was likely the first game they could include newer near source quality model data in real time but this isn't perfect yet and rather inefficient so we only get 8-12 car grids. Authenticity is more important than numbers on screen.
- GT 7 won't be done in a year unless its a 6.5, Kaz even eluded to that "port" possibility at the Jalopnik film festival.
- GT's track modeling is handled much more traditionally than their car modeling so they can build tracks quickly. And also have many from previous games.
- GTs priority is likely on lighting going forward more than detail as any artist will tell you "it's all about the lighting" and this will make even simple assets look brilliant and more believable in game.
- GT will never be the traditional game some of our younger forum posters want it to be, it's steeped in legacy and not built by gamers. They will always explore different things but they aren't likely to match op to expectations derived from other racing games.
- GT will continue pursuing real world partnerships and this is what helps keep GT a sophisticated product over a more traditional game, I just hope a real racing series universally accepted by all auto manufactures get into the mix...
- Kaz is the leading pioneer bringing automotive and gaming together, something we all get to enjoy the fruits of and I might add a contradiction, there are others but Kaz is way ahead and his battle is new industry vs old. I'm going the other way and making real unique cars that have huds like games in them (ray traced on windscreens)... And its a real David vs Goliath circumstance.
There certainly is a lot of thrashing here. First off, Turn 10 do a lot of outsourcing. A year ago I know that T10 core team was quite small if you exclude outsource partners and playground games... So I won't elaborate as I know outsourcing very well. I think PD are all (mostly) in house.
Its hard to know for sure but I do remember the 6 month statement Kaz made awoke back (GT 4 perhaps?) and I would guess this is the most extreme case as it was mentioned to reenforce their quality, not as much to excuse their timing (perhaps both). And PD's cars are expecting in detail, more realistic than any game and in order to do that consistently (games have very inconsistent art due to different artists) means they are modeling a whole lot more than what we see. I wouldn't be surprised if all the suspension, frame, cross members, etc. are also modeled. Though some detail is likely stripped out when it gets "optimized" for the game.
GT 6 will require a 30 gig footprint that is a forced install so they are already downloading much of the high res assets, to address another point made earlier in the thread. And because blu-ray drives are very slow compared to a DVD drive (limiting streaming capabilities).
Things have changed in recent years but with the wide use of normal mapping for high res textures processes have remained similar. But in the past we had to do very high res models, then numerous lower res models for LOD. There are tools that do the lower les models procedurally now as well as improved rendering "tricks". The point being it does take about the same time to model a 500k poly model as it did 5-6 years ago to make a 5-10k model. But there is a detail limit based on resolution of TV's back then so anything over 5k was lost in real time environments. No longer an issue.
That being said, I'm reasonably certain PD aren't spending 6 months per car... Maybe a few but the ROI is not there to do every car that way and that's just premiums.
CAD data doesn't really save you time either, it just gives you a good foundation but all cad data needs to be redone as it's horribly inefficient. I also build unique real race cars and we use cad (great for engineers building real world parts) and I'm contemplating doing a crowd sourced build (thanks for the opening for the shameless plug!) but when I have an artist make me a render of my cad car he will still need to build it mostly from scratch so it can be animated for videos.
Though I can't answer exactly as I'm not PD I'll give you guys an example for context and I know I'm not the only developer in these forums as I can tell by some people's replies in other threads (we know who we are), so they may jump in also.
Building a character asset for example is a 4-6 week process to model all LODs (level of detail), texture all LOD, rig it for animation, all facial bones (mostly procedural these days), etc. this doesn't include feedback and revisions which can easily double this number... Sometimes you might even add 2-4 more weeks if its a player character (not NPC) but many developers will claim they are never done with main characters
So the best conclusion I can draw is PD do "more" than most and take their sweet time to get it right (technical term sic) and this does show in the final product, if you are only measuring quality of assets.
On a more personal note, I believe this is the leading reason they struggle with damage. They build their models so accurate that they didn't plan ahead to account for damage beyond a very simple "morph" of textures. But they did do this for the rally cars (oddly absent in GT 6) and likely found it took twice as long because of their demanding quality benchmark and simply pulled the plug? Now with tesilation they could pull this off a bit easier so perhaps GT 7 will have more damage. But it will have to be a procedural system, and it's not easy.
In conclusion (since this is already long and I'm typing on a phone):
I hope this was informative, I've taken the liberty to add some of my own thoughts to the larger debates as I felt is was appropriate in the context of discussion.
- PD don't spend 6 month on ever car, that point was for marketing.
- They have spent 6 months on certain cars.
- 6 man months per car? Probably more actually but that's entirely different math, implying multiple people as animator a and character builders are different.
- PD have been doing models near premium quality in development since GT 3 but the source model data is to inefficient to use in real time and they are likely still looking to convert it over (fix standards) as they won't shrink their car numbers and they have potential to add hundreds more from other GT games.
- GT 5 was likely the first game they could include newer near source quality model data in real time but this isn't perfect yet and rather inefficient so we only get 8-12 car grids. Authenticity is more important than numbers on screen.
- GT 7 won't be done in a year unless its a 6.5, Kaz even eluded to that "port" possibility at the Jalopnik film festival.
- GT's track modeling is handled much more traditionally than their car modeling so they can build tracks quickly. And also have many from previous games.
- GTs priority is likely on lighting going forward more than detail as any artist will tell you "it's all about the lighting" and this will make even simple assets look brilliant and more believable in game.
- GT will never be the traditional game some of our younger forum posters want it to be, it's steeped in legacy and not built by gamers. They will always explore different things but they aren't likely to match op to expectations derived from other racing games.
- GT will continue pursuing real world partnerships and this is what helps keep GT a sophisticated product over a more traditional game, I just hope a real racing series universally accepted by all auto manufactures get into the mix...
- Kaz is the leading pioneer bringing automotive and gaming together, something we all get to enjoy the fruits of and I might add a contradiction, there are others but Kaz is way ahead and his battle is new industry vs old. I'm going the other way and making real unique cars that have huds like games in them (ray traced on windscreens)... And its a real David vs Goliath circumstance.
