Star Citizen developer says ray tracing "a massive headache"

Published by

Click here to post a comment for Star Citizen developer says ray tracing "a massive headache" on our message forum
https://forums.guru3d.com/data/avatars/m/53/53598.jpg
ViperAnaf:

by the time this game is released it will be 3 more card generations in the future...
By the time this game is finished and released we will actually be living and working in space and i will have been frozen for over 500 hundred years!!!!
https://forums.guru3d.com/data/avatars/m/270/270233.jpg
This goes without saying. Until ray-tracing is available to the masses - and available in a way that will produce acceptable framerates for the masses - it will be an extra feature, requiring further work from the developers. Only those who were taken in my Nvidia's "it just works" marketing hype would have thought otherwise.
data/avatar/default/avatar08.webp
I much rather them focus on getting the Vulkan client out the door then fiddle around with RTX/DXR. The games looks spectacular without raytracing, but needs optimaization/fps improvements more then anything.
data/avatar/default/avatar15.webp
HardwareCaps:

Wait this con-project still exists?
I'm actually really enjoying the game.
https://forums.guru3d.com/data/avatars/m/64/64284.jpg
Thanks Hilbert...I dont know anyone that plays it. It always looks as if NO LIFE in the game.
data/avatar/default/avatar12.webp
fantaskarsef:

Uhm.. the very question of the user on the forums is questionable in terms of being meant for serious... Does RTX save you work... no, how should it? Basically reprogramming most of your lighting... 🙄
RTX DOES Saves work, there is a reason Huang was repeating it just works, because engines that support it just toggle it on/off, no need to optimize lighting or shadows or whatever.
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
MegaFalloutFan:

RTX DOES Saves work, there is a reason Huang was repeating it just works, because engines that support it just toggle it on/off, no need to optimize lighting or shadows or whatever.
At the current state of any game developed that already runs (alpha), it does not. In general, it might help, but no game that started development already saves time by revamping all lighting... that's a no brainer.
data/avatar/default/avatar19.webp
nwc:

I'm actually really enjoying the game.
I thought it was dead tbh, I'm glad people at least enjoy it....
data/avatar/default/avatar40.webp
fantaskarsef:

At the current state of any game developed that already runs (alpha), it does not. In general, it might help, but no game that started development already saves time by revamping all lighting... that's a no brainer.
This is irrelevant you said that RTX doesn't save you work, no one was talking about non rtx here. If you use strictly RTX, it does save work. Actually it saves work anyway, here is example: You turn on RTX to see how proper shadows and lighting should look and based on this you can tweak non RTX mode to be as realistic as possible.
https://forums.guru3d.com/data/avatars/m/80/80129.jpg
MegaFalloutFan:

This is irrelevant you said that RTX doesn't save you work, no one was talking about non rtx here. If you use strictly RTX, it does save work. Actually it saves work anyway, here is example: You turn on RTX to see how proper shadows and lighting should look and based on this you can tweak non RTX mode to be as realistic as possible.
There is a difference between "it just works" and "this tool can save you work" "it just works" implies that you flip the switch and it's simply 100% done. That isn't the case. For performance reasons it only works with a subset of graphics effects - so right off the bat you need "non-rtx" lighting at some point to carry the rest of the scene. For the same reason (performance) we've seen DICE do a number of under the hood changes to way DXR works, at a basic level, to even get their subset effect to work at 60fps. On top of that the implementation in Unreal, for example, is missing several effects (no caustics), is very noisy and I believe only does one bounce. So yeah maybe at some point RTX running on a full pathtracer will do "it just works" but right now, at least for games, it's not even close. Also you say "no one is talking about non-RTX" here but the thread and the post that @fantaskarsef was referencing is literally about a game that uses non-RTX and asking whether using RTX going forward will save them work. The answer is clearly no. As for helping them accomplish things - maybe? As I wrote earlier in the thread, the lighting system in SC appears to be extremely complex to facilitate the multiple different environments the player encounters seamlessly. It's not like dropping an area light into an unreal map and saying "hooray!" - for example there is all kinds of lighting fancy pants stuff being simulated in the sky during descent onto the planet. I'm not sure RT would replicate any of that right off the bat without significantly changing the way the sky/planetary system works. Yeah it may help you simulate some interior lighting or whatever but even that is a gamble that requires a ton of tweaking and in the end it's typically pointless because @Fox2232 will come in and point out every misplaced light particle lol. I love the tech but Starcitizen might be the one of the worst games to try it in.
https://forums.guru3d.com/data/avatars/m/56/56686.jpg
with all the R&D money they got from crowdfunding they can do this I still think this whole game is never gona be released and yet they still getting crowd funding, I lost intrest in long time ago.
TheDeeGee:

I think they got a headache trying count all the cash.
like said above with all R&D the got from crowdfunding and last I checked there still getting more from people, and it still not done? something fishy imo dont get me wrong i was interested they moment this was announced cause wing commander franchise , but now it just like they trying to see how long they keep getting money and how long they keep the game in "development" before people get fed up
data/avatar/default/avatar28.webp
Well said, I believe most devs share similar opinions.
data/avatar/default/avatar35.webp
Denial:

There is a difference between "it just works" and "this tool can save you work" "it just works" implies that you flip the switch and it's simply 100% done. That isn't the case. For performance reasons it only works with a subset of graphics effects - so right off the bat you need "non-rtx" lighting at some point to carry the rest of the scene. For the same reason (performance) we've seen DICE do a number of under the hood changes to way DXR works, at a basic level, to even get their subset effect to work at 60fps. On top of that the implementation in Unreal, for example, is missing several effects (no caustics), is very noisy and I believe only does one bounce. So yeah maybe at some point RTX running on a full pathtracer will do "it just works" but right now, at least for games, it's not even close. Also you say "no one is talking about non-RTX" here but the thread and the post that @fantaskarsef was referencing is literally about a game that uses non-RTX and asking whether using RTX going forward will save them work. The answer is clearly no. As for helping them accomplish things - maybe? As I wrote earlier in the thread, the lighting system in SC appears to be extremely complex to facilitate the multiple different environments the player encounters seamlessly. It's not like dropping an area light into an unreal map and saying "hooray!" - for example there is all kinds of lighting fancy pants stuff being simulated in the sky during descent onto the planet. I'm not sure RT would replicate any of that right off the bat without significantly changing the way the sky/planetary system works. Yeah it may help you simulate some interior lighting or whatever but even that is a gamble that requires a ton of tweaking and in the end it's typically pointless because @Fox2232 will come in and point out every misplaced light particle lol. I love the tech but Starcitizen might be the one of the worst games to try it in.
BFv only does shadows, Metro Exodus does Global Illumination with RTX and it has no regualr light effects when RTX is enabled, so it just works. Performance optimizations are growing pains and done by nvidia and then shared to everyone, they doing the optimizations to better utilize what hardware they have right now. As far as ray tracing helping with non ray traced game, its been done before, just RT took tons of time and was expensive, now every indie with 2080ti can do it, he doesn't need 60fps to get a reference image, it can be 6 times slower running 10fps but in 6 times the quality Im not sure you can use dynamic lightning data for reference, but say room with static light, you can see how it looks Ray traced ie Realistic and then when you know how it actually is you can fake it 1:1, without ray tracing fake lightning is magic, its just artistic representation and effects but if you have RT reference, you can try to make your fake look as identical as possible.
https://forums.guru3d.com/data/avatars/m/243/243702.jpg
Denial:

There is a difference between "it just works" and "this tool can save you work" "it just works" implies that you flip the switch and it's simply 100% done. That isn't the case. For performance reasons it only works with a subset of graphics effects - so right off the bat you need "non-rtx" lighting at some point to carry the rest of the scene. For the same reason (performance) we've seen DICE do a number of under the hood changes to way DXR works, at a basic level, to even get their subset effect to work at 60fps. On top of that the implementation in Unreal, for example, is missing several effects (no caustics), is very noisy and I believe only does one bounce. So yeah maybe at some point RTX running on a full pathtracer will do "it just works" but right now, at least for games, it's not even close. Also you say "no one is talking about non-RTX" here but the thread and the post that @fantaskarsef was referencing is literally about a game that uses non-RTX and asking whether using RTX going forward will save them work. The answer is clearly no. As for helping them accomplish things - maybe? As I wrote earlier in the thread, the lighting system in SC appears to be extremely complex to facilitate the multiple different environments the player encounters seamlessly. It's not like dropping an area light into an unreal map and saying "hooray!" - for example there is all kinds of lighting fancy pants stuff being simulated in the sky during descent onto the planet. I'm not sure RT would replicate any of that right off the bat without significantly changing the way the sky/planetary system works. Yeah it may help you simulate some interior lighting or whatever but even that is a gamble that requires a ton of tweaking and in the end it's typically pointless because @Fox2232 will come in and point out every misplaced light particle lol. I love the tech but Starcitizen might be the one of the worst games to try it in.
:D You can bet I would. But in SC raytracing has bigger problem. That's scale. And Vacuum. Because it is using bounding volume checks and you are going to have big trouble to optimize scale of each volume. 1st issue is with deciding how big are going to be primary bounding volumes. They should be bigger than planets but distances in solar systems ask for much bigger than planets (maybe even 100 times), otherwise you are going to choke GPUs even before you find that ray has opportunity to hit planet size bounding volume. //info: distance between Sun and earth is 150M km. Diameter of is 12,7k km. This allows you to place 11772 Earth sized bounding volumes just in line between Sun and Earth. Computationally impossible on any consumer HW to check all volumes if Earth's size within Solar system in reasonable time. Therefore need for much larger. Then you are going to have spaceships moving in between. Maybe asteroids... Computational wastefulness would be crazy. = = = = Now this thing I wrote above is total BS. There is no reason to check for all or any rays in large distances within Solar system as there is mostly nothing to hit/check against. You simply illuminate planets and moons as intensity of light will be practically uniform at given distance. Only thing you need is to take each planet and it's close moons and see if they do cast shadows on each other. And then you go for directional bounces. -> From my point of view RT in Space game is meaningless. There are no soft shadows in vacuum (ships, space stations). Soft shadows are form large bodies to large bodies at "?medium?" distances. -> If light bounces from one space ship to another (or space station), it will not be exactly strong enough to make meaningful difference as bounced light from small body is several magnitudes weaker than unmitigated light from Star itself. And that weak reflection is going to be much weaker due to large distances than people are generally used to in planet surface based games. -> Self illuminated RT bounces within very small distances are reasonable. But that can be done via traditional GI methods too. Expecting RT to make meaningful impact from Spaceship to Spaceship with expected distances would be like shining light with flashlight (during pitch black night) on building 50m from you and trying to read book thanks to light bouncing from given building back.
https://forums.guru3d.com/data/avatars/m/258/258664.jpg
MegaFalloutFan:

This is irrelevant you said that RTX doesn't save you work, no one was talking about non rtx here. If you use strictly RTX, it does save work. Actually it saves work anyway, here is example: You turn on RTX to see how proper shadows and lighting should look and based on this you can tweak non RTX mode to be as realistic as possible.
Like @Denial said, in this scenario the user asked the dev of a game that's already in the making, adding a second way of lighting (in addition to non-RTX) only adds more work. If a game starts development from the start, I absolutely agree with you, and in the future, I'd like to see it that way too. But as the question was specifically directed at a SQ42/SC dev here, the question as I saw it, was directed at the specific situation SQ42/SC is in. Where indeed it would mean additional work to what they've already done.
data/avatar/default/avatar16.webp
Hilbert Hagedoorn:

Over on the Star Citizen forums, a player of the space sim asked, "Do you, as I, believe that RTX tech will save you devs so much work farther along and look absolutely spanking?"... Star Citizen developer says ray tracing "a massive headache"
Not really. If you consider that they were 12 guys with 6M$ and zero pipelines end of kickstarter in Nov.2012 plus a Beta in Q2 2020, they added more features and got faster than Publishers with thousands of devs and Billions $ at day one.
data/avatar/default/avatar10.webp
ViperAnaf:

by the time this game is released it will be 3 more card generations in the future...
You mean in Q2 2020 being Beta for Squadron 42 (Solo)? We can doubt AMD or Nvidia will release us three GPU generations 🙂
data/avatar/default/avatar08.webp
TheDeeGee:

I think they got a headache trying count all the cash.
They solved problem/ Deliver what Backers are paying for adding features Publishers will never do because.. they care about profit not gamers.
https://forums.guru3d.com/data/avatars/m/199/199386.jpg
nwc:

I'm actually really enjoying the game.
What were the review scores?
https://forums.guru3d.com/data/avatars/m/191/191875.jpg
JoeBlobers:

They solved problem/ Deliver what Backers are paying for adding features Publishers will never do because.. they care about profit not gamers.
You seriously think this project is being driven by nothing but the love of what gamers want? Maybe when it all kicked off but it's safe to say that bottom lines are being driven somewhere along the line now, after all with no further stretch goals being announced logic would say that they should and have all the money they need to complete this project... so why are they still asking for more? Okay let's be more generous by all means sell the game after all that's what most early access titles do... but the vast library of DLC that costs thousands to buy? The artificially created scarcity of certain addons that have prices that you would attach to an item that is considered rare because it took an artist years of work using rare materials to create rather than because some guy stuck a 5 in the database instead of a 5,000,000? The dodgy way they dealt with folk who DIDN'T want the additions but rather wanted the game they initially signed up for? The way they changed the EULA changing the refund policy due to failure to deliver on time, removing the deadline and giving them as long as they wanted? Yeah it certainly isn't about profit, it's all about what the gamers want. 🙄