Any Other "Big" Engines Comin'
polycounter lvl 12
Hello mates.
I am about to migrate from CryEngine these days. Plan is to drop it completely before Christmas.
The only option is Unreal Engine 4. But "currently" it lacks a reliable and high quality real time Global Illumination solution. You need to take light bakes.
It has been quite some time since I last basked lighting ( 4 years
) and I don't want to go back to that.
I had heard some rumors about some ıther " Big" engines going public. Like the FOX Engine and Snowdrop Engine.
I searched the net quite a bit and couldn't find much. So wanted to ask here.
Any of you mates hear anything around about these engine would go public/ subscription/ SDK ? Any clue would be appreciated.
Thanks and Keep Up the Goodwork everyone !
Here is a cookie ( melon pan) for great answers beforehand
I am about to migrate from CryEngine these days. Plan is to drop it completely before Christmas.
The only option is Unreal Engine 4. But "currently" it lacks a reliable and high quality real time Global Illumination solution. You need to take light bakes.
It has been quite some time since I last basked lighting ( 4 years
) and I don't want to go back to that. I had heard some rumors about some ıther " Big" engines going public. Like the FOX Engine and Snowdrop Engine.
I searched the net quite a bit and couldn't find much. So wanted to ask here.
Any of you mates hear anything around about these engine would go public/ subscription/ SDK ? Any clue would be appreciated.
Thanks and Keep Up the Goodwork everyone !
Here is a cookie ( melon pan) for great answers beforehand
Replies
No one knows when it'll be coming out, though.
http://www.indiedb.com/engines
1-) I really need that real-time GI because of the game I want to make. It is a RTS and baking that even for for a quick art check gonna take... forever.
2-) I really need a good renderer capable of high quality real-time visuals because I use game engines as a renderer for portfolio purpuses.
A lot of them have some great potential, but you need to be willing to rewrite swathes of them to get the results or flexibility you'd want.
i`m going to be very blunt about it but you also need to work on your art, no fancy rendering is going to make up for assets that arnt up to par
its great that you have experience with using engines though , just dont lose focus on whats most important. your portfolio tells me you want to be a character artist then you should be creating characters that look good no matter what engine they might end up in
sorry if it sounds harsh but if you get the message it will benefit you in the long run
Unity 5's stuff (Enlighten) is still basically baked, so your mileage there would still be totally limited.
Can I suggest you put what you think needs to be in one of these "other" engines into this Torque 3d thread: http://www.polycount.com/forum/showthread.php?t=143512
There are some programmers dedicated to making it better. PBR and possibly even GI are already in da works.
Hmm... I guess you are right.
@ Ravenslayer
Constantly working on characters in fact. You are right. Need to improve everyday.
Also I get what you mean. People's skills first, software and hardware second.
ambershee:
Yeah there is LPV, I guess the only way to find out is trying it out sometime on a large terrain.
[vv]108197771[/vv]
It is what Lionhead built it for!
Hmm Good
@ ambershee:
Exactly BUT current LPV in UE4 builds is not the onee Lionhead using right now. They improved it to a great extent. e are using the first version. That is why I was hesitating in the first place in fact
Nice, will do. I've not looked too much into Torque (I had assumed that the project had stagnated after the MIT source release). It's good to see that the project's reaching out to artists and designers, to improve usability- since that's usually the one thing holding open-source engines back.
Also, TAN- Not as graphically advanced as you may like, but Jmonkey Engine's pretty good both in terms of usability and implementation if you can stomach Java long enough to get things done in it.
http://jmonkeyengine.org/
God Damn it !
https://answers.unrealengine.com/questions/35329/lpv-causing-crash-with-translucent-material-in-a-m.html
Also this:
FFFFUUUUUU.....
1. There are plenty of good looking RTS games which do not have realtime GI, why exactly do you think you need it?
2. How are your portfolio and the RTS connected? Who ever will look at your portfolio will be able to look past the fancy realtime GI, it's a nice to have, but no must have. so i would say ramp up your art before you put too much thought into a engine.
For your game, get an engine your coders are comfortable with and make great art within it's boundaries. You really do not need a super nextgen pbr based photorealistic renderer for a RTS game, which is most likely going to be seen from very far away, so make sure it looks good from the players distance.
Quoted for truth.
There is also the Bitsquid Engine which is a nice data driven engine. (No realtime GI though)
WOW WOW, calm down mate
I simply don't want to bake maps. That simple. That is it really. Nothing else crosses my mind
Of course my portfolio got NOTHING to do with the RTS game I want to. In fact I simply use a 3-point lighting with an Environment probe for renders not even anything else. The reason for wishing for that is even simpler than the first one:
I pay a subscription fee. Why not using the same thing for also portfolio stuff and such instead of paying another subscription to another rendering program ?
See ? Calm down and read the posts mate and the truth... will reveal itself
It is often an easy distraction to think one has to touch a lot of technology first, rather then trying to master a little less. Being aware of technology is important, and toying with it... but there are good reasons to "hold back" and focus on less first.
Unfortunately the "mastering" takes time, and shows less visible improvement in short time, hence it's always more appealing to "add another layer" of complexity, which means one will make quickly progress with some new tech again. That is more rewarding, "easier", "quicker", the dark side
edit: saw your new reply, and I probably took it from a different angle as well
While we're somewhere near the subject of Unreal Engine 4, is there anyway to increase the shadow map cascades on dynamic shadows past 4? I don't really have any programmer or engine rendering background experience so I'm unsure how it works other than each additional cascade seems to increase the fidelity by two, but would something like that be simple or difficult?
Currently I'm finding a mixture of lightmapped and dynamic shadows to work the best, but ideally I'm working to figure out a way to have crisp dynamic shadows on both small and medium sized objects in a large area. Shadows on small objects (we're talking coffee cup sized or smaller, as small as pencils casting shadows on desks or the tiniest being the toggle on a light switch) look nice when right next to it but as soon as you take a few steps back the drop in visual detail is pretty noticeable. I find blending between dynamic and static shadows looks alright when about 10,000 units or more away but any closer and it's also pretty noticeable.
I could of sworn I was able to set the shadow cascades in the UDK to something like 10, manually typing it to some crazy number like 20 in Unreal Engine 4 drops it down to 10 but it looks the same as 4.
I have also lowered the bias down to be pretty low which helps, although I can't go too crazy with it before shadow acne of doom starts to occur.
I think there are scenarios where having GI is really important and other stuff (like pre-baking) isn't good enough. Example, a game using a flashlight to navigate a dark building at night. You need GI to simulate the light from the flashlight bouncing all across the room. If you pre-bake this, it would look silly (since the flashlight isn't permanently turned on everywhere at the same time).
Of course, GI isn't a magic button that makes all computer graphic problems go away. But just having a lightsource cast only direct illumination can get in the way of capturing realism. Having it in real time also means less waiting on the renderfarm.
Instead of prebaking 10 different lightmaps to correct an error or get a desired result, you could save time by tweeking the light bounces right in front of you.
I am entirely calm, still i do not see how you really need dynamic GI for a good looking RTS
you can do a lot with ambient cubes to get some bounces going on, or using other ways to fake the light bounces. Of course Dynamic GI is nice to have, but why do you need it? My questions remains
What does your content look like right now?
Driveclub on PS4 is an example of the flashlight GI I was talking about.
See how the rear lights bounce from the wall and back into the car? This couldn't be done if it was pre-baked.
It's true games still feature flashlights, which I don't deny. However, they were never as realistic, if they cast light without any secondary bounces. It's also understandable why they haven't had GI up until now. Hardware has gotten a lot more powerful, we should expect it to be common place either in this generation or the next.
Unreal DOES allow you to use non baked lighting through Cascaded Shadow Maps. It's actually been available since the January 2012 UDK build and I've used it for a project or two. Works very well. Yes, you won't have real time GI but considering the genre, I personally am not sure you need that and just want it.
So why exactly are you switching from CryEngine? It's a cheaper sub fee, it seems to have the tools you want...I'm a bit lost as to what you're looking for?
I... DON'T.... WANT TO..... BAKE ..... MAPS
That's it. I Swear to Jesus Christ ! Really ! No reason else ! I am NOT lying. I swear on the name of Her Majesty and Country ! Really ! I have no ulterior motives !
Believe me ! I am not GUILTY ! You GOT THE WRONG MAAAAN !!!
You believe me now ?
Ahhh the most important question.
There really isn't much of a community going on these times and highly probable, won't go ever again.
Over the last 6-7 years I covered most of visuals, characters, level setups, visual scripting ( flowgraph) , physics. I can even write my own entities in C++ that does stuff that I want thanks to a great guy's tutorials.
But going further from that is not much possible because there is simply no sources to learn from and no great community like UE4 to ask around.
Also with latest scandals with Crytek I am not sure how the engine will fare in the near future honestly speaking.
That is why.
TL;DR, UE4 is going to be the next best thing to CryEngine when it comes to features and quality out of the box. Goodluck getting what you want with other engines these days that meet the same "quality" of UE4/Cry.
i do believe that is just the screen space glossy reflections doing its thing on the car because there is no bounced diffuse lighting present on the ground
i do not think he was questioning your dynamic motives but rather your need for bounced light at all - just a well-set ambient level would be suffice for most purposes
This, meant to mention this in my post. If it is "global illumination" it is a hack-y one at best, as juniez mentioned... The ground and surrounding objects receive no bounce light, and the overall car itself seems to not be affected other than the screen space reflections.
and on the topic... who says you need to bake maps?
most rts have no prebaked maps, due to the world being dynamic to some degree.
if you want to use enlighten, go for it, buy a license or wait for unity 5. people bere are trying to help and to understand better what you are after...
That's totally feasible in many engines, including UE4. You can skip all of lightmass in UE4 if you want to and rely on just on the dynamic shadow/lighting solutions, with a skylight to help fake bounce. That's how many RTS games do it, I imagine. Neox was questioning why you would need actual bounce in an RTS, since it's unlikely the light subtlety would even be noticeable from a traditional RTS camera.
Regarding the bounced flashlight GI stuff---that drive club example seems a little tricky to tell if it's actual bounce...or just direct lighting. Regardless, the flashlight trick is done pretty nicely in Alien:Isolation..the flashlight bouncing does a tremendous job of giving you navigable light mixed with a dim/spooky flashlight. But there was probably some baking involved with this solution--not baking lighting, but baking color influence into a 3d grid so that the light color bouncing can be 'averaged' optimally. To the best of my understanding, that's kind of how enlighten and a lot of "realtime" GI stuff work...it's realtime, but still involves "precomputing".
Ohh.. Okay then. Sorry
I don't exactly "in-need" of a bounce light system. Maybe I should have said it this way:
" I want a real-time lighting solution for very large maps which can offer both high quality and high-performance. "
So maybe I managed to clear it now ?
what is quality wise good enough, what is performance wise good enough...
what is the look & feel you want to achieve, technology is a just a way to achieve something, not a purpose
Around 16 km squares.
Includes Vegatation, Destructable buildings ( possibly alembic) , bodies of water, a skybox
Characters:
General RTS setup
Max. unit count on the screen when cameras is at highest altitude facing downwards can amount around 100, ıncluding form infantry to vehicles, mechs, aircraft.
Of course buildings constructed on battlefield.
What I "need" is " after I have done everything to increase performance during creation of assets, coding in optimize-wise etc" a dynamic lighting system that would allow around 30 FPS on an average PC.
I am bad at abstract things and I prefer visual things. Here is 4x4 map that I usually play around for fun with infantry and tanks slaughtering each other. The most I could without dropping below 30 FPS was 2 tanks 45 infantry and 3 turrets fighting against each other.. But please keep in mind these were assets optimised for a FPS- quality game. So of course they were a bit "overkill "
So what are your takes on this mates ?
At that distance bounced lighting or dynamic global illumination really isn't going to make that much of a difference. You'd be better off on focusing on making the large details seem "real" at that specific distance. IE making textures that look more like a view from an airplane or helicopter and less focusing on up close blades of grass. Being able to create textures that do not look obviously repetitive at that specific distance would be foremost, making the elevation and broad landscape interesting without looking strange, focusing more on atmospheric effects (maybe some particle clouds or volumes or whatever rather than standard fog), having an ocean that doesn't appear to tile, etc. would all take a higher priority.
No one's really going to notice whether there are light bounces and reflections on a soldier's helmet when they're the size of an ant, unless you zoom in to first person size. And imo, I think there's a reason RTS's typically stick to making the details fit for whatever level the camera is predominately set at. (Larger environments = more work = more stress on the graphics card, which is migrated by LODs but you have to ask, do you have time to make everything look first person quality in a game where you can view the whole island?) Then again, who knows if you think about procedural elements or modularity to save some of the time it takes.
I'd recommend taking a look at wargame: air land battle/red dragon for a rather nice example of beautiful textures at a distance, which also allows you to zoom in detail-wise to the jets and tanks and still look great. (I think it might have to do with blending between different textures depending on the distance away from the object, not too sure)
The skylight source also does a pretty decent job of creating ambient light with a fully dynamic light, and you can of course get away with making everything looking a bit more cohesive and visually interesting by messing with the post-process settings all day. https://docs.unrealengine.com/latest/INT/Engine/Rendering/PostProcessEffects/ColorGrading/index.html
hint: messing around with the curves of the overall values can introduce some nice effects, like darking the midtones or specific colors, creating a more contained spectrum of color, etc. Sort of like how the ole' cinematic blue lens cap seems to be applied to every recent battlefield game to make it seem a bit more "real" in a camera-man hollywood sense. Just think about how almost cold and gray the colors feel in the opening of Saving Private Ryan compared to some of the rather over-the-top vibrant colors in some environments of Inglorious Bastards for example. (Queue the cool versus warm tones)
Anyways, there's a bunch of random crap from my head. Take from it what you will
If your game doesn't need to operate on a galactic level you will probably not need zooming this intense and your requirements will be somewhat relaxed for art
That is not an issue since 4.5. I think i should update that post.
Also you can get pretty good results if you mix LPV with a movable Skylight.
usually the player car has realtime relfections (wich is very expensive) the cars of the opponents probably use a hybrid of prebaked cubemaps when they are far away and screenspace reflections when they are closer to the player.
This. the above screenshot definitely looks like SSR to me, especially with the "light" hitting the inside of the dash and AC vents which is behind cover of the "source".
I really wish I could find a link definitively saying the Fox engine will have a indie/subscription/educational release. Have u found anything?
Also...
Major face-palm to all the noise second guessing yer GI aspirations.
I remember the same scanline is more than enuff... No need for GI when it all can be faked arguments for software rendering in the early 2000's. ( ie. before there was a GI solution fer Maya )
More often than not, by Renderman users/fans with no GI in the near future.
As far as they knew at that time.
A position I remember Pixar reinforcing in the forums themselves. ( anyone remember loomis listserve > highend3d > now creative crash? )
That is until Larry released Entropy and suddenly GI was important enough fer Steve Jobs to sue Larry Gritz for patent infringement/trade secret theft
( seriously doubtful and denied despite settlement and the end of entropy. Larry sold Exluna to Nvidia and developed Gelato )
making way fer Pixars implementation.
Back then it was the same lecture based on the assumption that being excited on such a new technology meant "u wished for that single mythical make cool button".
I just don't get how anyone can second guess an artist's technology inspirations? Otherwise we would all be painters or cel animators
( not that they don't have their own respective tech )
instead of inspired to create the art of the times.( when interactive CG immersion replaces FILM as the dominant art form )
Now that GI software rendering is commonplace I find that there has never been a make cool GI button and that critiques of rendering on the rendering forums are just as tuff and complex, held up to scrutiny as much as any animation or modeling forum's WIPs!
( and I imagine the most inspired r suffering, putting in the most effort/hours/reflection and getting the most inspired results the same as with any other advanced tech's tools! )
I wonder if it were not for those excited "back then" whether or not the tech would have developed as fast when there was so much nay-saying and resistance at the time?
And from that past experience I would say that the audience has to be that inspired for development to be taken seriously. Catch 22: If u make it they will come vs They came in their pants because u made it.
Maybe the promise Geomerics Enlighten originally inspired is not ready/stable enuff yet? ( still
But from what I have seen from Frostbite and Battlefield for instance...
There seems to be nothing "inspired"? ( thanks in advance fer examples I may have overlooked )
Considering the tech seems far from a single "make cool" button:
[ame="
and is capable of varied dramatic and subtle cinematic lighting ( coming from lighting background in broadcasting, it sounds like the promised land of c-stands and bounce sheets for the future of real time! )
Don't know personally if the workflow example is actually possible or a pipe-dream they were just shooting for. But if even simple interior inspirations are possible with today's hardware... I find that extremely exciting!