Today I did a few experiments with gameguru. I was noticing that no matter how bright I made my lights, the only thing that the specular map would reflect was the sun. I use static lights, so I've always tried to use them only to make for nicer shadows, but I could never seem to get the nice shader effects that I can in my favourite 3d tool:


bumpy 'n glossy

I wondered why this was and started to develop a hypothesis that I tested today and proved myself correct. Specular maps in gameguru actually DO NOT REACT TO STATIC LIGHTS.

Here is the test to prove my findings. First: our room setup:


2 of these rooms are lit. to begin, we will only use static lights. one per room. 500 range.

looks pretty nice, normals are bumpy but only white on the parts emitting specular. This is the issue. (also no semblance of "light bounce")

so to test my theory we put 2 dynamic lights in the same place as the statics with the same colour. the statics have a range of 500, the dynamics 100. This was the result:


youll notice the tiniest inkling of specular emission on the ceiling. This is what we're going for. I decided to see if increasing the size of the dynamic lights would effect how "specular" everything was. I changed the size of the dynamic lights to 500, and kept the statics 500. this was the result:


above: architectural visualization???

so my conclusion proves my hypothesis. Gameguru specularmap shader does not consider static lights. This is a bug, and I think it should be looked at, considering we're already diving headfirst into PBR.

if this wasn't fixed, though, it wouldn't be the end of the world. crafty gameguru users should notice this gives us increased control of the specularity of our surfaces. the ability to choose the size of the dummy-dynamic light that we use to cause specular reflection, along with the per-entity specular control of the object itself built-in opens up a whole new world of possibilities.

Another thing I think everyone should note is how the use of dynamic and static lights creates a sort of "global illumination" effect with the dynamics creating specular reflections AND causing all walls to be somewhat lit, but the static lights creating shadows.

I can't wait to see what the forum produces with this knowledge!

interesting find Bugsy illumination if you pardon the pun is eluding me at the moment

good one graphix!
I cant wait to see what sorts of things yall make. I hope someone posts something here!

I don't think they ever did, i'm pretty sure the current shader for specular doesn't reflect anything static except sun light. I do't think it's a bug, more that it's just not designed to do what you want.

Also, with the addition of the current iteration of PBR it looks like Lee is moving away from specular, and onto amore metallic kid of workflow.

this tutorial proves that is incorrect
I consider this a bug, why should static lights not affect specular? That doesnt make sense.



that also doesnt make sense considering specular mapping is a part of the PBR workflow. PBR maps still require normal/specular/diffuse (or as they called now "albedo" maps). they just also require roughness, glossiness, displacement, emission, opacity and optionally several other maps.

=============
anyways
=============

I've attached some examples to show how this is best used.
In a high lighting/low specularity situation, or a scene lit mostly by the sun, there is almost no need to use this, it does almost nothing except make the room artificially brighter:


figure 1: static lights only. good.


figure 2: dynamic lights added: not better


This works best in low-light situations and interiors. It does an excellent job drawing the players eye to a light, or making a clean area look less empty. 0 ambience is where it is most noticable:


figure 3: no dynamics used: no specular emission


figure 4: identical dynamics used with statics: scene looks much nicer, light draws you down the hallway, floor looks super clean

it also helps bring some light into shadows and create the effect of light "bouncing" into nearby crevices and rooms, which would be explained by the reflective materials that illustrate this technique the best.


figure 5: no dynamics, middle of room almost black, floor and ceiling textures almost indescernible


figure 6: identical dynamics used. floors glisten and every tile can be seen even in low light. whole room has light appearing to bounce throughout.

The effect of this "GI" effect is best seen in the shadows of things close to lightsources. with 0 ambience, they unrealistically are near pitch black using no dynamics, but with the use of a dynamic and a static, the bloom is much more reactive and the shadows realistically appear to reflect the colour of the light:


figure 7 ugly


figure 8 nice!

This effect is not infallible. it doesn't work well all the time, and the cracks begin to show when creating more detailed realized levels with it: GG can only handle 3 dynamic lights onscreen at once. any more, and lights start going off. Also, specular reflections on floors can be seen past walls:


figure 9- small visual bug

According to most of the various tutorials I've seen pbr can use specular yes, but not in the same workflow, you either use a metallic roughness workflow or a specular gloss workflow , the new pbr system currently uses a metallic roughness (although lee for some reason has named the roughness gloss even though its an inverted gloss map which is really a roughness map) workflow.

As to whether current shaders are meant to reflect static light sources other than the sun, I'm only going by what I was told when I asked about why only the sun effected specular a while ago. ( I don't think I wrote it right previously, I meant static light sources, dynamic do reflect on specular, just not very well).

ah well, hopefully this tutorial clears it up! the sun is not the only thing to affect specular. only dynamic lights affect specular.

i wonder why lee wouldnt go with both
I have also noticed that roughness or "gloss" maps can be made very easily from a good spec map.

Sorry just edited my post above, I wrote it quick, when I said its not designed to reflect any light other than sun, I meant "any static". From what I have seen so fall it looks like Lee is trying to add both workflows so we can choose as he had added specular to pbr, but the metallic (he's called it metalness) is the one that's working best so far.

Roughness gloss and specular maps are very similar maps and can be made from eachother, yes, but roughness and gloss are the exact opposite to each other, where one is white for shiny the other is white for not shiny etc. Which is why they don't work well in the same workflow.

That's some nice investigative work and lovely illustrative screen shots Bugsy - thanks.

I think the release of PBR shaders is going to cause a lot of debate and comment until we get used to the workflow . Whether it fixes the static light issue or not I don't know but dynamics are much improved.

Cheers.

Just to clear:


Specular workflow has nothing to do with specular reflections in the scene.
Specular workflow just use the specular texture to generate a metallic texture ( nothing else ) , the PBR shaders are the same and do the same.

PBR do not use a specular textures , the specular reflection is calculated using the metallic/roughness textures ( simply from the material you defined in your textures ) , so specular cant be controlled directly in PBR , only by the way you make your PBR materials.


Bugsy: I think the bug is that the lightmapper do not generate specular reflections, the specular light should already have been in the lightmap textures

So why does substance painter have spec/gloss workflows, and why does the unity and unreal documentation discuss spec/gloss pbr workflows, and why does SP's pbr tutorial state that thee are two main pbr workflows, being spec/gloss and metallic/rough if spec doesn't work in pbr then? Now I'm totally confused because almost every professional tutorial from the big companies like these i've watched or read is saying the opposite of what you are stating.

Dynamic lights go through walls, this has always been the case in GG/FPSC:R.
Use a smaller dynamic light to control it better; specifically in the 100-200 range.

One of the issues with generating specular lighting from static lights is that the effect is view dependent and will change based on the viewing position and angle of the player. Static lights are used to bake lighting into the scene and so will not change as the player moves around the scene. The solution is, of course, a GI approach to lighting, so all lights (including static ones) are considered in real-time for correct lighting. Alas, this affects performance quite a bit with a forward renderer, which is why the approach has not been used.

Right now we're working on the PBR renderer for the next update, which might provide the means in time to handle a specular effect from static lights, but the moment these extra lights are part of the real-time lighting model, the pre-bake may become somewhat redundant. A deferred renderer does a better job of collecting lighting information from many light sources, and I would be tempted to roll 'infinite lights' into the deferred rendering work if it gets to the top voted spot.

I for one would definitely like at least an increase in the dynamic lights we can use at once, it would solve a lot of issues, especially (need to double check as I haven't checked in the last couple of betas) considering how some models textures look completely different in dynamic than they do under static rendering, I'd live to be a me to just use realtime without needing to baje, it was something I mailed you about Lee, but I can't remember whether you found the issue or not.

Belidos:

Yes there is spec/gloss , but when you talk specular workflow , the specular texture is only used to generate metalness for the PBR shader.

So when you have a spec/gloss workflow , the engine will use the specular texture to generate metalness, the specular texture will not have any direct impact on the specular reflection in the shader , that is always calculated from the material using metalness/roughness no matter if your using a specular or metallic workflow.

Lee: 'infinite lights' now your talking

Hi Bugsy ,
are you working with the public release and finding all theses things
or in the private beta testing update ?
If your in the public release it's more than possible that anything you find that works
in that release might not function at all once the PBR release is out.
if your using the private beta release wouldnt it have been best to post this
in that forum instead so that those in the public forums dont get too confused ?

Just asking...

thx

UNIRD12B

Heck I'd settle for more than 3 at any given time.
Infinite lights would definitely be a step in the right direction but then you run into the issues that dynamic lights present, namely their ability to ignore walls and barriers.

Still I think in terms of a cheat, Bugsy is getting pretty good results from his method (which mirrors what I presented in this article:

http://gamegurureport.blogspot.com/2016/07/technique-evaluation-advanced-lighting.html

Basically using static lights to set the 'tone' of an area then using dynamic lights to provide the shadowing on dynamic objects, specular, etc. Still though I'd recommend using a smaller dynamic light of 100-200 (generally below 150 but greater than 100 since nothing below 100 works).

I suppose the real question though is what is the long term answer for higher quality lighting that won't crush performance in game-guru?

its the Public release not BETA

hi Graphix ,
thx for adding that , so then , all this discussion could be made moot pretty quick
in the new release when it comes.

Thx

UNIRD12B


Thx

UNIRD12B

Can't wait for the DX11/PBR update to come! It's gonna be awesome!

So it doesn't use specular for specular reflection but instead for metallic, gotcha, thanks for clearing that up. PBR's confusing because so many people have different ideas how it works.

Ime sure this is clearing thing up for a lot of people,my head is spinning.we need one export setting ie unity or whatever and then import into gg .

Dave

wow so glad theres such a discussion here!


this is what i was trying to say i guess. how can/can this be fixed?


how/can this be fixed? seems like if changing to PBR-mode is required to make specular reflections, which won't even be handled as specular reflections any more from what i am gathering will make our games slower and smaller across the board for a simple effect we should have had day 1 (or perhaps documentation of phenomena like this for advanced users included)


in my experiment I did that, however I notice that the way to get a level of specular reflection most consistent with your light power is to use the same range as your light. this is a very limited technique, however as we can only have 3 dynamics at once. it works best in a situation with very few lights per room/building and not much view into other rooms. an FPSC game setting basically


I wasn't aware FPSC had this approach, but it seemed to handle specular reflection somewhat, (although it looked like the light was always casting from the player). Is there perhaps a smarter or even similar way to "Fake it"? maybe some sort of global specular cubemap that is just faked and muted slightly in shadows? I ask because I notice that when you use dynamic lights, the "sun" doesnt really effect an objects specularity indoors anymore, it only seems to when you have an entire level with no dynamic lights, and you happen to get a weird glisten. because of the way youre facing


"might" ? I dont understand. can you provide examples?


not if dynamic lights dont even cast shadows. if you included this global illumination it would take prebakes to a whole new level. as it is now, it seems prebakes and dynamics are both required, cutting down the size and scope of a level with high graphic fidelity.


the public release. I guess nobody around here has deemed me enough of a hardened gameguru user to send me the beta


dang here i was thinking i was so cool and original


so what you're basically saying is something with a low roughness and high metalness is what we will now use instead of something specular. That is cool, but i worry reflections will look odd, especially when combined with stuff not using PBR. I also fear that with our already extremely limited capacity for media in our levels and games, and even stuff I'm hearing about games not building due to overloading gameguru, or not working on all computers will mean that this PBR business will be almost useless for actual "games" and more for small one room walking simulator projects, given how many more maps everything needs. if it's possible to use only 1 pbr object and go with regular norm/spec workflow on other stuff, that would be pretty useful, but as said earlier, combined with actual pbr effects, our underperforming specular effects might drag down overall visual quality

It sounds like we want to further develop the light mapper to generate some pre-baked data we can use as part of the dynamic lighting system for things like dynamic lights that do not go through walls and handle specular effects properly. It's also worth pointing out that the light mapper does not produce PBR texture sets and uses non-PBR assets. The engine will attempt to render a non-PBR asset under the PBR shader (including lightmapped geometry) when the PBROverride is active (on by default). As stated above, specular in the PBE universe a combination of contributions that involve the environment map plus some clever math, so it might be best to discuss the issue of specular in the context of what we have now under the PBR renderer.

but what about including something in the static lighting system to produce specular reflection? when PBR is introduced, will updates and support continue for old fashioned norm/spec workflow?



define "attempt"? have people tested this?

Is there some way I can test the PBR renderer or see pictures of tests with it? has anyone tried just a simple untextured room with a mirror and a few objects of different material with perhaps a few different lights?

Hi all

For me the whole problem with lighting comes from the need to light map the level to get shadows indoors, but the light mapping process ALWAYS produces unwanted artefacts somewhere.

Most likely it is me missing something important in the model making\texturing process, but whenever I light map a level there is some part of the map that is ruined by it.

Models may be washed out (especially when trying for whiter lookin walls in sci-fi stuff) , or you can get ridiculously bright areas - like those produced by the skylight in the original post images - or weird shadows somewhere on the terrain outside.

Lol - when I press the F keys by accident and invoke the light map process, I always kill GG with task manager, because the end result will not be good.

By the way Lee,
as a suggestion, the F1 keys should invoke a dialog box so you can choose the level of light mapping you want, and them Confirm or Cancel you decision, obviously my choice would always be Cancel


Reliquia...

hey Bugsy ,
it's really NOT that hard to get in to the BETA TESTERS GROUP....
in point of fact.....LEE even allows me to participate , so , that shows you how low the bar is set.
I think it would be a great idea for you to just send him an email and ask to participate as this would give you
a full blown opportunity to see to what degree your findings will apply to the more recent updates that have defined
more clearly how PBR textures will act and look in the next Public release and to possibly suggest things to Lee
before they are all written in stone.

Keep up the great detective work......we ALL need help in grasping the PBR texture concept as it applies to
Game Guru.

Once you get in to the Beta , you might want to look at the Pbr textures in the Entitybank / CityScape PBR folder
and see how they have been applied and how they are structured as it might give more insight to the overall direction that
Preben and Lee are taking with them.
For example , if i'm understanding it right , when I opened one of the textures in GIMP , it showed that it had 12 layers
and no way I could even start to understand all that at this point.

Soo ALL HELP is needed from everyone to see if we can all get our heads around this PBR things and have fun...!!!

Thx

UNIRD12B

Yes I've been testing it quite a bit and it still eludes me

welcome to the club. make sure you use f3 for lighting, use walls that are more than 25 units thick so the sun won't shine through, and all media that you use in a lightmapped scene MUST have specular and normal textures. even then, the lightmapper isn't super amazing or anything and will leave you ticked off frequently without you spending a WHOLE LOT of time learning the ins and outs of it and how to best apply its "eccentricities"


you're telling me all it takes is an email? is the BETA run through steam or is it standalone?


layers or channels?

Yes. it's as easy as that, send Lee an email asking to joint he beta, he will send you a code, then in stem go to the gameguru properties, choose the beta tab, enter your code, then you will be able to switch to the beta in the drop down.

and then i can select which version to run in properties?
is it possible to have 2 listings? one that always will open the newest beta and one that will open up the public release?

unfortunately not no, I setup a test PC in the office and I run the Beta on that I have left my desktop with the public preview so if you have a spare PC its an ideal setup if not then you will have to choose one or the other

Yes, copy your GameGuru Folder and rename it to GameGuru - Old or whatever you want, then update to beta on the steam, whenever you want to run beta run it through steam, whenever you want to run the live version run it from the gameguru.exe I your gameguru-old folder.

I wasn't aware that you could run gameguru without steam! this is good.

I run it most of the time without steam,not a great lover of steam .

Dave

Of course you wont get any updates this way.. So once the dx11 is released you may have to reinstall..

wow seems awful hard to access the beta and public release at the same time...

im not understanding why a beta archive couldn't just be available at a link, and the beta was what you used off-steam...

Because its so much easier to auto update via steam and its protected against any Tom Dick or Harry getting hold of the link.
You can easily opt out and the public is reinstalled ... No biggie if you back up your " Files " folder before going beta ..

yeah gonna have to make a day of it i guess