HMart

Like AluminumHaste said MSAA is for edge of geometry/polygons, if you want to affect textures/shaders/transparencies, you need to use a post-process AA like TAA (it will blur the scene a little contrary to old school MSAA thou). You can use Reshade for TAA. Ops: MirceaKitsune sorry you already knew msaa was for geometry only, didn't read your reply well sorry...

Holly s*** that graphically, for a TDM mission, is very impressive indeed, very professional.

I don't deserve that credit, using parallel lights to light windows is user thebigh process.

Sorry for the off topic but nice out of the box thinking thebigh. But IMO doing that light from behind effect, would be best done through the window material using a blend add stage. Or I would use a cheap blend light instead of a real parallel light. (just put the global material keyword 'blendLight' in the light material and do a blend add) Blend lights don't support specular and normal mapping, but for such effect I don't think that matters anyway, but I could be wrong. Plus you can project a texture with blend lights as well and so you could create a blend add (black&white) mask texture that would light the glass parts and not the wood parts of the window, where on real life, light wouldn't never show though.

Personally have really no good opinion on this, mostly because I have almost no knowhow, on how TDM (or idtech 4 for that matter) low level render works. So I really don't know, if killing old parallel lights, would be good or not, but if you think they are more trouble than good... And how many daylight missions exist in TDM anyway? Honest question. But, to those that perhaps want to do day light, would making sun light, still be mostly the same using the parallelSky? One thing that IMO was cool about the parallel light, compared to other lights, is that it doesn't care where it is on the world, it only cares about its axis rotation, and it afaik was capable to "project" light from any direction, is that conserved with parallelSky? I'm asking because the sky part, in its name to me makes it sound, like it only can cast light in a semi-sphere, like a dome. Edit: Reading this on TDM wiki explained a few more things, it seems I was wrong when I thought parallel lights didn't cared where they were on the world, it seems they do for portal flowing at lest.

Indeed I saw a big increase in performance has well. This new light culling system is doing wonders, fantastic job. Btw to test this new version, I played Volta 2 for a while and in that part where you need to swim against the current, to get too the gold bottle, I was unable to do so, no matter how much i tried, but I remember being able to do that very easily, does the swimming system got changed or is a bug?

Agree on FXAA indeed but it would be cool if you looked at SMAA (Enhanced subpixel morphological antialiasing) as well, is a upgraded version of MLAA (rival of FXAA) from AMD, made by Crytek and Universidad de Zaragoza. Is always subjective but many say it looks better than both FXAA and MLAA while being almost as fast. Never tried it in TDM but I know that Reshade has a SMAA shader, could be a way for you to test how it looks in comparation to FXAA ingame and later decided if is worth to write a inhouse one. And IMO no need to replace multisampling it can work along side post process AA systems very well.

$speaker_sausages.setKey("s_volume", 2); That is a spawn time script function meaning what it does, only happens when a entity spawns, so would need to restart the mission for it to make effect, if you want to change something during gameplay, you need to call other functions like /** * Fades the sound on this entity to a new level over a period of time. Use SND_CHANNEL_ANY for all currently playing sounds. */ scriptEvent void fadeSound(float channel, float newLevel, float fadeTime); or /** * Set the volume of the sound to play, must be issued before startSoundShader. */ scriptEvent void setSoundVolume(float newLevel);

I'm not anti frame rates if that was the idea i passed ;D nor I think playing at 30fps is equal to 60fps or higher, I'm just saying that in a uneven frame timing scenario, playing at constant 30fps can feel smoother (less stuttery) than at very uneven 60fps or higher and that is true, but in a equal situation, I will prefer 60fps (or more) above 30fps any day! (real frame rates not AI fake ones...)

Not only that but IMO today with Nvidia AI frame generation (or any equivalent tech that AMD may eventually come up with), frame rate numbers lost all their meaning (at lest with frame generation turned on), with DLSS3 FG you can have a "60" fps game that can feel like a or less than a 30 fps game ( on input lag and if that is the base frame rate from where DLSS AI is working from). High frame rate numbers are a good way to sell GPU's ...

I'm sure they will create a stable 2.12 version, when its done! But the official game release is at version 2.11! The 2.12 is still in heavy development and so, will have bugs and most probably break very easily, it shouldn't be used for real play, only to help TDM code team, to test, find bugs and regressions. This dev version is accessible to the general public, only by courtesy of the TDM team, so we can experience and see how the game engine is developed, bugs and all, don't make them regret it, if more people start complaining, instead of helping, they will more than likely closed it behind doors, like they did before and do the public testing, only way later when is more stable. But that also makes it longer to release, because they will have less people testing, will have to do most of the testing themselves and because, afaik 1 or 2 people, are working on the engine at the moment, it will take way more time to develop and then test to the final release. And it will very probably come with more bugs, than if they did early public testing like they do now. Unlike paid comercial developers, TDM team doesn't have a dedicated paid bug testing team, we the players are the tester, is the lest we can do for be giving this fantastic free game. And so IMO complaining the development branch is not stable, doesn't help, they know that, if you want a more stable version and not experience bugs, then use only the oficial 2.11 release, wait for the final oficial 2.12 release, again, when its done.

I was thinking about this the other day, after seeing that Source engine add a func_occluder entity, while idTech4 has not, Valve at lest thought they were still useful for something, I bet they are used to help optimize the HL 2 outside areas, that game has plenty of those. That lead me to search about the subject and saw this two links, imo they have some useful info. https://www.gamedeveloper.com/programming/rendering-the-great-outdoors-fast-occlusion-culling-for-outdoor-environments https://bazhenovc.github.io/blog/post/gpu-driven-occlusion-culling-slides-lif/

Then is something else thanks for the info.

Just for curiosity, If any of you are running with unlocked frame rate, lock the game to 60fps (or 60hz) and see if the problem goes away?

I comprehend but you do lose volumetric effects with stencil but only the really recent missions have volumetric lights. Btw In reality there's a big difference between what shadow maps are capable off and stencil maps are capable off, unfortunately because of backwards compatibility with old missions and the fact that for a long time, only stencil was available, TDM team add to limit shadow mapping capabilities mostly to what stencil can do.

MSAA is heavy specially at 16x because is more or less a type of super sampling (rendering the scene at higher rez than displayed), but unlike the later is way faster, but like SSAA, MSAA still increases pixel count. Yes i'm not surprised AA was the main culprit good it was not something else. Btw Stencil shadows are mostly CPU based, shadow maps are GPU based and because you were already GPU bound, changing to to maps wouldn't make any diference, perhaps even make it worse, because maps are normally slower than stencil, thou today because of some exclusive effects to maps, I would recommend them to everyone.

Then the problem is also the material, yes. I would assign a opaque dark material, to the inside faces of the light housing, minus the glass front, so they don't blend with the background.

Was this model made in DR? The uvmap is all over the place, meaning it falls outside the 0 1 area. But that is not the problem, the problem is that It add a missing face in the back of the lamp thing, so you could see to the outside thought it. I put a face there that should solve the problem. mod_playerlamp_model.lwo

Link the model here and I will solve that, is easy (famous last words.... ). To me no need to link the material or textures just the raw model. Btw is the model lwo? I can only open .lwo models in Modo my tool of choice.

That is slightly less than 4k but imo is still a really a high rez, specially for this engine, that is ~7.3 million pixels that the engine has to calculate/shade, per pixel 60 times or more, every second! also why MSAA (multisampling anti-alising) can be very expensive on this engine, it increases pixel count. So like nbohr1more said, if you have it on, lower it or better disable MSAA, at that rez IMO you don't need it anyway and it has no effect on geometry faces/surfaces, only on geometry edges and if you really want AA, also like nbohr1more said, force a post process one (FXAA or better if possible TAA) using the Nvidia driver. You can also try to lower the rendering scale and increase the sharpening filter to counter balance the loss of definition in the video options menu. About the CPU the fact you have low CPU utilization and high GPU utilization, shows you are getting GPU bound in this case, so increasing thread count in the game console, I don't think will help but try it. Btw sorry if you already know this but if you don't, at lest may explain to you why this "old" game, is so heavy even for your modern very expensive GPU, this game uses a modified Doom 3 engine, with better support for multicore CPU's and other modern stuff making it way faster than original Doom 3 engine, but is still using the old forward render and unlike more modern renders like deferred render or forward+, that can render hundreds of (non shadow casting) dynamic lights, before getting slow, a old forward render like of Doom3/TDM, can get really slow with a relatively low number of dynamic lights in a scene, worse if they all cast shadows, and on this game all lights are fully dynamic and in many missions most of them cast shadows, real time shadows, there's no light baking of any kind.

If dmap doesn't show a leak and there's a leak into the void (a gap in the world into the void or a model origin in the void) that is really strange, too see if is a leak and dmap somehow fails to show it, try to start your map with "devmap mapname.map" instead of using "map mapname.map" devmap afaik ignores leaks, if the map/mission opens then is a leak. But the fact that you are having AAS out of date warnings after dmap, sounds like you are opening a old version of the map not the one you dmap a new, perhaps you have DR badly setup and it is saving the fresh maps and files into a different folder than you think?

Personally I really don't know if that is available or not but that does sound too a specific guard behavior, so perhaps is not defined? If is possible to have that? Yes, if is not available already, unfortunately there's no other way but code it into the guard AI, in c++ or script yourself, if you know how to code.... Btw the civilian AI does have "fear" and runs away from danger, is that AI type really entirely separate from the guard AI, or the guard behavior is just "dormant" on the civilian AI and vice versa? If all AI's are done through the same AI system then perhaps is just a matter of turning some stuff on or off. If they are totally separate AI's, they are still so similar in parts that perhaps so you could adapt/transfer the civilian fear behavior into the guard AI for specific types of danger. But all of this requires coding. Sorry for not being more helpful than this, I really don't know much about TDM or even Doom 3 AI stuff.

Yeh that is also how Doom 3 weapons were done, but perhaps this can be worked around? Next to weaponDepthHack there's also a modelDepthhack bool I assume for every other model, perhaps that just needs to be put in the model def file or enabled by a spawn argument? I really don't know, or if that spawnarg even exists, if not then is something to put in the roadmap.

Is the lamp a weapon? Unless TDM weapon system is different from Doom 3, weapon models should have the "depth hack" thing enabled, that "hack" (that is how idSoftware calls it ) makes sure the weapon never intersects with geometry. in the code: bool weaponDepthHack; // squash depth range so view weapons don't poke into walls Don't know if there's a way to enable that without messing with c++, stgatilov or others may know.

Perhaps but the game, has gameplay features that can muffle your footsteps sounds, carpet and other ground textures, the vine arrows also, so by making them more silent/quiet, imo you may end removing the incentive, for mission makers and players to make use of those. To me, If a player thinks the footsteps are to loud then maybe the mission maker, didn't put enough ways for the player to walk more silently in the level and so he always hears loud footsteps and thinks that is bad/broken. Btw what about incentivizing mission makers to do certain things, by the older more experienced mission makers, agreeing on a small list of "best practices for mission makers" for performance, looks and gameplay and pin that at the top of the forum? Then any new developer may learn something from that and avoid some rookie mistakes, like for example, not putting enough ways in a level for a player to walk more quietly. Just my two cents.