zergrush
-
Posts
296 -
Joined
-
Last visited
-
Days Won
2
Posts posted by zergrush
-
-
Thief used torches so we use torches. People are familiar and comfortable with them whether they make sense or not.
I'm not entirely against the use of torches, like I explained, but this is just about the worst argument to use them. If everything TDM can do is to copy Thief for the sake of familiarity, one might as well just play Thief instead. It's a slippery slope that can easily justify turning TDM into nothing else but a Thief clone.
This not to mention that there are already plenty of fantastic maps that do not use indoor torches whatsoever.
-
Open, extinguishable gas flame lights seem to be the closest approximation to a torch but they certainly provide a more upscale feeling wherever they are placed, and are not appropriate sources of lights in dungeons, stables, back alleys or what have you. I would love to find an alternative that makes sense, while also retaining it's in-game function.
We can either do open gas lamps with a larger flame or create a model for a wall mounted oil lamp with a wider radius. There were plenty of ancient lamps with more than one wick for this purpose. Here's an example:
-
1
-
-
I would think anyone worried about the historical accuracy of TDM would find dozens of things more egregious than the overuse of torches.
This has little to do with the aspects that are not based in reality (such as steam machines and robots) and more with the ones that are (such as how fire and smoke behaves in real life). Naturally there always aspects of gameplay that are adapted or streamlined to make things fun, such as hiding in minimal shadow rendering you near invisible. Torches on the other hand, serve no such purpose with all the alternatives already available. Being aware of this is something that can even help mappers add some nice details to flesh out missions. For example, if you still want to use indoor torches, placing them in a surface that is not flammable and with an air vent right above.
-
1
-
-
That's true, they scale with modern CPUs very well though (Intel Coffee Lake setup here). While shadowmaps can give a fairly midrange card (GTX 1060 6 GB) an intense workout.
IMO the problem is TDM doesn't look like a modern game in terms of visual fidelity, so players who aren't familiar with very specific tech we have here might feel disappointed with its very high hardware requirements.
I guess that is because most of the things that make TDM impressive are either more subtle (such as shadow fidelity) or tend to run under the hood (like AI behaviour). Making performance smoother would certainly improve appeal for those looking for a cool, easy to run free game, and eventually I guess improved textures will bring graphics to a new standard while optimizing performance (I've been very impressed with your new builder texture pack, for one).
-
1
-
-
The torch. To most players this simple light-emitting object is the staple inanimate foe in both TDM and the original Thief games. It represents what any half-decent sneak artist must avoid in order to stay alive, an icon that has been seared into the minds of stealth game players ever since they took their first steps into the Metal Age.
Outside of Thief, torches are also frequently depicted as the most common light source in both pre-modern fantasy and historical fiction settings alike, and can be found as permanent fixtures on the walls of every fictional palace, castle or peasant hut, from the Lord of the Rings film series to the Gladiator.
But what does history actually say about the usage of torches before the invention of gas and electrical lights, and how sensible and commonplace was their usage? I've debated this matter on the thread of a recently released FM, and I thought it would be worthy to further put this into deeper discussion.
Not as bright as you think
What happens is that contrary to popular perception torches are an impractical and unreliable source of light, and their usage indoors is a particularly poor idea. While still being used around the world today in ritualistic ceremonies, such as processions, torches primarily served either as offensive weapons to set buildings ablaze, or as beacons to temporarily signal a location or send a message from afar.
The reasons for why torches were simply not used as a portable or wall-mounted light source become pretty self-evident if one has ever been close to an indoors wood fed fire source without any direct ventilation: torches do not burn cleanly without making lots of smoke, their light is unstable and obfuscating at close to medium range, and their primarily fuel source lasts for half an hour at best. In fact most movie prop torches are generally cleverly disguised gas torches, burning without any smoke. Moreover, placing a torch near a wooden structure, or any other generally flammable material, such as wallpaper, would present a serious fire hazard at its worse, and a source of soot, stink, and burned spots at its best. As personal light sources, torches are far from being handy objects either, starting by the fact their sheer size and amount of heat produced make for a fairly uncomfortable object to carry around, added to the previous general disadvantages of smoke and stink.
History researcher and youtuber Lindybeige further describes the problem with torches in the following video series.
So what was used instead?
In ancient times the light sources were actually plenty, very accessible and quite easy to carry. People would build their houses to maximize the use of natural light by default, but for night time there was a wide variety of decently lasting clean light sources available, both expensive and very very cheap. Commoners would mostly use pieces of rush weed embedded in fat, which could burn cleanly sometimes for more than hour. Richer households would be able to afford tallow candles or even beeswax candles, which would become cheaper and eventually ubiquitous in the centuries past the middle ages. If you're curious about details I would recommend you reading this article on medieval light sources.
So what does this mean for TDM?
As of the moment there is an unholy amount of torches on plenty of TDM fan missions as part of this misconception, and my goal with this thread would be precisely to change that. While TDM is a fantasy franchise it is important to notice people are incidentally educated of many historical notions through fantasy representations, and as such I feel it is important to keep at least some aspects of daily live grounded in actual reality for this purpose. This does not mean you should rush to delete all the torches from your current mission, but maybe be more wary of their placement in the future missions you create.
And what are the alternatives?
Fortunately TDM already has plenty of alternatives to torches, from oil lamps to lanterns, which produce a similar effect. To have a model consistent with the larger light radius of a torch, perhaps an oil lamp with several flame ends could be added, as well as diversifying the amount of candelabra both for oil and candles. Some mappers already well-aware of this notion. A good example of proper use of candles and lamps replacing torches can be seen on the Wm. Steele series.
In what situations would it still make sense for me to use a torch?
In every situation where it would either make sense to have a beacon meant to be seen far away (such as tower) or as an impromptu flame source on outdoors locations such as encampments. Although not truly historical, this usage can at least be justified by common sense through some extent.
One good example of allowed indoor torches would be magical flames created by paranormal events or entities. These flames generally take a unusual color (such as green or blue) to signal their unnatural source.
What about dungeons?
On another example of a classic misconception, dungeons were never lit in ancient times, precisely for safety reasons. If a prisoner could not see where he was going, there would be little chance he could find his way out in the event of escape. For this purpose, only guards and visitors would carry light sources. In fact the absence of light in dungeons was such that it would not be uncommon for prisoners to go completely blind overtime.
-
3
-
-
Stencil shadows on the other hand can be a bit pressing on the CPU, and in my older AMD laptop it causes it to overheat fairly quickly. Performance, however is clearly superior, like you mentioned. Maybe the team can start spending a bit more time on maximizing performance once all current experimental features are rolled out. So let's hope for the future.
-
In terms of performance, ideally every mapper still needs to do the tuning step and manually disable shadows for select materials/lights/entities
With or without foliage shadowing I, as a player, expect to see 60 fps on a modern $100 GPU in 1080p with eye candies off.
We might add a cvar to control foliage shadowing in general but it probably won't qualify to end up in options GUI
I personally think the game should run just as smooth on legacy machines, at least on less complex missions. The thing about games available gratis is that they often find an audience on people with less money and weaker computers. It's the main reason why free MMOs like Tibia are still so popular on poorer parts of the globe. I feel accessibility should always be kept in mind when developing any libre game, and to its own credit TDM already does a fairly good job at that at least when running on stencil shadows. While a big part of optimizing performance will always be left to mission designers, there is still a lot that can be done on the engine and artwork side to make things run smoother.
-
I don't think it is wise to add any additional graphic upgrades on top of experimental features that would further stress the engine without getting performance ironed out first.
-
As someone who lives in a house with a traditional wooden board floor, I can say wood can be very loud. Not so much the step noises, but the wood itself creaking under pressure.
-
1
-
-
I committed changes to the trunk for the adjusted visibility factors.
See the last note in this issue.
Thanks for addressing this so fast, Grayman.
Since I run GNU/Linux myself, I would like to ask anyone who has access to the windows builds to record a video of the changes on all the difficulties, so we can have a good look at it.
-
Hi Dragofer, I found a bug in one of the first rooms in the boat at the beginning of the first episode. Apparently the AI can see you through the wall due to a collision bug. Please see the video below and Grayman's comment on the matter:
@zergrush:
The problem in your videos has nothing to do with the visibility problem we've been discussing.
The func_statics shown in this image:
have no clip models, so when the AI is at the far end of the room, looking back in the player's direction, a test for visual occlusion fails, since there's no clip model to occlude the AI's LOS to the player's eyes (pink cube).
Result: he can see the player hiding up on the package.
So this is a map issue.
To check, I removed the no clip model designations and rebuilt the map. The AI comes all the way back to my side of the room and doesn't spot me until he has cleared the corner of the shelf I'm hiding behind.
Mischief managed.
P.S: A secondary request Any chance you can update the loader background images with higher resolution versions? As of the moment they're mighty pixelated.
-
@zergrush:
The problem in your videos has nothing to do with the visibility problem we've been discussing.
The func_statics shown in this image:
have no clip models, so when the AI is at the far end of the room, looking back in the player's direction, a test for visual occlusion fails, since there's no clip model to occlude the AI's LOS to the player's eyes (pink cube).
Result: he can see the player hiding up on the package.
So this is a map issue.
To check, I removed the no clip model designations and rebuilt the map. The AI comes all the way back to my side of the room and doesn't spot me until he has cleared the corner of the shelf I'm hiding behind.
Mischief managed.
Thank you for clarifying this! I will report it to Dragofer in the respective FM thread.
-
I don't think replacing all current models makes any sense, but at least newer ones can be optimized better. That will require different approach though, basically not using current texture material base (consolidating materials, making new textures etc.), so it will add to the mod package a bit. It's a difficult situation to get out of, no doubt about it.
This doesn't paint a good picture for older missions though. Even if one were to replace older textures and materials with more efficient versions, it means pretty much every FM would have to be revised.
-
@Grayman, I've only lightly played the mission so far, and I gotta say: what a magnificently built level! The atmosphere, structure, and design all seem top notch, and I can't wait to have a bit more extra time to tackle this one.
So far, my only fair bit of criticism goes towards the intro and the loading screen, something that also affects previous WS missions: I'm a fairly fast reader and having to wait for the paragraph texts to slowly appear can be frustrating to say the least. I also think the display image for the intro should start zooming out at a decreased size, so it doesn't appear so pixelated at the beginning. Personally I think WS begs for a more traditional Thief-like voiced over slideshow, like on the two default TDM missions, but I understand it might not be your cup of tea. Finally, because I absolutely love the style of the loading screens for WS, I wonder if you could render them at a higher resolution, so they look prettier and less pixelated as well.
-
-
Recorded with inaudible on, at a sharper angle closer to the wall and leaning left, just to match shoulder width with FOV. If I'm not leaning the result is the same.
-
A spiritual remake of Thief 1's "The Sword", with a TDM flavor to it.
It would be called "A Simple Mind". The player would be hired to break into the home of a reclusive member of the inventor's guild and access his personal lab in the basement to steal an artifact. The house would be a plain, inconspicuous country manor with only some guards and servants going around the inside, and the player would be in the possession of a very detailed map of the whole house... except of the basement.
Once access to the basement is gained, the player would first find a labyrinth of tunnels followed by a system of catacombs filled with clever traps and highly sophisticated robotic guards.
Upon reaching the deeper level, the player would finally enter the lab where he would confront the inventor himself, a decrepit old man connected to an iron lung sort of device. The old man would reveal the player he was one of many lured to break into the mansion as a test, and that he was the first one to succeed. As such he was free to claim all the valuables and knowledge in the lab, and as a last request, the old man would ask him to unplug his respirator device. The player would then return to the surface via a hidden elevator and the mission would be over.
-
1
-
-
@Grayman, I found an interesting one while playing Down by the Riverside. I'm playing this on Forgiving while using the adjusted AI reactions configuration you posted here. At an early section of the level one AI is apparently able to see you through a wall corner. Now this guard is a notoriously hyper-aware one, as far as his mistrust and reactions go, but even if the player is fully lit, getting spotted through a plain blind angle makes no sense. The question now is whether this is a map bug or game one.
I've attached the save as well, since this is an easily reproducible one.
-
The good thing is, the game does seem to scale with CPU as well. I switched from first-gen i7 to coffee-lake i7 some time ago, and an average "drawcall threshold" in FM rose from ~3000 to ~6000. That's a nice improvement, although we're talking 8 years between CPU mfg. dates, so it's not as impressive as you might think.
One thing you might want to know about is that the whole game graphics pipeline is currently very disrupted, so it's hard to measure things properly. Complex multi-stage materials and multi-material models are the main culprits (basically "anything goes", no limits). I'm trying to get back to more "by-the-book" setup with my custom environment, and while things look promising here, it's still far from comprehensive model and material library on my end.
So does this mean significant gains in performance would only come if most TDM main package textures and materials are remade and guidelines for mappers are imposed on creating custom ones? If so, that sounds like an overwhelming task ahead.
-
Yup, I mentioned that as well. If you have time, you can download a thingie called MSI Afterburner/RivaTuner. It's a bit complex to set up, but you can use it to monitor CPU and GPU use while you change options.
That's why I'm not so sure about shadowmaps replacing stencil shadows, at least in the current state of things. If you upgrade your rig / switch to a new-gen CPU, you'll see that stencil shadows work very nicely with it (you should have a lot of overhead here), while shadowmaps switch that load onto GPU, which is already busy with other things. I hope I'm wrong, but I got the impression that we may run into performance problems here as well.
As of the moment, on my old laptop, stencil shadows cause my CPU to overheat quite badly (while using experimental multicore support), but the game runs very smooth. In opposition, shadow maps make for a choppy framerate but are much more forgiving on the CPU. My question as of the moment is whether future updates will bring any significant improvements to this, especially with the further upgrades to multicore support planned along the line. It's no secret TDM is a resource demanding game due to the amount and detail of light sources and perhaps also due to the complexity of AI routines. I was just wondering what is the ceiling to be expected in the long run, especially for low-end PCs, since I consider support on legacy devices an important factor on the accessibility of most free games.
-
The AI will never see you as an enemy "immediately". There will always be a short period of reaction time entering combat state where the AI pauses to assess what he's seeing.
I think I expressed this back in the day as what to expect if you jumped down right in front of an AI. His reaction is going to be "Huh? What the hell is this? Hey! I found him, brothers!! To me!!". That thought process is going to take a bit of time.
Yeah, I agree with this. But indeed, let's do some further testing in the meantime.
-
Exactly, which is why I have been going back and forth testing AI and making these videos. I'm thinking of making a small map with light switches and guards placed on corridors to make the testing of AI reactions much more easy and standardized in the future. Since this is such an important component of TDM I found it surprising this bug managed to slip through since 2.04, which in my opinion calls for more thorough testing of AI every time a new release is produced. If you can also test this yourself in the meantime, please record a video and let us know your input!
-
1
-
-
I did some close range testing, and you're right, it does seem to be working fine for the most part, although I'm not entirely comfortable with the excessive delay on Nearly Blind:
I think there's an expectation that if the player is fully-lit and close by, that the AI will easily spot you and climb quickly to alert index 5 (combat) regardless of how good his vision is. Perhaps that's a factor that should be added to the current calculations. (??) That change, however, might work to the detriment of the player in less-than-fully-lit situations, which means some maps will play differently than they have for the past 6 years.
I always take it players engage into a simulation considering three important factors by this order:
1) What does the player expect based on the knowledge he has from reality?
2) What does the player expect based on what he knows from video game logic?
3) What does the player perceive as being a fair challenge given the previous points?
I think knowing how to balance a certain dose of common sense realism with just enough fun and challenge is a core pillar of good gameplay design. Based on this, I think certain aspects of AI behaviour should be difficulty agnostic. If you bump against an AI, it should immediately trigger level 4 or 5 awareness. The same should happen if you jump into full light at less than 3 meters distance from AI. Conversely I expect AI to hold a reasonable doubt if I'm at a considerable distance, even if in full light. However, audio cues should always be provided to make the player aware he has been detected, so as to make it fair even if not fully realistic, and so on.
I don't particularly think this would be something that severely alter how most maps are played, and then again, it could always be tested. So I think that we're on the right track to balance AI visibility just fine.
-
CuTN does look fine, but enabling this configuration leads to poor AI at close range as seen in the end of the Bakery Job video. I'll record another example from Tears of St Lucia later on.
Doing away with torches indoors: an appeal to mission designers
in TDM Editors Guild
Posted · Edited by zergrush
My point is that TDM is not (or at least no longer is) a Thief clone to begin with. It has a game base that is similar to Thief but it is much more developed in terms of AI, mechanics, physics, even lore. This challenges the assumption you should keep using certain tropes in mission design just because old Thief players are familiar with them, and ignores a whole potential player base of people who are not familiar with Thief.
Gas fueled flames can be put out with water, and besides that most TDM indoor lamps look like oil ones, not gas.
As for using arrows to break lamps, that would require adding new specific breakable objects and creating new "broken" versions of the mesh, which not only is a lot of work, but it also alters an element of difficulty that is supposed to be tolerated by players. I'm not saying torches should be banned or removed from missions, I'm just questioning their choice of placement from a pure design perspective, not a gameplay one. It's just that it looks better if mappers don't use them inside buildings. I am not requesting a feature here, nor for the game to be changed. And again, we already have a variety of TDM maps that do just that.