cabalistic
-
Posts
1579 -
Joined
-
Last visited
-
Days Won
47
Posts posted by cabalistic
-
-
No, but you *can* do it on a subset and make reasonable statistical extractions about the plausibility of the overall numbers. Which, again, leads to my question what kind of massive mistake you expect to be hidden in those numbers? If COVID is not the cause of the excess mortality we are seeing worldwide, what is it then?
-
1
-
-
59 minutes ago, chakkman said:
The question is rather whether or not they died OF Corona. Which needs extensive post mortem examination, as usual. That's the important and interesting thing, when talking about the harm done by a disease.
The way you are wording this, you seem to believe that COVID did not cause a significant number of those deaths. May I ask why? What do you expect those extensive post mortems would find?
-
1
-
-
9 minutes ago, chakkman said:
Clinically confirmed eh? Just like the amount of Corona infections, I guess.
I can assure that they don't have the means to examinate 13.000 thousand people in such a short perdiod of time.. It's very expensive to do such investigations.
Very well, deaths of people with clinically confirmed Corona infections. So yeah, a few of those might have died even without Corona, and more will have been missed as they weren't tested.
-
Two wrongs don't make a right. Everyone involved in the terror attacks in France or elsewhere can go straight to hell. And all the people who were and are cheering them on, fuck them. If any Uighurs were cheering them on, fuck them, too.
That does not justify in any way what China is doing to the Uighurs. China isn't doing this because they fear terror attacks by the Uighurs, and even if they were, putting an entire collective of people under a general suspicion and punishing them is not acceptable and a gross violation of due process and the human rights we should hold high. @KurshokI don't know why you found it important to inform us of your political affiliations, as if that should excuse or justify your stance? But it really doesn't cast a great light on you when you are condemning the barbaric acts of Islamic terrorism and in the same post sympathize with barbaric acts of a different kind. If you want to maintain a moral high ground, you need to do better.
-
4
-
-
7 minutes ago, chakkman said:
You're right. Probably wrongly added a 0 to those numbers... what I did read though is that the numbers are a rough approximation. For the obvious reason that you can't do a full post-mortem examination for everyone.
They are a statistical estimate based on excess mortality. Certainly not accurate to the last count, but the order of magnitude is probably quite accurate. The amount of actual clinically confirmed influenza deaths is typically in the low thousands or even below. The current 13,000 COVID deaths reported for Germany, by comparison, are clinically confirmed cases.
-
14 minutes ago, chakkman said:
Regarding the 1.3 million Corona victims: Be careful with such numbers. There were lots of cases of countries counting "deaths with Corona". It's not a coincidence that the yearly deaths by influenza are counted in dimensions like 200.000 to 600.000 (in Germany for example), because obviously they can't do a post-mortem examination for every case, just like they can't with Corona. 1.3 million world wide is also not very much, I just stated the deaths by influenza in Germany alone. And, the WHO already corrected the mortality rate a lot downwards.
I don't know where you pulled those numbers from, but they are not even close to true. Influenza deaths in Germany typically range in the thousands to ten thousands, with approximately 25,000 reported for the particularly nasty season of 2017/2018, see e.g. https://de.statista.com/statistik/daten/studie/405363/umfrage/influenza-assoziierte-uebersterblichkeit-exzess-mortalitaet-in-deutschland/
600,000 deaths per year would be almost 1% of Germany's population...
-
1
-
-
Note that there are no gamepad input mappings active by default (except for the main menu). That's still something we need to sort out. Follow this guide to actually make the gamepad work: https://github.com/fholger/thedarkmodvr/wiki/Gamepad-support
-
2
-
1
-
-
uBlock Origin is the best adblocker. I don't know if that's what you are referring to, because there's also one simply called uBlock, and they are not the same. But either way, any blocker is only as good as the blocklists you've configured - perhaps you need some extra ones for the sites you are using?
-
On 9/19/2020 at 10:54 PM, Kurshok said:
It gets worse. A man was arrested at. the historical Speaker's Corner in London for "speaking against Sharia" or something along the lines of criticizing Muhammad and the Quran. Speaker's Corner is a historical place of Freedom of Speech in the world, with many famous people having had speeches there.
Source, please. I regularly hear these stories circulated about some arrest in some western country for speaking against Islam, and so far not a single one has turned out to be true...
-
1
-
-
I'm very curious to see how the 3080 and 3090 compare in real-world benchmarks. The price difference between those two is utterly ridiculous. It also feels like we might see a 3080(Ti) with 16 or 20 Gb in the future in-between those two...
-
1
-
-
FYI, I have a few other things I need to take care of at the moment, so development may slow down a little.
But so that you know what to expect, this is what I intent to work on next:
- Further performance experiments. I specifically want to try and implement fixed foveated rendering. The most notorious performance offender at the moment is a rasterization/fillrate limit due to stencil shadowing. With fixed foveated rendering I could reduce the pixel density at the outer regions, which should help significantly in those cases at a negligible quality decrease. It will, however, increase geometry processing costs, so I'll still need to do the experiment and see if and how much it actually saves in practice. On Nvidia cards, I can probably eliminate the extra geometry cost and even reduce it below what it currently is due to their support for the multiview extensions. Existing AMD cards, however, are SOL on that front...
-
After that, I will slowly move towards motion controls and room-scale tracking. Not all at once, obviously, and it'll probably take a few alpha versions before it's really playable that way, but at least it'll get going in the right direction
Unfortunately, since SteamVR's OpenXR support is not quite ready, yet, I will have to focus on the OpenVR backend, first. So initially only the OpenVR backend will feature controller support and roomscale, and only once I got that to a certain point I'll backport it to OpenXR.
-
2
-
5 hours ago, grodenglaive said:
Nice, this fixed needing parallel projections for the Pimax (5k+). I tested it out in the creepy Langhorne Lodge. Xbox 360 controller works to move around and to select the start menu items, but I wasn't able to map any of the buttons. Do I need to use a 3rd party app like Xpadder?
No, just follow the instructions here: https://github.com/fholger/thedarkmodvr/wiki/Gamepad-support
-
1
-
-
Alpha 6 is out: https://github.com/fholger/thedarkmodvr/releases/tag/v2.0.9_alpha6
It contains some minor improvements to culling with experimental support for non-parallel projections for canted displays (Index, Pimax). And I added experimental gamepad support, which some of you may prefer over keyboard/mouse until tracked controller support arrives
-
4
-
-
It's also not too bad to actually use sockets directly. For the most parts, Windows and Unix sockets are similar enough to make this manageable. I've done that a couple of times in the past.
-
35 minutes ago, stgatilov said:
I think it does not.
It does. r_ambientGamma directly affects the outcome of the ambient lighting stage, which is also applied to lightgem calculation. It's very easy to test - go to a dark corner, then increase ambient gamma and watch the lightgem get brighter...
Edit: actually, it's fine in 2.08, so this is a regression in 2.09. I'll look into it.
-
2
-
-
Note that r_ambientGamma has gameplay impact - it increases overall brightness and thus also player visibility. So setting that value high is not a good long-term fix. The issue is on the roadmap for 2.09.
-
1
-
-
I do mean core. The issue seems to have vanished as mysteriously as it appeared. I linked a current AppVeyor build, because I don't know when exactly the issue was fixed and if the latest tdm_installer build thus works or not.
-
1
-
-
On 7/6/2020 at 7:36 PM, stgatilov said:
I'm afraid I have no idea what could have caused it.
I don't have any idea what could have broken it, either. Just as I have no idea what fixed it in 2.09, but the overlay *is* working for my current dev build
If anyone wants to test this, here's a current dev build: https://ci.appveyor.com/api/buildjobs/ddl4n6spjmevbqy7/artifacts/TheDarkMod.7z
-
1
-
-
Trust me, the equivalent OpenGL or DX11 tutorials are significantly simpler
Well, the advantage is that it gives you a lot more control over the graphics pipeline, which also eliminates a lot of implicit assumptions and verifications needed from the driver. That *can* result in better performance and significantly reduced driver overhead, if you know what you're doing. Imho, Vulkan is aimed at professional engine developers, not hobby developers. Not that the latter can't use Vulkan, but it will eat a significant chunk of your dev time for questionable benefits.
-
1
-
-
4 hours ago, Diego said:
@STiFU Have you considered skipping ahead to Vulkan? From what I heard it drops a lot of inconveniences we have been dragging with opengl and dx and is easier to work with.
You have been misinformed on the "easier to work with" part
If you'd like to get an impression of what it's like to work with Vulkan, just take a quick glance at the https://vulkan-tutorial.com/ section for "drawing a triangle" and see what is involved in even just setting up an extremely simple scene in Vulkan. It's beyond crazy.
-
Glad you're enjoying it!
You can actually disable the deadzone by setting the cvar vr_decoupledMouseYawAngle to 0. You can also fully extend it to 180 degrees, I think - although without currently any other method to rotate the view, I don't think that would be playable
-
1
-
-
18 minutes ago, kerrle said:
I just tried this, streaming it wireless to a Quest via virtual desktop.
It worked great, and while I do miss the soft shadows and ambient occlusion, I didn't really have any performance issues in either Mission 1 or the first Penny Dreadful FM.
Congratulations on the work - even this early it's very much playable. Going forward, even without full touch controls I think bindings for touch controllers would make sense, and, eventually, it'd be cool to see some of the persistent UI(light gem, compass) maybe reworked into your off-hand model. I'd also suggest at least an option to decouple aiming (both frob and weapons) from the view. There's a Quake2 VR port that optionally does that while remaining mouse/kb driven, and I found that to work very well.
Even in this early state, though, inhabiting the City in VR feels great. I would definitely play missions this way.
Haha yeah, soft shadows, I'm afraid, would need an improved implementation of shadow mapping to become viable in VR (soft stencil will just never reach the necessary performance). And while that would be an interesting technical challenge, I think right now there are other things to spend my time on that would improve the VR port more. Ambient occlusion, though, isn't actually as big a deal in VR, interestingly. The thing is, I think a major part of its appeal in the flat game is that it helps improve the depth perception. But you already have plenty of that in VR. So while it does add some nice shading to the environment, it just doesn't have the same wow factor in VR - at least that was my experience when I did enable it.
Right now, I'm basically monitoring feedback and bug reports to try and get the experience as smooth as possible. I'll also try and see if I can improve the performance a little more so that there might be a little more headroom for at least 2x anti-aliasing, which would already be an improvement. But then motion controllers are in fact the next major item on my todo list. It will go through several stages, first steps will be very primitive and basically work more like a traditional gamepad. Eventually they will control the actual hands and weapons, and yeah, putting the lightgem on the hand model is an idea I also had - however, that all depends on whether I can find someone who can help me with the models for that. I'm just a coder
-
3
-
-
18 hours ago, Bienie said:
edit: Forgot to mention that it seemed to me like the lightgem is just a hair brighter in the VR version? While playing CoS 1 I noticed the AI giving first alerts in places I know for sure I never got them in standard TDM. Unless that is a 2.08/2.09 issue, I only played the map in 2.07 before.
It's difficult to say. If you have a specific place in the map in mind (ideally coordinates with getviewpos), I can check it out. I just played through Tears of St. Lucia in VR, and the lightgem seemed alright. Doesn't mean that there might not be a few instances where I messed something up, though
It might also help to recalibrate your seated position. If it's not properly calibrated, your actual player avatar might be in a slightly different position than you think, which could also explain some lightgem value differences.
-
2
-
-
3 hours ago, Bienie said:
I feel like maybe performance is slightly lower than in earlier builds though, as I ran into a little bit more reprojection than before. Rain seems pretty taxing (Langhorne Lodge seemed to run fine anyway, though I noticed my fps dip down to low 40's in the outdoor scenes). Played a bit of CoS 1: Pearls and Swine tonight and got some reprojection in scenes with multiple lights, but nothing too terrible. Going to finish it tomorrow evening. Not sure if I dare try to play The Painter's Wife for performance fears, but I guess I should try it as well to see how it fares.
So, out of curiosity I installed TDM 2.05 - which is dead simple with stgatilov's new tdm installer, btw. Holy shit, was that first VR alpha bad. Hard to believe I ever released it in that state
As far as performance goes, it's a bit of an interesting picture.
The new builds win leaps and bounds when it comes to CPU performance - that is not surprising to me, as I've invested a lot of effort in that regard. CPU performance in the old builds does cause quite a bit of reprojection in certain scenes e.g. in "A new job", whereas with the new builds I can go through the entire mission without reprojection.
However, GPU performance may have deteriorated somewhat. The old builds don't have the profiling tools of the newer ones, so it's difficult to pinpoint what's going on. But looking at SteamVR's frame graph, the GPU is spiking a bit more in the newer build. The baseline looks very similar to 2.05, but every third or fourth frame GPU time spikes, and as a consequence there may be less GPU headroom. Which, consequentially, could cause more reprojection in GPU-heavy scenes, as scenes with many lights potentially are. I think this is probably a "regression" in the base game, but definitely worth looking into. Although, it isn't really a big difference, and I also need to point out that the comparison is a little unfair, as the old build does have some graphical glitches.
Also, I have briefly tried The Painter's Wife, and as you would expect, it is not a smooth ride throughout. The open city parts are prone to reprojection - there's just too many lights at once. It wasn't outright bad, though. Might still be very playable, depending on your tolerance for reprojection.
-
3
-
The Game Awards 2020
in Off-Topic
Posted
I've only played TLOU2 and like 30 mins of Hades from those nominees. Good games, but GOTY? My personal GOTY would be Half-Life: Alyx, by a mile, I actually have to think hard what other games I even played from this year