Jump to content

Lisias

Members
  • Posts

    4,361
  • Joined

  • Last visited

Reputation

4,890 Excellent

Profile Information

  • About me
    Boldly crashing what no Kerbal has crashed before!
  • Location
    Universe ! Virgo ! Milkway ! OrionArm ! SolarSystem ! Earth ! America ! SouthAmerica ! Brazil ! SãoPaulo ! Capital ! Home ! LivingRoom ! MyChair
  • Interests
    I felt a great disturbance in the Force, as if millions of lines of code cried out in Null Reference Exceptions and were suddenly flooding the KSP.log...

Recent Profile Visitors

13,680 profile views
  1. Because I'm addressing your objections one at time. The last link was about how human eyes work, addressing your terrible statement that human eyes would be a kind of special devices completely unrelated to the Law of Physics. [snip] Exactly. What hints you don't need animations using fps faster than 75Hz, the whole point of the discussion. One just can't "perceive" anything faster that the brain can process it. Eyes are not autonomous devices, they need feedback from the brain in order to work - for example, to size the iris aperture to cope with the current illumination levels. Why do you think first-responders illuminate the iris with a flashlight to check the brain's responses? Not to mention the eye's cones themselves (as the rods are not useful on gaming as their "resolution" is too low, se my previous video) also have their limitations. The cones have a point of saturation where they don't work properly, and measuring these timings get us some very interesting data: This article I linked is terribly dense, hard as hell to extract layman understandable data - I took some time to get it), but in essence what it is saying is that once the rods are "overexposed", they took about 13 ms (interesting number) to cease working, and then another 7ms to get back to work. Monitors (and VR headsets in special) are nothing more than millions of small flashlights pointed to your eyes, by the way. Not being enough, I found an article where the rods and cones response time are being measured (forget about the brain, now we are talking directly to the photoreceptors on the eyes). These tests were made using saturation (and using photoreceptors from different animals so they are not directly applicable to humans). But some numbers are consistent with the discussion. Rods saturates after approximately 10 to 15ms (cones saturate way faster), and once they are saturated, they cease to work as expected. Unrelated to the discussion (we are talking about how fast the brain can perceive visual stimuli, not how slow the stimuli can be before annoyance), I found this interesting article - perhaps this could help to explain why you "perceive" some things differently from what science says:
  2. Apparently my lack of understanding is pervasive in the scientific community.. On the other hand, I learnt that the minimum FPS for preventing headaches (that could be severe to the point of causing nausea) is 90Hz per eye. Appears to be related more to eye strain and eye fatigue than perception though.
  3. Nope. The argument is exactly about how low framerates looks on the movie, and yet people still prefers it by reasons out of scope of this discussion. Being "technically better" doesn't automatically improves the experience. Nope. Your brain takes at least 13ms to process an image. Then it needs time to propagate the muscle's commands from the brain, another 16 to 25ms. So you have something like 26 to 51ms between the stimulus and the reaction (something between 40 to 20 Hz!!). So, you see, we have a significative lag between what we see and what we do about what we see. Considering that we perceive visual clues way faster than we react to them, 75Hz appears to be the best compromise. Not exactly. What's happening here is that the input controls are (wrongly on my opinion) tied to the framerate, Since we have a delay of about 26 to 51ms between the perception of the stimuli and the reaction, and the stimuli themselves takes down to 13ms to be processed, we have an additional 13 ms to confirm the results of that reaction. Let's assume a very responsive youngman whose reaction times are on the top of the species: 13ms to process the visual stimuli, and 13 ms to propagate the synapses into muscle's signal receptors. 26ms total, or 2 frames at 75Hz. You need an extra frame to process the results of what you did, so between the original stimuli and the processing of the feedback, you "waste" 3 frames and the cycle ends at the 4th frame (remember Nyquist - anything happening between two sampling points are detected on the second sampling point, unless it's duration is less than the delta-T of the sampling points when it is plain lost). On a faster framerate with the inputs attached to the frame, the computer will be able to show you the feedback earlier on that pipeline. Not because of the higher framerate (because you don't see anything below 13ms anyway), but because the processing of your reaction is on the critical path of the frame pipeline, what delays the resulting frame for you - and so fastening the whole pipeline ends up accelerating showing the resulting frame to you. So, yeah, on a system where the input is attached to the frame, going 120 or 150Hz will improve perception because the processing of the resulting frame will be ready before you can perceive a frame. But this is happening not because doubling the framerate is perceptible by you, it's happening because someone tied the input and the respective processing to the frame's critical path, delaying the results to you. Had the computer be programmed to receive the input and process the reaction at it maximum speed instead of be forced to pace down by the frame rate, you could have the feedback displayed already on the start of the 3rd frame from the stimulus that triggered the chain process. You can't "feel" anything faster than 13ms. Feel free to provide any material supporting your argument. Here follows mine: I don't know how VR headsets works nowadays, but from the days we were toying with it (we were using Nintendo Power Gloves and LCD shutter glasses both tied to a parallel port), the problem is that each eye receives half the framerate, as the computer needs to draw two frames instead of one, one for each eye. So the 60Hz CRT monitors we were used to have would render 30Hz for each eye tops, and people using more expensive 90Hz monitors reported better results for fast pacing animation, not to mention way less strain on the eyes (Anyone played Sega Master System with a 3D glasses? I did, lots of fun - but could not play it for too much time! ) In a way or another, the sickness is reported to be related to the disruption between the aparent movement shown in the headset and the lack of real movement reported by your body. This is disruptive for your brain. Our brains are biological/chemical computers and the Laws of Physics still applies. We have limits on our sampling rates we use to perceive reality. Sources, please.
  4. O didn't said 24fps is acceptable for games, I said it's still preferable for movies. People's perception was the core of the argument. agreed on all accounts! I think we are not on the same page anymore? Low fps don't make things slower, make them 'chumkier'. An object that crosses the screen from on side to another in 2 secs will do it in 2 secs no matter.bow many frames it takes - 50, 100, 120 or 288. About the input, you don't need to attach the input to the framerate. Doing input on the critical path of a time critical computing is far from being a good idea a anyway. Every single academic text I ever read about real time computing advocates againt doing I/O on time critical code. And, in a way of another, the real bottleneck for a smooth animation is the Monitor's refresh rate. 4k monitors do 60 to 70Hz, and most affordable 1080p oned like mine does 75Hz. So stressing the GPU beyond that just don't improve the experience. Not to mention studies about human vision stating that we can process images as fast as 13ms (from 80 to 13, as I read) and so Monitors faster than 75Hz are just not helpful - 1000 / 75 is 13.33333, the most common limit for the brain to process human vision. I think that 75Hz with vsync is probably the same experience as a 144Hz image. But 'cheaper'.
  5. On the 80s, Americans used to comply when watching TV on Europe, as they were used to 60Hz and on Europe they used 50Hz as the frame refresh rate - and the flickering was annoying. But the 60Hz also flickers the same - only on a faster pace. So what was the problem? Perception. After some time, people got used to the frame rate and stopped to perceive it. You see differces between 60 and 144 fps because you want to see such differences. Once you start to pay attention to the game itself, you stop to perceive it. There's a reason most movies are still at 24fps nowadays. TLDR: 60fps or even 120 fps are better technically, but not necessarily desirable. Most people get more satisfaction at current 24fps, and that ends up settling the matter.
  6. that's the point in which I think things are going to change. Once you sign anm NDA and agree to oy share what they want you to share, the influencer reliability is co.promised. it's interesting to see how youtubers are being forced to specify every possible relationship with the promoted product. The point is: once you sign an NDA or have some other profit from the product maker, the influencer's credibility is somewhat hindered - who would expose an inconvenient true if this would hinder a profitable relationship with the product maker? This is exactly what lead the old media to lose public confidence at first place. on the other hand, my friends that bought the game told me that they were going to wait for the game to be stable instead of asking a refund because they liked the game very much. What would be the point to ask for a refund if they were going to buy the game again later? No trailer would do that. on that, we have an agreement. Me too.
  7. Alfred D'Harling's "Aero-unicycle". Source Not sure if this one is related (but it's remarkably similar): Source. Marvelous deat-trap. I say…
  8. Feel your pain, same thing here. I have some parts standing on the storage for 9 months. And the backlog doesn't stop to grows… (sigh) Once I finally manage to fix something, another one breaks (when not two) and... Nice rig. I bought a Dell SC1425 for peanuts two years ago, spend a bit of money expanding the CPU and the memory, and now that damned thing is one of the fastest computers on the house!!! (and the loudest… ) Probably not. But the final word on the subject is on the motherboard's service manual. Some high end motherboards are also used on Workstations, where non ECC memory is preferred (as they are slightly faster than ECC counter-parts, and for some tasks this matters), but since the thingy is on a 1U case, it appears to be highly customised as a server and I doubt it will accept non ECC memory sticks.
  9. The majority of people relies nowadays on third-parties to inform themselves - some of them we call "influencers". For the good and for the evil, they are here to stay. You talk to a few, these few talk to a lot, and every one of the latter ones will talk to another few. It works, and it's also pretty cheap. The "good old days" of mainstream media are over (again, for the good and for the evil). I beg to differ. "My friends are playing this, and they told me it's fun" is a hell of a fact. What happens is that the "facts" that matter to the users are not usually understandable by software developers (if by anyone at all). I once read an article about Gerry Anderson, and how he was annoyed by having "puppet show" as his most famous work, apparently he would like to be remembered by something more… "grown up", as a drama or something. The problem, as he explained once on a interview, is that you just don't have control about what's going to be a success and what's not - he said he didn't knew what made Thunderbirds such a huge success, he only knew that once it happened, how to keep things going on (at least, for some time). Of course nowadays there're people specialised on building factoids, that so can be used to promote things and get some advantage, but usually these things just don' "stick" for much time. Why the first Star Wars trilogy was such a huge success? Why the second trilogy was so controversial, besides technically superior (by a parsec)? The one that manages to answer these questions will probably dominate the World... But they can be easily refunded nowadays, so the pretty trailers are not enough anymore. There was a time in which once the sell was made, it was done. Good luck trying to get a Refund for your Windows 98. Now people are returning games - Steam allows you for a "no questions asked" refunding for a few hours after the buy, and some countries' legislation (as mine) demands a full refund for up to 1 month (or something) for goods considered unfit if they were bought online (and this can hurt a publisher badly if software would be considered a good, instead of an asset licensed to be used as is usually done nowadays). Didn't knew that, but I know about Mandalorian's virtual set using Unreal - so, not exactly a surprise. (but amusement parks? they are using Unity to control roller coasters? Jeez… #boeingFeelings) Been there, screwed by that. Once this happens, the company has little to no incentive to invest on technical excellence, as marketing (both good and evil), P/R stunts, and even not exactly capitalist tactics will render better results on the short run. In essence, it's the other way around - anyway pursuing technical excellent is expelled, as he/she would be a threat to the "business model".
  10. Hi! Thanks! (and also with older ones, TweakScale is working fine down to KSP 1.4.1! ) Yep, I can confirm it. It's happening on KSP 1.8 and newer, I just tested this from 1.12 downto 1.7.3, and on 1.7.3 the bug didn't manifested. You found a new way of doing the problem - I was aware of this bug when scaling up, when parts would be misplaced outwards: https://github.com/net-lisias-ksp/TweakScale/issues/163 I reproduced your bug report on this issue, by the way. It will help me later. In time, another workaround is to detach the part that is placed wrongly and re-attach it. This is easier when the part is scaled down, as it will be misplaced inwards the attachment point. As a side note, the problem is happening probably on the OnCopy thingy inside KSP, that changed somehow on KSP 1.8.0 - probably the order of events changed a bit, and this screwed up the way TweakScale does things. This is on TweakScale, the add'ons on your list are innocent on this one (say it mentally with the voice of the Ghost Rider! ). And thanks for the demonstration, it helps to have new ways of reproducing the bug! Anyway, since there's a workaround for it (two, with yours) and this is something that does not propagates (i.e.,, once the part is placed right, it stays there on launch - what hints me this is a problem related to the Editing Time code, not on the runtime), I will keep my previous decision to work on it on the future release 2.4.5.3 . (I'm working right now on 2.4.5.2, updating the patches to support new parts from KSP 1.11 and 1.12 - and some others that ended up being forgotten in time! Rushing it a bit to have it ready for this WeekEnd - but since I said it, chances are that Murphy had heard it and will try to screw me up on this… ). Cheers!
  11. And then we would play Wack A(toms) Mole on them! #hurray!
  12. Where was these patches? On the Companion? If they were on the Companion, it's something I need to investigate... In a way or another, great you solved it! Cheers!
  13. ANNOUNCE KSP Recall 0.2.0.3 is on the Wild, featuring: Mishaps on the sanity checks were detected and fixed This Release will be published using the following Schedule: GitHub, reaching manual installers and users of KSP-AVC first. Right now. CurseForge. Right now. SpaceDock (and CKAN users). Saturday Night. The reasoning is to gradually distribute the Release to easily monitor the deployment and cope with eventual mishaps, as I will have a hell of a Week and can't handle too many issues if something wrong happens. Cheers!
×
×
  • Create New...