EdZ
-
Posts
19 -
Joined
-
Last visited
EdZ's Achievements
-
That's a really impressively short throw, and already has a mirror in the beam path. Would be really interesting to replace that mirror to make a short-throw dome screen.
-
The Rift comes with an XB1 controller. It's included at-cost from Microsoft (i.e. cost of plastic for mouldings, cost of PCB fab, ~$10) so you can resell it at a profit if you already have one. A good chunk of VR games will be using a controller rather than motion controlls, because: - A controller is what has been used for the past 3/4 years to develop games with - Many games do not work well with motion controls Whether games o nthe Vive will support the Steam Controller is entirely up to game developers. Probably less likely due to it not being bundled with the Vive, so devs can;t assume people will have one.
-
Vessel username: edzieba Videos: https://www.vessel.com/videos/LCoY5zfFf https://www.vessel.com/videos/JYZEYDYx0
-
There are definitely some technical deficiencies with the Vessel platofrm compared to even Youtube: - Video appears to max out at 1080p at 4800kbps (if the page source is to be beleived) at 30fps, with no stream selection granularity - No way to disable autoplay - Vessel does not respond to middle-click correctly (will hijack the same tab to open links rather than opening a new tab as desired) - No download copy (an expectation for any paid-for distribution platform) for offline/mobile viewing, particularly when this is so trivial with Youtube - It appears that all LMG videos have been uploaded with (or autoconverted by Vessel with) TV levels rather than PC levels Basically, apart from videos arriving earlier, Vessel is an inferior way to actually watch videos. Compared even to Youtube, which is hardly a shining bastion of usability.
-
That's my point: it's easier to work in than much larger cases due to the much more open access. e.g. it's a lot more easier to reach through the back of the case to release the PCI-E latch (which with an mITX board is right near the edge) to remove a GPU while lifting it out from the front, than it is to reach down from the front and try and slide your fingers undernearth the GPU to reach the latch. Not bad for a case that you could actually put inside the main cavity of a full size ATX case.
-
It's actually incredibly easy to work in, because all but the rear and bottom of the case can be removed to leave it completely open for access. Much easier to work in the my previous 550D and P180.
-
The unfortunate truth is: it's never going to work well due to fundamental technical limitations of available sensing technology. The IMU-based portion (sensing what the orientation of the band is, and by extension, the orientation of your arm) will work as well as any other IMU. Great for roll and pitch, Yaw drift dependant on local magnetic conditions (or just bad yaw drift, if the magnetometer is ignored entirely). But the myoelectric portion is never going to work well. For medical applcations, there are two ways to apply electrodes for effective myoelectric sensing that work in practice: 1 - A trained technician applies the electrodes to specific muscle groups, with the aid of the user performing tasks to flex those groups in order to lcoate the sensor properly 2 - For prosthetic applications, the same process is used to locate the electrodes initially, followed by the electrodes being embedded inside a stump-cup to keep them in a fixed position over the stump Just wrapping the arm in a handful of electrodes that cover a tiny portion of the arm, are not positioned over selected muscle groups, and may shift not only between uses but while in use, is not going to get good results. With a good training process and good filtering it may be able to reliable discriminate celnched-fist (many muscle groups activated) from relaxed-open-palm (few muscle groups activated), and maybe some degree of the current direction of wrist flexion and extension force*, but that's about it. *Not wrist orientation, but the force currently being applied to the wrist joint. Because the Myo can only sense muscle activation, it cannot tell your wrist bent all the way out and trying to superextend from your wrist bent all th way in and just beginning to swing back.
-
Not quite true. The DK2, which uses a Samsung OLED panel (the same panel as the Note 3) expereinces an issue nicknamed 'black smear'. This is due to pixel switching time (or pixel emitter charge time) between a dark and light pixel, combined with the high-speed gaze scanning due to the VOR (Vestibular Oculomotor Reflex) when turning your head. Because the Samsung panel specifically does NOT implement overdrive, this is an issue with pure black backgrounds and bright objects. Currently, the two ways to mitigate the issue is to raise minimum output levels (lose true black) or use 'post'-overdrive by modifying the values sent to the panel per-frame (truncates maximum and minimum brightness range, reducing contrast). This is more an implementation issue, but OLEd panels DO need overdrive, particualrly when illumination times are short (low persistance / lightboost / ulmb / etc).
-
DSR does not use naive bilinear/bicubic sampling to scale, so is not particuarly affected by non-integer scaling, and is fundamentally different to the way a display controller in an LCD panel will scale an image. DSR uses a 13-tap gaussian filter. i.e. a 13x13 pixel region of the 'source' image is sample and weighted to prodiuce every pixel of the output image. This obivously means that each pixel of the input image contributes to multiple pixels of the output image with different weightings, as this moving 13x13 region (called a 'kernel') will overlap.. Scaling a 3840x2160 image to 2560x1440 will look better than scaling to 1920x1080, because you have more output pixels available, which means you take more samples of the input image.
-
Looks like parts fo the EU have excellent pricing for the U3415W at the moment. In the UK, the cheapest it is available for is £680 from Amazon UK. Official pricing from Dell is £885. I've just ordered one from a German store, and even with shipping it works out to under £570 (€750).
-
Dead green subpixel. A 'stuck' pixel will display one or more colours (red, green and/or blue) on a black image. A dead pixel will fail to display one or more colours (red, green and/or blue) on a white image.
-
No, but I WAS using milliradians by mistake. The subtended angle as actually 8.6 arc-minutes, so even worse.
-
From the site: "As Steve Jobs said: "It turns out there’s a magic number right around 300 pixels per inch, that when you hold something around to 10 to 12 inches away from your eyes, is the limit of the human retina to differentiate the pixels."" This is false. 300ppi at 12" has each pixel subtend an angle of 2.9 arc-minutes. Even regular 'Minimum Seperable' acuity is 1 arc-minute, and Vernier acuity/Hyperacuity can resolve down to around 1 arcsecond. That website is sinple incorrect. Steve Job's definition of a 'retina display' is 100% marketing, with no basis in reality.
-
A 'retina display' is purely a marketing term, with effectively no scientific backing. PPI is measingless without knowing the distance from the eye to the display. There are multiple measures of human visual acuity, due to the multiple ways in which the brain processes signals from the eye, and all of these are based around the angle an object subtinds with respec to the eye (generally in the range of arc-minutes to arc-seconds).
-
If you need a HDMI 2.0 output, your current choices are either the GTX 970 or GTX 980. That's it.
