May 28, 2014 at 02:19Why does the Surface Pro 3 digitizer have a total of 256 pressure levels?
- Transfer
Now on reddit passes the AMA of the Surface development team. One of the questions was why in the third surf, they didn’t use the Vakomovsky digitizer, but N-Trig which offers a total of 256 pressure levels. People are worried that this will not be enough to confidently work with the stylus in drawing.
The answer turned out to be exceptionally entertaining and intelligible, and therefore I made a free translation. Some are not interesting things and lowered, so do not scold much. At the same time I will say that I translated some terms like touch digitizer as I could, so if you know an established translation, write in a personal.
Hi, this is Stevie B. I was expecting a digitizer question, and therefore pondered the answer a few days before AMA. I hope you and others interested find it useful.
At the moment, there are three types of pen digitizers: electromagnetic, passive-capacitive (in which the tip of the pen "pretends" with a finger), and active-capacitive.
Electromagnetic based on the receiving circuit board usually located under the screen and backlight. This board has nothing to do with the finger digitizer, which is usually located in front of the screen. On the board are a bunch of flat coils that emit an electromagnetic field (such as one part of a conventional transformer). The second part of the transformer is located in the lane. When the pen is brought to the screen, this part enters the EM field and adds load. This load is distributed over several coils and the pen position is interpolated from this data. This field works up to 15 mm above the screen and this makes it possible to realize a hover mechanism. Data from the pen (level of pressing, buttons) is transmitted by modulating the frequency. A secondary grid of coils and a bit of trigonometry are usually required to get the orientation of the pen. Orientation is needed to compensate for mechanical parallax. Of all the technologies, this is the oldest.
Passive-capacitive styluses simply imitate a finger, and work with a finger digitizer.
Finally, active-capacitive solutions began to appear in the early 2000s. They also work with a finger digitizer, but the tip of the pen introduces an electrostatic signal that is received by the capacitive lines of the digitizer. Imagine a mini-walkie-talkie that transmits signals to miniature antennas built into the screen. The crosshairs of capacitive lines (the place where the vertical and horizontal lines intersect), which receives the strongest signal, corresponds to the position of the pen. Of course, for the operation of such a pen, it requires batteries. By the way, you probably heard how a couple of years ago we bought the developer of a cool active-capacitive digitizer, Perceptive Pixel. I am sure this is the best pen technology for large screens. The market for this type of digitizer is actively developed by several manufacturers,
Now let's think about what makes the stylus different from the point of view of iron.
1. Accuracy rules the ball. The more accurate and stable the pen tip draws on the screen, the easier and more convenient it is to use. The accuracy problem depends on three parameters.
2. Tactile and sound sensations. The pen should feel and sound like you are writing / drawing on real paper.
3. The pen itself - its weight, material, ergonomics.
4. Stable and accurate pressure assessment.
5. Buttons for switching modes.
6. Delay in drawing. It is highly dependent on the application, and good developers strive to reduce it to a minimum.
7. Blocking the palm, in order to exclude false touches.
8. Integration with the device.
Now that we have a little understanding of the topic, let's discuss some of the advantages and disadvantages of the above technologies. I will deliberately skip the discussion of passive-capacitive digitizers in view of their apparent flaw. Also, I want to say that among these technologies, in general terms, there are no obvious leaders. It all depends on the specific implementation, and the integration of a specific digitizer into a specific device. I personally observed the miserable implementations of all three technologies by various vendors. So using some type of digitizer alone does not guarantee success.
1. The accuracy. Such styluses can be very accurate, but it depends a lot on the implementation. In order for such a pen to work well along the edges of the screen, the receiving printed circuit board behind the screen should be slightly larger than the screen itself. Also, metal objects and magnets near the screen or even near the device are strictly contraindicated, since they introduce distortions that greatly impair accuracy. These requirements limit the manufacturer to the screen design and materials used. And since EM fields drift depending on the environment, the vendor needs to calibrate the digitizer well, and also take into account situations where the user places a metal object in front of the device (for example, a case with a metal insert).
2. Visual parallax - depends on the thickness of the screen, and none of the technologies has a significant advantage in this regard.
3. Electronic parallax - since the receiving board is buried deep in the device, the digitizer must calculate the orientation of the pen and calculate the position from the orientation. The calculation of the position often depends on where exactly on the screen the tip of the pen is located. Therefore, you cannot write one mathematical transformation for all points on the screen. Rarely does anyone bother with this and usually choose a simpler path.
4. Stability and accuracy across the screen. The easiest way to check is to take a ruler and draw some diagonal lines across the screen. Estimate the evenness of the lines. Most likely they will not be perfectly smooth, because this is difficult to achieve.
5. The pen itself - its weight, material, ergonomics. EM stylus cannot be made of metal. They can be of any size and shape, from thin and uncomfortable (but which can be stuck in a socket in the device), to large and more like handles. They also do not need batteries.
6. Stable and accurate pressure rating. Usually, EM styluses are fine with this. By the way, it depends more on the pressure curve than on the number of bits being replicated. I will explain in more detail below.
7. Integration with the device - since the EM digitizer is separated from the finger digitizer, it will require about 0.4-1 mm thickness, several millimeters around the screen and several tens of grams.
1. The accuracy. In the past, I saw here (transl.: In MS, I suppose) not the best implementations, but I am very pleased that we achieved in Pro 3. We really raised the bar. The pen has become much more accurate and across the screen. The first thing I hear from artists who work with this pen for the first time is praise for accuracy.
2. Visual parallax - depends on the thickness of the screen, as I said. In Pro 3, we reduced it to 0.75 mm, and this is one of the lowest numbers I've seen on pen tablets.
3. Electronic parallax - since capacitive lines are located directly under the glass (which is only 0.55 mm thick!), Parallax is quite small, and this is one of the reasons why our feathers feel more accurately.
4. Stability and accuracy across the screen - do a test with a ruler!
5. The pen itself - its weight, material, ergonomics. Since we are not limited by material, we made a cool pen made of adonated aluminum, which fits comfortably in the hand. Yes, she needs a battery, but with a battery you can radiate a more powerful signal, which allows, for example, by clicking on the end to open OneNote from a distance of one and a half meters! You cannot do this with a passive pen.
6. Delay - as I said, the delay is highly dependent on the specific application. We really have one problem - a small lag is noticeable in hover, but this is only in hover.
7. Integration with the device - in comparison with the EM digitizer, there are practically no restrictions. For example, our Type keyboards cling to the surf through magnets, which would prevent the EM field.
Well, as promised, let's talk about the main Fedor, about pressure. Some here are worried that 256 levels of pressure are not enough. You know, you can promise any resolution, 10, 12, 14, 16 bits ... But in the end, even if the system produces a 16-bit number, this does not mean at all that it contains 16 bits of useful information. It’s like in cameras with 20 megapixel sensors - it also doesn’t always work out 20 megapixels of useful pictures. You can do the experiment that I did last weekend. I took the best EM tablet that I know and compared with Pro 3. First I downloaded a software called “digiInfo”, which can record Windows messages. I set it up to record pressure messages on both tablets. Then I made a small device that held the stylus and pressed on the screen with a force of about 50 grams(transl .: sorry I'm mixing grams with newtons) . I recorded the reported statistical pressure, imported it into Excel, and this is what I saw: for an EM pen with 1024 levels, the standard deviation was three times greater than for our pen with 256 levels. So, as a result, the performance is the same despite the extra two bits of resolution.
In other words, the Pro 3 pen measures pressure in the range of 10-400 grams and projects onto one of 256 levels, and nonlinearly, because human pressure is nonlinear. But about 1-1.8 grams per level is obtained. And the EM-pen measures in the range of 10-500 grams and theoretically gets about 0.4 grams per level.
...
from the translator:
Then I lost the thread of the conversation. Stevie started talking about scales and signal averaging.
But after a thoughtful analysis, it seems to me that he offers to take any exact scales and see how accurately you can set pressure with your finger. He apparently means that the 10-bit pen has excessive accuracy, which leads to excessive noise, a strong signal averaging, and, as a result, lag.
The answer turned out to be exceptionally entertaining and intelligible, and therefore I made a free translation. Some are not interesting things and lowered, so do not scold much. At the same time I will say that I translated some terms like touch digitizer as I could, so if you know an established translation, write in a personal.
Hi, this is Stevie B. I was expecting a digitizer question, and therefore pondered the answer a few days before AMA. I hope you and others interested find it useful.
At the moment, there are three types of pen digitizers: electromagnetic, passive-capacitive (in which the tip of the pen "pretends" with a finger), and active-capacitive.
Electromagnetic based on the receiving circuit board usually located under the screen and backlight. This board has nothing to do with the finger digitizer, which is usually located in front of the screen. On the board are a bunch of flat coils that emit an electromagnetic field (such as one part of a conventional transformer). The second part of the transformer is located in the lane. When the pen is brought to the screen, this part enters the EM field and adds load. This load is distributed over several coils and the pen position is interpolated from this data. This field works up to 15 mm above the screen and this makes it possible to realize a hover mechanism. Data from the pen (level of pressing, buttons) is transmitted by modulating the frequency. A secondary grid of coils and a bit of trigonometry are usually required to get the orientation of the pen. Orientation is needed to compensate for mechanical parallax. Of all the technologies, this is the oldest.
Passive-capacitive styluses simply imitate a finger, and work with a finger digitizer.
Finally, active-capacitive solutions began to appear in the early 2000s. They also work with a finger digitizer, but the tip of the pen introduces an electrostatic signal that is received by the capacitive lines of the digitizer. Imagine a mini-walkie-talkie that transmits signals to miniature antennas built into the screen. The crosshairs of capacitive lines (the place where the vertical and horizontal lines intersect), which receives the strongest signal, corresponds to the position of the pen. Of course, for the operation of such a pen, it requires batteries. By the way, you probably heard how a couple of years ago we bought the developer of a cool active-capacitive digitizer, Perceptive Pixel. I am sure this is the best pen technology for large screens. The market for this type of digitizer is actively developed by several manufacturers,
Now let's think about what makes the stylus different from the point of view of iron.
1. Accuracy rules the ball. The more accurate and stable the pen tip draws on the screen, the easier and more convenient it is to use. The accuracy problem depends on three parameters.
- Visual parallax - from the tip of the pen to the ink on the screen. This is where you think the tip of the pen is located.
- Electronic parallax - from the tip of the pen to the digitizer. This is where the digitizer thinks the pen tip is located.
- Digitizer accuracy and distortion linearity over the entire screen.
2. Tactile and sound sensations. The pen should feel and sound like you are writing / drawing on real paper.
3. The pen itself - its weight, material, ergonomics.
4. Stable and accurate pressure assessment.
5. Buttons for switching modes.
6. Delay in drawing. It is highly dependent on the application, and good developers strive to reduce it to a minimum.
7. Blocking the palm, in order to exclude false touches.
8. Integration with the device.
Now that we have a little understanding of the topic, let's discuss some of the advantages and disadvantages of the above technologies. I will deliberately skip the discussion of passive-capacitive digitizers in view of their apparent flaw. Also, I want to say that among these technologies, in general terms, there are no obvious leaders. It all depends on the specific implementation, and the integration of a specific digitizer into a specific device. I personally observed the miserable implementations of all three technologies by various vendors. So using some type of digitizer alone does not guarantee success.
So, let's start with electromagnetic styluses.
1. The accuracy. Such styluses can be very accurate, but it depends a lot on the implementation. In order for such a pen to work well along the edges of the screen, the receiving printed circuit board behind the screen should be slightly larger than the screen itself. Also, metal objects and magnets near the screen or even near the device are strictly contraindicated, since they introduce distortions that greatly impair accuracy. These requirements limit the manufacturer to the screen design and materials used. And since EM fields drift depending on the environment, the vendor needs to calibrate the digitizer well, and also take into account situations where the user places a metal object in front of the device (for example, a case with a metal insert).
2. Visual parallax - depends on the thickness of the screen, and none of the technologies has a significant advantage in this regard.
3. Electronic parallax - since the receiving board is buried deep in the device, the digitizer must calculate the orientation of the pen and calculate the position from the orientation. The calculation of the position often depends on where exactly on the screen the tip of the pen is located. Therefore, you cannot write one mathematical transformation for all points on the screen. Rarely does anyone bother with this and usually choose a simpler path.
4. Stability and accuracy across the screen. The easiest way to check is to take a ruler and draw some diagonal lines across the screen. Estimate the evenness of the lines. Most likely they will not be perfectly smooth, because this is difficult to achieve.
5. The pen itself - its weight, material, ergonomics. EM stylus cannot be made of metal. They can be of any size and shape, from thin and uncomfortable (but which can be stuck in a socket in the device), to large and more like handles. They also do not need batteries.
6. Stable and accurate pressure rating. Usually, EM styluses are fine with this. By the way, it depends more on the pressure curve than on the number of bits being replicated. I will explain in more detail below.
7. Integration with the device - since the EM digitizer is separated from the finger digitizer, it will require about 0.4-1 mm thickness, several millimeters around the screen and several tens of grams.
Active Capacitive Stylus
1. The accuracy. In the past, I saw here (transl.: In MS, I suppose) not the best implementations, but I am very pleased that we achieved in Pro 3. We really raised the bar. The pen has become much more accurate and across the screen. The first thing I hear from artists who work with this pen for the first time is praise for accuracy.
2. Visual parallax - depends on the thickness of the screen, as I said. In Pro 3, we reduced it to 0.75 mm, and this is one of the lowest numbers I've seen on pen tablets.
3. Electronic parallax - since capacitive lines are located directly under the glass (which is only 0.55 mm thick!), Parallax is quite small, and this is one of the reasons why our feathers feel more accurately.
4. Stability and accuracy across the screen - do a test with a ruler!
5. The pen itself - its weight, material, ergonomics. Since we are not limited by material, we made a cool pen made of adonated aluminum, which fits comfortably in the hand. Yes, she needs a battery, but with a battery you can radiate a more powerful signal, which allows, for example, by clicking on the end to open OneNote from a distance of one and a half meters! You cannot do this with a passive pen.
6. Delay - as I said, the delay is highly dependent on the specific application. We really have one problem - a small lag is noticeable in hover, but this is only in hover.
7. Integration with the device - in comparison with the EM digitizer, there are practically no restrictions. For example, our Type keyboards cling to the surf through magnets, which would prevent the EM field.
Well, as promised, let's talk about the main Fedor, about pressure. Some here are worried that 256 levels of pressure are not enough. You know, you can promise any resolution, 10, 12, 14, 16 bits ... But in the end, even if the system produces a 16-bit number, this does not mean at all that it contains 16 bits of useful information. It’s like in cameras with 20 megapixel sensors - it also doesn’t always work out 20 megapixels of useful pictures. You can do the experiment that I did last weekend. I took the best EM tablet that I know and compared with Pro 3. First I downloaded a software called “digiInfo”, which can record Windows messages. I set it up to record pressure messages on both tablets. Then I made a small device that held the stylus and pressed on the screen with a force of about 50 grams(transl .: sorry I'm mixing grams with newtons) . I recorded the reported statistical pressure, imported it into Excel, and this is what I saw: for an EM pen with 1024 levels, the standard deviation was three times greater than for our pen with 256 levels. So, as a result, the performance is the same despite the extra two bits of resolution.
In other words, the Pro 3 pen measures pressure in the range of 10-400 grams and projects onto one of 256 levels, and nonlinearly, because human pressure is nonlinear. But about 1-1.8 grams per level is obtained. And the EM-pen measures in the range of 10-500 grams and theoretically gets about 0.4 grams per level.
...
from the translator:
Then I lost the thread of the conversation. Stevie started talking about scales and signal averaging.
Think about both those numbers and that is both super super sensitive ... the best weight scale I have can do .1 gram increments .... The only reason it works is cause it averages the heck out of the numbers which adds a significant amount of lag ... this lag one cannot do on a stylus ... so you are stuck with a nosier signal comparatively in a stylus.
But after a thoughtful analysis, it seems to me that he offers to take any exact scales and see how accurately you can set pressure with your finger. He apparently means that the 10-bit pen has excessive accuracy, which leads to excessive noise, a strong signal averaging, and, as a result, lag.