|
On February 24 2012 06:39 Medrea wrote:Show nested quote +On February 24 2012 03:48 Jooms wrote:
Just got my Kana.
Is there anyway to turn off mouse prediction?
Having the same horrible problems with angle snapping as I did with my mx518. No, since every mouse has prediction and needs it to stay functional. Increase the polling rate. Like you could have done on your MX518. There might be a firmware update for the Kana to recalibrate the prediction though.
I would like to nip this in the but right now, but the polling rate of a mouse cannot change the amount of angle snapping in a mouse, period. You'll still want to increase your polling rate for other reasons, just not to combat angle snapping because it can't.
I'd go into more detail but I have an 9K word post incoming soon (I said it was long, lol), and am waiting on verification over a number of items. In the meantime, this is what you want to look at:
http://www.esreality.com/?a=post&id=1933667#pid1933667
|
Please refer to page 22 of this thread as well as the source material.
On January 17 2012 09:56 Medrea wrote:Show nested quote +On January 17 2012 09:36 Neurosis wrote:On January 17 2012 09:15 Medrea wrote:On January 17 2012 09:01 Neurosis wrote:On January 17 2012 08:46 Medrea wrote:On January 17 2012 08:05 Neurosis wrote:On January 17 2012 07:50 ashLoo wrote:
Anyone have any negative feedback for the Steelseries Sensei? Just wondering if there is any as I've already asked a handful of CS 1.6 and CS:S pro's and have found none thus far. Sensei has hardware acceleration that you can't get rid of. If you're looking for a mouse without a major flaw you won't find one, it doesn't exist yet. MX 518 or 518 legacy and Intellimouse exist and have nothing stupid. Neither feature insane DPI though so if you NEED 3600 DPI for whatever reason, I dunno heart monitor, then these wont work. Although I dont think you can adjust the DPI on the intellimouse, which leaves just the MX 518 in my eyes. I think the G500 and the G400 continue the legacy though. mx518 has prediction. So does every single mouse in existence. Point? Prediction is a necessity or the mousing experience is awful as the cursor jumps at every thread. Not every mouse has prediction, not sure what you're talking about. http://www.overclock.net/t/1151416/mouse-prediction-questionFor instance, I'm pretty sure deathadder doesn't have prediction. Yes it has prediction. Every single modern optical mouse has prediction. A mouse has to have prediction in order to fill in the blanks between going from point A to point B and to make sure surface problems are ironed out. You can have too much prediction yes. The MX518 1800 DPI has more prediction than its 1600 legacy variant. But you cant have none. If you want zero prediction you might as well go to McDonalds or whatever and buy a trash 5 dollar McMouse because your mousing is gonna be garbage anyway. In order to avoid too much prediction in your mousing experience, the polling rate has to be raised. That way the mouse has to predict less. Make sense? I have a picture somewhere that demonstrats it. Found on that exact same website in fact. if you poll a 518 to 1khz its perfect. EDIT: Found the article. http://www.overclock.net/t/771171/which-mouse-has-no-accel-correction-etc/60![[image loading]](http://i7.photobucket.com/albums/y267/turbosupercharger/image004.png) ![[image loading]](http://i7.photobucket.com/albums/y267/turbosupercharger/image005.png)
Now these mice still exhibit prediction. And it gets magnified by your polling rate (that is, inverse of your polling rate). And while you can never truly get rid of it completely, you can reduce its impact on your mousing experience to very low. To a point where it stops pushing your cursor around except for the most fastest of hand movements.
EDIT: Actually allow me to clarify this a little bit.
Prediction and Angle snapping are not the same thing. However the visual impact angle snapping gives is multiplied by the amount of prediction occuring. MX518 has at default a low polling rate, and some angle snapping. When the polling rate is amped up, the multiplied impact that angle snapping gives is lessened, as in the above explaination.
A lot of this stems from MSPaint drawings which IMO are not the letter of the law.
Also Bullveyr says that this occurs at the sensor level, it in fact occurs at the firmware level. If it truly occured at the sensor level you would not be able to change it via firmware updates but we have seen updates correct this (g400). The sensor itself is a camera that takes many more pictures than what the update rate of the mouse says it is. AFAIK prediction occurs at the sensor level, angle snapping is a formula that is firmware.
One again, this is a mouse with zero prediction.
![[image loading]](http://i7.photobucket.com/albums/y267/turbosupercharger/image001.png)
Not literally mind you but you get the idea.
|
Either way this is unrelated to Jooms' problem. If his problem is anything like the last user who had the exact same problem with the exact same mouse on the exact same OS, it is that Vista is borked. The Youtube video had the pixel jumping around quite a few pixels vertically and horizontally at a time.
Standard prediction and angle snapping is nowhere near as intrusive, even at the bottom polling rate.
|
On February 23 2012 17:49 Medrea wrote:Show nested quote +On February 23 2012 16:48 Blasterion wrote:On February 23 2012 09:45 IPS.Blue wrote:On February 23 2012 09:26 Blasterion wrote:should be about right maybe I am underestimating the length of 1.5 inches. I am actually not sure myself hahaha ![[image loading]](http://i.imgur.com/z725I.png) It's like you tried to break Warp 10 and destroyed your whole mouse setup in the process. You should not confuse mouse speed with actual hand speed. A fast mouse does not automatically make you good or God. It's more like your mouse should be fast enough, but not faster. You have to do the rest. How fast is warp 10? Depends on the Star Trek canon you are reading I suppose. http://en.memory-alpha.org/wiki/Warp_10#Warp_ten_and_aboveUh, most recently it was infinite speed, occupying all space in the universe at the same time.
Wow that's pretty fast. So I am still trying to get used to no acceleration mouse. but I think I'll get used to it soon
|
On February 24 2012 07:16 Medrea wrote:Please refer to page 22 of this thread as well as the source material. Show nested quote +On January 17 2012 09:56 Medrea wrote:On January 17 2012 09:36 Neurosis wrote:On January 17 2012 09:15 Medrea wrote:On January 17 2012 09:01 Neurosis wrote:On January 17 2012 08:46 Medrea wrote:On January 17 2012 08:05 Neurosis wrote:On January 17 2012 07:50 ashLoo wrote:
Anyone have any negative feedback for the Steelseries Sensei? Just wondering if there is any as I've already asked a handful of CS 1.6 and CS:S pro's and have found none thus far. Sensei has hardware acceleration that you can't get rid of. If you're looking for a mouse without a major flaw you won't find one, it doesn't exist yet. MX 518 or 518 legacy and Intellimouse exist and have nothing stupid. Neither feature insane DPI though so if you NEED 3600 DPI for whatever reason, I dunno heart monitor, then these wont work. Although I dont think you can adjust the DPI on the intellimouse, which leaves just the MX 518 in my eyes. I think the G500 and the G400 continue the legacy though. mx518 has prediction. So does every single mouse in existence. Point? Prediction is a necessity or the mousing experience is awful as the cursor jumps at every thread. Not every mouse has prediction, not sure what you're talking about. http://www.overclock.net/t/1151416/mouse-prediction-questionFor instance, I'm pretty sure deathadder doesn't have prediction. Yes it has prediction. Every single modern optical mouse has prediction. A mouse has to have prediction in order to fill in the blanks between going from point A to point B and to make sure surface problems are ironed out. You can have too much prediction yes. The MX518 1800 DPI has more prediction than its 1600 legacy variant. But you cant have none. If you want zero prediction you might as well go to McDonalds or whatever and buy a trash 5 dollar McMouse because your mousing is gonna be garbage anyway. In order to avoid too much prediction in your mousing experience, the polling rate has to be raised. That way the mouse has to predict less. Make sense? I have a picture somewhere that demonstrats it. Found on that exact same website in fact. if you poll a 518 to 1khz its perfect. EDIT: Found the article. http://www.overclock.net/t/771171/which-mouse-has-no-accel-correction-etc/60 Now these mice still exhibit prediction. And it gets magnified by your polling rate (that is, inverse of your polling rate). And while you can never truly get rid of it completely, you can reduce its impact on your mousing experience to very low. To a point where it stops pushing your cursor around except for the most fastest of hand movements. EDIT: Actually allow me to clarify this a little bit. Prediction and Angle snapping are not the same thing. However the visual impact angle snapping gives is multiplied by the amount of prediction occuring. MX518 has at default a low polling rate, and some angle snapping. When the polling rate is amped up, the multiplied impact that angle snapping gives is lessened, as in the above explaination. A lot of this stems from MSPaint drawings which IMO are not the letter of the law. Also Bullveyr says that this occurs at the sensor level, it in fact occurs at the firmware level. If it truly occured at the sensor level you would not be able to change it via firmware updates but we have seen updates correct this. The sensor itself is a camera that takes many more pictures than what the update rate of the mouse says it is.
The information/graphs you are referencing are coming from a marketing employee at Zowie, who was incorrectly creating links from polling rates to the amounts of prediction in their EC series mice. This information sounds good, but there are serious flaws once you get into the actual technical operation.
How often a host system initiates the gathering information cannot influence whether or not the DSP in the Sensor (I can't confirm %100 it is actually the dsp yet), or the MCU changes the delta x/y values that the IAS records. The fact is in mice with angle snapping, these values have already been changed either by the time they reach the MCU, or at the MCU, before they are translated into USB or PS/2 signals and sent back to the host, meaning that the polling rate cannot influence "how much" prediction there is, no matter how "pretty" someone presents their information. And prediction, angle snapping, drift control, or any other term you've heard of are ALL used to describe this specific path correction algorithm, meaning they are exactly the same.
Also, Bullveyr clarified in a different thread that in regards to steelseries mice like the sensei and xai, angle snapping is coded at the MCU and not at the sensor. I still am waiting to have some things clarified for me but I also should mention that the Avago 3080 and 3080E all have angle snapping, and you can't remove it by flashing the firmware. The updated G400's use the newer 3090 and 3095 which have two firmwares, one with angle snapping, one without, from what I remember. The latter might be something like the 9800 which has a specific address set for angle snapping, but again, I'm waiting for clarification on this.
|
Oh. Well in that case yes path correction is not influenced by the speed by which the buffer is read but when the OS is forced to draw a line between the polling points higher polling rate clarifies the line.
The end user inaccuracy remains the same, this is true, but we are not talking about a whole heck of a lot of pixels. Less then a mouse with 0 prediction for sure. You can never be perfect afterall.
To show people what wolfwood is referring to. Look at the following:
+ Show Spoiler +
I am a white man so Im not the most exact circle drawer, the red line is MSpaint auto circle tool. Black line is me.
Can you tell which one is 125hz polling, and which is 1khz polling?
+ Show Spoiler +The left one is 1000khz polling
Youll also note that despite this mouse have an "angle snapping problem" I can keep the line very well. The straight points are not even longer than the straight points on the original circle.
On February 24 2012 08:29 Blasterion wrote:Show nested quote +On February 23 2012 17:49 Medrea wrote:On February 23 2012 16:48 Blasterion wrote:On February 23 2012 09:45 IPS.Blue wrote:On February 23 2012 09:26 Blasterion wrote:should be about right maybe I am underestimating the length of 1.5 inches. I am actually not sure myself hahaha It's like you tried to break Warp 10 and destroyed your whole mouse setup in the process. You should not confuse mouse speed with actual hand speed. A fast mouse does not automatically make you good or God. It's more like your mouse should be fast enough, but not faster. You have to do the rest. How fast is warp 10? Depends on the Star Trek canon you are reading I suppose. http://en.memory-alpha.org/wiki/Warp_10#Warp_ten_and_aboveUh, most recently it was infinite speed, occupying all space in the universe at the same time. Wow that's pretty fast. So I am still trying to get used to no acceleration mouse. but I think I'll get used to it soon
Bet its nice being able to click on more than a quarter of your screens pixels for a change.
|
On February 24 2012 09:02 Medrea wrote:Oh. Well in that case yes path correction is not influenced by the speed by which the buffer is read but when the OS is forced to draw a line between the polling points higher polling rate clarifies the line. The end user inaccuracy remains the same, this is true, but we are not talking about a whole heck of a lot of pixels. Less then a mouse with 0 prediction for sure. You can never be perfect afterall. To show people what wolfwood is referring to. Look at the following: + Show Spoiler +I am a white man so Im not the most exact circle drawer, the red line is MSpaint auto circle tool. Black line is me. Can you tell which one is 125hz polling, and which is 1khz polling? + Show Spoiler +The left one is 1000khz polling Youll also note that despite this mouse have an "angle snapping problem" I can keep the line very well. The straight points are not even longer than the straight points on the original circle. Show nested quote +On February 24 2012 08:29 Blasterion wrote:On February 23 2012 17:49 Medrea wrote:On February 23 2012 16:48 Blasterion wrote:On February 23 2012 09:45 IPS.Blue wrote:On February 23 2012 09:26 Blasterion wrote:should be about right maybe I am underestimating the length of 1.5 inches. I am actually not sure myself hahaha It's like you tried to break Warp 10 and destroyed your whole mouse setup in the process. You should not confuse mouse speed with actual hand speed. A fast mouse does not automatically make you good or God. It's more like your mouse should be fast enough, but not faster. You have to do the rest. How fast is warp 10? Depends on the Star Trek canon you are reading I suppose. http://en.memory-alpha.org/wiki/Warp_10#Warp_ten_and_aboveUh, most recently it was infinite speed, occupying all space in the universe at the same time. Wow that's pretty fast. So I am still trying to get used to no acceleration mouse. but I think I'll get used to it soon Bet its nice being able to click on more than a quarter of your screens pixels for a change.
Well it's pretty nice and I don't have over lapping boxing marines as much why more.
|
On February 24 2012 09:02 Medrea wrote:+ Show Spoiler +Oh. Well in that case yes path correction is not influenced by the speed by which the buffer is read but when the OS is forced to draw a line between the polling points higher polling rate clarifies the line. The end user inaccuracy remains the same, this is true, but we are not talking about a whole heck of a lot of pixels. Less then a mouse with 0 prediction for sure. You can never be perfect afterall. To show people what wolfwood is referring to. Look at the following: + Show Spoiler +I am a white man so Im not the most exact circle drawer, the red line is MSpaint auto circle tool. Black line is me. Can you tell which one is 125hz polling, and which is 1khz polling? + Show Spoiler +The left one is 1000khz polling Youll also note that despite this mouse have an "angle snapping problem" I can keep the line very well. The straight points are not even longer than the straight points on the original circle. On February 24 2012 08:29 Blasterion wrote:Show nested quote +On February 23 2012 17:49 Medrea wrote:On February 23 2012 16:48 Blasterion wrote:On February 23 2012 09:45 IPS.Blue wrote:On February 23 2012 09:26 Blasterion wrote:should be about right maybe I am underestimating the length of 1.5 inches. I am actually not sure myself hahaha It's like you tried to break Warp 10 and destroyed your whole mouse setup in the process. You should not confuse mouse speed with actual hand speed. A fast mouse does not automatically make you good or God. It's more like your mouse should be fast enough, but not faster. You have to do the rest. How fast is warp 10? Depends on the Star Trek canon you are reading I suppose. http://en.memory-alpha.org/wiki/Warp_10#Warp_ten_and_aboveUh, most recently it was infinite speed, occupying all space in the universe at the same time. Wow that's pretty fast. So I am still trying to get used to no acceleration mouse. but I think I'll get used to it soon Bet its nice being able to click on more than a quarter of your screens pixels for a change.
No, the amounts of angle snapping cannot be influenced at all. Not by the polling rate, not by the OS when it updates the registers. This is where Bullveyrs post is fairly crucial. We have to make the distinction between the path correction algorithm, and the "perceived" correction of a low polling rate.
The sensor sends the movement data (delta x/y) to the MCU, usually more than 1000/s, and the MCU then sends it to the PC.
So the polling rate doesn't affect the communication between the sensor and the MCU.
Lets say your mouse runs at 125Hz and the current data send to the PC consists of delta x = 40 and delta y = 2.
With 1000Hz this data package would be split in 8 packages which could look like that.
x = 3 | x = 3 | x = 4 | x = 4 | x = 4 | x = 4 | x = 5 | x = 5
y = 0 | y = -1| y = -1 | y = -1| y = 0 | y = 1 | y = 2 | y = 2
In both cases overall x/y =40/2 and your cursor/crosshair will end up at the same position allthough in Paint with 1000Hz the line would look less straight, with a bump down and a bigger bump up, than the line with 125Hz.
Because angle snapping is done on a sensor level the sensor still reads x/y = 40/2 but may send 40/0 to the MCU (actually more like 20 times x/y = 2/0) and you would draw a perfect straight line.
So with angle snapping you would end up 2 pixels below the the point your hand movement represents.
A higher polling rate is smoother and more "precise" and looks less corrected, at least in Paint, but over a longer time period (long in the sense of 8ms) it doesn't affect the position of your crosshair and doesn't affect angle snapping.
In this example, both lines that are drawn start and end in the same spot, but they look slightly different because the host was able to update the registers more quickly in the line that was drawn at 1000Hz. This is not angle snapping (the correction algorithm), nor is it actually "correcting" a path (though this is how it is perceived in paint), this is a limitation of the current USB spec (I am trying to verify some things about USB 3.0 that "could" completely change the landscape, if I read them correctly) where the registers are updated less frequently, and therefore the path is less accurate in showing your actual movement.
Obviously this strongly argues that a low polling rate is undesirable because our cursor movements will look less like our actual movements at lower Hz, but our movements are less precise because the delta x/y values are being updated from the host at slower rates, not because they are "actualy" being corrected, which is what angle snapping does.
From what I understand about the USB 2.0 spec, the OS will never be forced to draw lines of any kind. When the host updates the register it will move the mouse by that amount, as in steps and not a fluid line, but we have to remember that these updates are taking place at speeds that we cannot humanly detect where we would actually notice this blocky, or choppy behavior. As an example, by the time that a user reaches the 500Hz polling threshold, we are already traveling at ~1.5in/s, which is absurdly slow.
|
You pretty much repeated what I just said >.>
I must have been unclear somewhere.
|
On February 24 2012 09:57 Medrea wrote:
You pretty much repeated what I just said >.>
I must have been unclear somewhere.
btw turning off acceleration, I had to tone down the dpi. I could no longer function @ 5600 dpi, now riding at like 3600.
|
On February 24 2012 10:07 Blasterion wrote:Show nested quote +On February 24 2012 09:57 Medrea wrote:
You pretty much repeated what I just said >.>
I must have been unclear somewhere. btw turning off acceleration, I had to tone down the dpi. I could no longer function @ 5600 dpi, now riding at like 3600.
Yeah so it is pretty much exactly as I thought.
Also make sure that the sensitivity option in SC2 is off and that windows sensitivity is still 6/11.
If that is too slow, bring the DPI back up. If you havent done this already I mean.
|
On February 24 2012 09:57 Medrea wrote:
You pretty much repeated what I just said >.>
I must have been unclear somewhere.
Yea I think I did. Some of the wording at the top just seemed a little murky and I wanted to make sure that everything was crystal clear. :D
|
On February 24 2012 10:12 wo1fwood wrote:Show nested quote +On February 24 2012 09:57 Medrea wrote:
You pretty much repeated what I just said >.>
I must have been unclear somewhere. Yea I think I did. Some of the wording at the top just seemed a little murky and I wanted to make sure that everything was crystal clear. :D
Heh. That's probably my bad. I have a better example of what I was talking about that I will probably post in spoilers. Short story is. Over prediction (path correction) is bad for obvious reasons. No path correction (prediction) is also bad because you need to iron out tiny flaws in the surface you are mousing on. While its romantic to think that a mouse with no correction at all provides perfect results, we just aren't there yet. You'd need a pure titanium surface to provide perfect tracking.
The key players we need to watch out for in mice:
Acceleration
Lift-off problems
Tracking issues
Maximum speed (for about the first 1.5 m/s after that its pretty unrealistic speeds)
Bugs
Prediction is hard to qualify. Perception often leads to alteration of course. Also it is sort of funny. But its a bit odd we are hard on a mice that draws straight lines, when we are in fact trying to draw straight lines on purpose. >.> The irony is not lost on me.
|
On February 24 2012 09:44 wo1fwood wrote:
... I am trying to verify some things about USB 3.0 that "could" completely change the landscape, if I read them correctly ...
Sounds interesting. Please keep us up to date!
|
btw @ Medrea I started to train for No accel 5600 dpi again to see if I can pull it off.
|
United Kingdom20326 Posts
On February 25 2012 02:55 Blasterion wrote:
btw @ Medrea I started to train for No accel 5600 dpi again to see if I can pull it off.
Good luck with that
*goes back to 450dpi with 0.5 on the fly sens for splits and blink engages*
|
On February 25 2012 04:50 Cyro wrote:Show nested quote +On February 25 2012 02:55 Blasterion wrote:
btw @ Medrea I started to train for No accel 5600 dpi again to see if I can pull it off. Good luck with that *goes back to 450dpi with 0.5 on the fly sens for splits and blink engages*
Damn 450 dpi You must have to use the entire mouse pad.
|
On February 25 2012 04:50 Cyro wrote:Show nested quote +On February 25 2012 02:55 Blasterion wrote:
btw @ Medrea I started to train for No accel 5600 dpi again to see if I can pull it off. Good luck with that
^^
On February 25 2012 04:50 Cyro wrote:
*goes back to 450dpi with 0.5 on the fly sens for splits and blink engages*
Seriously? How's your experience with this setup so far?
|
For folk (like myself) that were interested in trying out the CM storm spawn, Amazon has a pretty good deal for it. It's currently listed at:
$60 - $30 - $10 rebate = $20.
|
On February 27 2012 11:37 Kambing wrote:
For folk (like myself) that were interested in trying out the CM storm spawn, Amazon has a pretty good deal for it. It's currently listed at:
$60 - $30 - $10 rebate = $20.
Wow, anyone who's a high sens claw/fingertip gripper should definitely look into this if they're looking for a new mouse.
|
|
|
|
|
|
EPT![[image loading]](http://dl.dropbox.com/u/34863088/Polling.jpg)
