Overclock.net › Forums › Components › Mice › patent detailing how mlt04 works
New Posts  All Forums:Forum Nav:

patent detailing how mlt04 works - Page 3

post #21 of 28
so why does my WMO have lower tracking speed than my 1.1 frown.gif
post #22 of 28
Quote:
Originally Posted by kyotkyotkyot View Post

Back into pc gaming after a long break. Keep reading about the mlt04 preferance and I'm wondering if the MS blue optical is included with these famed mice or was it the ugly duckling of the bunch? I have one from back in the day, but only got it because friends said it was good at the time.

The Intellimouse Explorer 4 was a bad mouse and you cannot overclock the polling rate.
post #23 of 28
Blue optical is supporting mlt04, yes.

IE4.0 is a totally different mouse.
post #24 of 28
Quote:
Originally Posted by Brightmist View Post

Blue optical is supporting mlt04, yes.

IE4.0 is a totally different mouse.

The Intellimouse Explorer 4 isn't a MLT mouse?
post #25 of 28
Not sure about that, I used it for a brief amount of time 10+ years ago but it might not be.

At least it isn't listed as an MLT04 mouse in the sticky Mouse Sensor thread.
post #26 of 28
Quote:
Originally Posted by kyotkyotkyot View Post

Back into pc gaming after a long break. Keep reading about the mlt04 preferance and I'm wondering if the MS blue optical is included with these famed mice or was it the ugly duckling of the bunch? I have one from back in the day, but only got it because friends said it was good at the time.
Well, it very much is the ugly duckling of the bunch. But it does also feature the MLT04 sensor. wink.gif
Great mouse really, I have recently gone back to it myself because I like the shape more than that of the WMO.

OT: As for the patent documentation. Chances are if you couldn't extract anything more interesting from those than the framerate thing then I surely can't either.
post #27 of 28
Thread Starter 
well i still need to figure out how exactly the equations in pg19 are derived
seems to be some typos there...
Quote:
The auto correlation component for the case Where

both a and b equal 0 is alWays equal to one,
given their formula (sum of absolute differences), AC5 should be equal to 0

and quadrant 0 equations should end with (C4 - C2) not (C4 - C1)

edit:
yea ok i cant get these equations to spit out any numbers which make sense.......
even in the trivial case where the cross correlation and auto correlation are identical, the numbers dont come out right
Edited by qsxcv - 11/8/15 at 7:13pm
main
(15 items)
 
old
(14 items)
 
 
CPUMotherboardGraphicsRAM
4770k maximus vii impact nvidia gtx 970 crucial ballistix tactical 16gb 
Hard DriveCoolingOSMonitor
crucial mx100 noctua nh-c14 windows 7 ultimate sony cpd-g520 
KeyboardPowerCaseMouse
kbp v80 matias quiet silverstone sx500-lg ncase m1 v3 logitech g100s with mcu replaced by teensy2.0 
Mouse PadAudioAudio
allsop raindrop xl chord mojo hifiman re-600 
CPUMotherboardGraphicsRAM
i7 920 evga x58 sli le galaxy gtx 460 crucial something 3x1gb 
Hard DriveCoolingOSMonitor
intel 330 180gb scythe kotetsu windows 8.1 pro sony cpd-g520 
KeyboardPowerCaseMouse
logitech k120 silverstone st75f-gs nxzt h440 evga torq x5 
Mouse PadAudio
allsop raindrop mobo 
  hide details  
Reply
main
(15 items)
 
old
(14 items)
 
 
CPUMotherboardGraphicsRAM
4770k maximus vii impact nvidia gtx 970 crucial ballistix tactical 16gb 
Hard DriveCoolingOSMonitor
crucial mx100 noctua nh-c14 windows 7 ultimate sony cpd-g520 
KeyboardPowerCaseMouse
kbp v80 matias quiet silverstone sx500-lg ncase m1 v3 logitech g100s with mcu replaced by teensy2.0 
Mouse PadAudioAudio
allsop raindrop xl chord mojo hifiman re-600 
CPUMotherboardGraphicsRAM
i7 920 evga x58 sli le galaxy gtx 460 crucial something 3x1gb 
Hard DriveCoolingOSMonitor
intel 330 180gb scythe kotetsu windows 8.1 pro sony cpd-g520 
KeyboardPowerCaseMouse
logitech k120 silverstone st75f-gs nxzt h440 evga torq x5 
Mouse PadAudio
allsop raindrop mobo 
  hide details  
Reply
post #28 of 28
Quote:
Originally Posted by popups View Post

At slow speeds the MLT method feels more accurate than the PixAart/Avago method. At higher speeds the Pixart/Avago method seems fine. Maybe the MLT method would limit the malfunction speed compared to the PixArt/Avago method and maybe the PixArt/Avago method will allow better surface compatibility. I guess the method used for the 3366 isn't implemented in the AM010/3320.

In terms of (my) perception, I like how the MLT feels at all speeds, I can't say the same for the 3366.
Quote:
Originally Posted by qsxcv View Post

at very low speeds (1 pixel at a time, drawing a figure 8 or something in paint) both 3366 and am010 feel much better than mlt04, i believe due to the deadzoning thing
once you get up to typical low speeds, like controlling recoil in cs, imo mlt04 feels a bit more "natural" to use, if that makes any sense.

personally i don't notice the framerate transitions in 3366
Quote:
Originally Posted by popups View Post

Maybe it's the rest mode transition that make it feel that way?

I skimmed over the MLT document. It says something about only using a portion of the frame and a single buffer. Then something about zones that determine when to add a new frame into the memory. Does this make the sensor feel more consistent than using the full frame and low FPS, then transitioning to a MLT type method?


Quote:
Quote:
"In an embodiment, the feature detector 222 only selects a fixed size portion of a frame of image data. As an example, if the captured frames of image data of the navigation surface 106 are 19×19 pixel frames of image data, the feature detector 222 may select only 7×7 pixel portions of the frames of image data. In other embodiments, the feature detector 222 may select any appropriate size portion of a frame of image data from a range of sizes. As an example, if the captured frames of image data of the navigation surface 106 are 19×19 pixel frames of image data, the feature detector 222 may select 6×6, 7×7, 8×8 or 9×9 pixel regions of the frames of image data. If the feature detector 222 determines that a frame of image data does not have such a portion with prominent trackable features, then the processing unit 220 operates in the first mode of operation. However, if the feature detector 222 determines that a frame of image data does have such a portion with prominent trackable features, then the processing unit 220 operates in the second mode of operation, if enabled to do so.

The cross-correlator 224 operates to cross-correlate either full frames of image data of the navigation surface 106 or portions of the frames of image data. In the first mode of operation, i.e., when the feature detector 222 determines that a particular frame of image data does not have a predefined portion with prominent trackable features, the cross-correlator 224 cross-correlates that particular frame of image data with a previously captured frame of image data using the entire or full frames of image data. That is, the current frame of image data is cross-correlated with the previously captured frame of image data using every pixel of the particular frame of image data and every pixel of the previously captured frame of image data. In the second mode of operation, i.e., when the feature detector 222 determines that a particular frame of image data does have a predefined portion with prominent trackable features, the cross-correlator 224 cross-correlates the selected portion of that frame of image data with a corresponding portion of a previously captured frame of image data, rather than using the entire or full frames of image data. A corresponding portion of a frame of image data is the portion of the frame of image data having the same size as the selected portion with prominent trackable features and positioned in the same location in the frame of image data as the selected portion in the particular frame of image data. The results of the cross-correlation are then used to produce the output directional delta X and Y displacement values by the processing unit 220.

In the second mode of operation, the number of calculations needed to perform cross-correlation for displacement estimation will be greatly reduced. This is because the number of multiplications needed for a full frame cross-correlation is n to the fourth power (n4), where n is the number of pixels along a single side of square frames of image data that are being cross-correlated. Using the example of 19×19 pixel frames of image data and 7×7 pixel portions of the frames of image data, the number of multiplications required for cross-correlation is reduced from 130, 321 to 2,401, which is approximately an improvement by a factor of 54. In addition, since weak features in frames of image data do not significantly contribute to tracking using cross-correlation, the reduction in the number of multiplications required for cross-correlation can be achieved without significant tracking performance degradation because the weak features are most likely outside of the selected portion with prominent trackable features. The reduction in the number of multiplications required for cross-correlation can translate into reduced power consumption for the optical navigation system 100, assuming that the frame rate of the system is unchanged. Thus, the second mode of operation can be used to reduce the power consumption of the optical navigation system 100.

Instead of or in addition to reducing the power consumption, the second mode of operation can also be used to increase the tracking performance of the optical navigation system 100 during high speed usage. Usually, the number of multiplications required for cross-correlation is the bottleneck for frame rate increases. Thus, a reduction in the number of multiplications required for cross-correlation will directly enable the optical navigation system 100 to increase the frame rate, which will result in better tracking performance during higher speed usage." -Avago Patent

http://utmalesoldiers.blogspot.com/2015/11/rivaltrackinglag.html

I guess enabling the second mode of operation on the Avago/PixArt sensors makes it perform "worse" than a MLT in terms of tracking your motions, but gives you a higher malfunction speed if you increase the frame rate as well. Maybe using the second mode on the 3310 is the reason for reports of malfunctions on QCK mouse pads.

The first mode of operation could make the mouse similar or "better" than a MLT mouse depending on various factors. It would take more computation power, which could limit frame rates, thus the feeling of smoothness. I wonder how high you can push the frame rate with the first mode of operation and what will the malfunction speed be reduced to.

If companies use mode 2 it would be better to use the highest frame rate as possible. However, the 3310 has a limited frame rate. So, maybe using mode 1 is the better option for the 3310. Maybe mode 2 is okay on the 3366 and 3988 if the frame rate is 12000 at all times.
Edited by popups - 11/9/15 at 11:26am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Mice
Overclock.net › Forums › Components › Mice › patent detailing how mlt04 works