Overclock.net › Forums › Components › Mice › Mouse from Steelseries - Rival 100
New Posts  All Forums:Forum Nav:

Mouse from Steelseries - Rival 100 - Page 53

post #521 of 1498
Well the G300 had a retarded sensor position. So that would leave the G100 then?
post #522 of 1498
well it's just a guess given that logitech usually does (the technical) things sanely.
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 #523 of 1498
Quote:
Originally Posted by woll3 View Post

......

Quote:
Originally Posted by qsxcv View Post

30 is a unitless number.
you dont pull out a unit of 1/inches out of thin air

the number of pixels in an array does not affect resolution at all. as in you can build a sensor that's 2x the physical size of everything in the 3090 and it's base cpi would be halved

actual cpi depends on pixel density*lens magnification*scaling.

pixel number only affects it indirectly, as with more pixels you scale more

Quote:
Originally Posted by uaokkkkkkkk View Post

So apparently going over 500dpi with a 3050 based mouse will make it explode.

Actually is there any consensus on what has been the best 3050 implementation so far?

http://media.digikey.com/pdf/Data%20Sheets/Avago%20PDFs/ADNS-3050.pdf
Quote:
Note: As the sensor resolution increases, slight performance degradation on certain surfaces may be observed. For higher than 500 dpi setting, use 12-bit motion reporting to achieve the maximum speed.

Is that the same for the ADNS-3090?

You won't get the same performance as a 3090 at the same CPI. That is my point.

From what I heard, the 3050 is problematic past 500 CPI.
Edited by popups - 9/30/15 at 7:00pm
post #524 of 1498
what the datasheet is means is that beyond 500dpi, the perfect control speed of the sensor is likely limited by the 8bit motion registers since counts are limited to -128 to +127 each time the mcu reads motion (aka this issue: http://www.esreality.com/?a=longpost&id=1265679&page=4, see how the response plateaus for 125hz). in the 12bit mode the limitation is -2048 to +2047, and the sensor itself will malfunction before it exceeds the range of the the motion registers.

3090 only has an 8bit mode, so it always suffers from this issue. e.g. do a fast swipe with a 3090 mouse set to 4000dpi and see what happens.

30x30=900 does not mean 3090's base cpi is 900
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 #525 of 1498
Quote:
Originally Posted by popups View Post

From what I heard,

There found the problem.
post #526 of 1498
well i think it's pretty plausible that there'd be visible jitter at 1000dpi for 3050. 750 may be tolerable though. i assume 500 is the base cpi?
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 #527 of 1498
using some of the sensor images that qlettersv got from the 3366, I once did correlation by hand in an image editor and got a difference in correlation of 16384 across the entire image for moving the whole thing by a single pixel (I forgot if this was diagonal or not)

divide that in 2 because the 3366 was spitting out 7-bit data for some reason (instead of 8-bit) and we have 8192 levels for a single pixel of movement

Sensor resolution would affect DPI but the numbers that the sensor gets out of the correlation function are so absurdly high in the first place that *every* sensor scales down the values in some way before sending them to the MCU or whatever.

supposing that the 3050 has the same bitdepth and coverage, the scaling difference for them for the same dpi should be around 2.1x (40 vs 19 pixels wide)

someone on the logitech forums once wrote "Pixart 3366 sensor is native at ALL DPI settings. However its only jitter free up to ~3600 DPI, so don't go above that. Once you reach 4000+ DPI, you start getting jitter."

3600/2.1 is ~1700

Supposing that the 3050 is significantly inferior in other areas relating to jitter, that 1700 can easily be cut in half (for example, if the illumination is poor, or the shutter time is too short, or the sensor just picks up more outside noise)

i should really write a correlation simulation software as a learning experience....
Edited by wareya - 9/30/15 at 8:39pm
post #528 of 1498
16384=2^14...? if you have it still could you show your work/equations? there's quite a few ways to go about doing it.
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 #529 of 1498
Quote:
Originally Posted by qsxcv View Post

16384=2^14...? if you have it still could you show your work/equations? there's quite a few ways to go about doing it.

there are 1600 pixels capable of 256(128 actually for some reason) levels of brightness so it's not that absurd

I think I used this image http://i.imgur.com/9kBcSbg.png

The first step of correlation is an unsigned difference filter and the second is accumulating the resulting difference pixels together, but I was lazy so I just blurred the difference image altogether and multiplied by the number of pixels. should only give a quantization error of half the number of pixels as a number of correlation levels though

because different surfaces have different levels of noise and contrast, if they just outputed certain things verbatim, it would make surfaces with less contrast output less decorrelation from the "center" pixel. So for grayscale sensors they definitely normalize the correlation array in some way before outputting the distance of the apparent decorrelation peak from the center pixel into their scaling buffer (otherwise they would need to use interpolation or line math to find the apparent peak, which is an approximation and introduces more artefacts). And if they don't use a scaling buffer but use a series of deadzone checks and don't throw away the memory image until a single pixel of movement is achieved instead, then they might not need to normalize the correlation array, but I'm not sure, they might need to do some scaling on a value that's both tied to surface properties and necessary to avoid being locked at integer pixels of movement for a given speed (but I have a hunch that this type of sample handling responds poorly for image rotations or high noise floors -- but who knows! it might be the other way around, or totally not important at all)

I'm quite tired so it's possible that this post has a logic mistake in it.
Edited by wareya - 9/30/15 at 9:19pm
post #530 of 1498
uh isn't the procedure like
<
below is (reasonable) speculation
>
shift by (rounded) previous frame's motion counts
calculate correlation
shift by a pixel
calculate correlation
shift again
etc...

and then you'll have (like discussed previously) a 3x3 or whatever matrix of correlation values. the question is how to combine the values in the matrix to figure out where the "peak" is with sub-pixel accuracy.
e.g. if the matrix looks like
0.0 0.0 0.0
0.0 0.0 1.0
0.0 0.0 0.0
then obviously the peak is at (1,0)

but if it looks (more realistically) like
0.0 0.2 0.4
0.1 0.4 0.8
0.1 0.0 0.1
then the peak is probably actually something around (0.7, 0.7)
and then you add (0.7,0.7) to the previous frame's motion counts to get the current frame's counts.

the question is how to calculate the (0.7, 0.7) or whatever. i strongly suspect that not doing this correctly is the reason for the variance issues in the 9500/9800/3310/3988 family

on an undetailed surface or when the lens is unfocused, the peak would be blurrier so that would complicate things. it wouldn't be surprising if the dsp applied a high-pass or edge detection filter to the images before the correlation process to minimize the effects of things like illumination uniformity.
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
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Mice
Overclock.net › Forums › Components › Mice › Mouse from Steelseries - Rival 100