Unfortunately for you, your argument via verbosity doesn't scare me.
>1. What do you mean by “any analog”?
>2. You want to say, that ball mice have same limitations as ICS?
Any "analog -> digital conversion". Basic english.
>3. By quantization you mean CPI?
You *CANNOT* convert analog data (motion) into digital format without introducting quantization, and quantization is *THE* limitation of *ALL* movement response systems. If you don't understand basic signal processing, you have no right to teach *ANYBODY* about it.
>4. What noise are you talking about?
>11. Which flaws exactly do you mean?
I did say all of them.
>12. How can you compensate them?
I already explained this. Your argument is "How?" after I've alread explained myself several times.
>13. What do you mean by “easy”?
Even an intern could do it.
>14. With which errors it will be compensated?
This is ungrammatical.
>15. Will it lead to different flaws?
>16. Why do you think it’s possible?
Because it's actively done in the real world.
>17. What things exactly?
The things you claim keep it from working optimally.
>18. What do you mean by “perfectly fine”?
Are you seriously asking this question?
>19. Why do you think that these things don’t massacre it?
Because we know that they don't from a combination of "how DIC mathematically works" and "real world mice work".
>20. Any proof on 7 bpp?
>21. What is this based on?
qsxcv's debug images from the sensor are 36x36 with 7bpp. That is the amount of information that they contain. You can easily verify this yourself with any image editor capable of displaying color hex codes.
>22. What do you want to say by this?
This is not a question.
>23. How did you get this number?
>24. What do you want to say by this?
I ran an unsigned difference filter on shifted 32x32 matrixes from qsxcv's debug images and accumulated the result of the difference filter on each pixel. This resulted in a difference of 16384 raster levels. This is exactly how DIC operates and how it detects movement. A very small movement results in this large of a number. This is the level of precision that optical mice work under.
>25. More than enough for what?
100k DPI, for one thing.
>26. What is enough?
I think that 64 levels per sensor count would be enough in an ideal world, but multiplying that number by a couple hundred seems like a good idea Just In Case (tm).
>27. Is it grounded on anything?
The competent experience of literally every single person in the world that's using a good optical mouse.
>28. If accuracy can’t be more than 100%, how can it be too much?
Accuracy cannot ever be 100%, and we are far, far past the point of diminishing returns. If you want an LED that's so bright that it burns your eye, I'm sure that we could get 1% better than existing DIC implementations, but it's not worth it.
>29. Do you mean hardware resolution or interpolation?
The hardware resolution is 36x36, but this has no relation to the
>30. What do you mean by high?
More than 800
>31. How DPI is related with tracking accuracy, which we are talking about?
High DPI is more accurate, or do you want to claim that we shouldn't push the limits like you just said we should?
>32. On what CPI?
>33. What do you mean by negligible?
>34. Do you know it’s exact influence?
It's low enough that we can't detect it, ever.
>35. How do you know, that sensor doesn’t hide jitter by smoothing and/or any other kind of (post)processing/path correction?
Because it doesn't need to. Actually, the FinalMouse does, but it's only necessary at 5000 DPI, and it doesn't reduce drift, because the smoothing implementation that the FinalMouse has is a simple infinite response filter that doesn't have any properties of reducing random walk.
>36. How can technology have no flaws and limitations?
You said "a lot", not "any".
>37. Do you want to say, that ICS has no limitations?
No; it has the limitations of quantization and sensor noise, just like literally every single other possible motion sensing method.
>38. What is this statement grounded on?
>39. Why did you except Niquist frequency from ICS limitations?
ICS operate on raster images. Raster images are mathematically equivalent to PCM waveforms. PCM waveforms have nyquist frequency.
>40. Why did you except pixel array size and framerate from ICS limitations?
Pixel array size and framerate are obvious forms of quantization. Having two doesn't suddenly make it worse.
>41. How do you know, that it’s possible?
Every new "next gen" sensor increases these things over the previous new sensor.
>42. How could it lower influence of limitations?
Increasing these reduce quantization.
>43. Why don’t you think it’s possible?
>44. How exactly will you do it?
All optical mice do this already. It's a basic element of DIC. It's mentioned in that interpolation patent that qsxcv posted in the other thread, if you want to verify it for yourself.
>45. Which measurement errors will have this?
None. It mathematically has zero drawbacks. You only need to increase the margins of the sensor picture array without increasing the size of the correlation array, then you get it for free.
>46. What is this statement based on?
>47. How can prediction add no artifacts?
It's not "prediction" in the sense you use it.
>48. Any proof?
1+1=2, pigs don't have wings, and moving over the frame of reference in a picture doesn't make it look different. It's essentially an optimization so that they can perform fewer computations and get the same result.
>49. What do you mean by theoretical problems?
Problems that don't exist on the real world.
>50. How can theoretical problems disappear?
>51. Which theoretical problems exactly do you mean?
You just asked a question using a term that you asked for a definition for and before receiving the definition. Invalid.
>52. How is FPS related with sensor quality?
Higher framerates reduce quantization error.
>53. How you think brightness of LED can eliminate fundamental flaws?
>54. Which limitations can it possibly eliminate?
Brighter LEDs allow for higher framerates and lower exposures without increasing sensor optical noise.
>55. Aren’t they popular because of invisibility?
My DA's IR LED emitted visible light in addition to its IR light.
>56. What is practical limit of DPI?
Right now? Looks like 10000.
>57. Why do you think, that jitter happens only on some CPI settings?
Jitter has many causes, but the cause of jitter that I'm referring to here is cheap interpolation of the correlation accumulation array causing jitter.
>58. Doesn’t jitter happen on level of sensor measurement?
No. Quantization happens on the level of sensor measurement. Quantization actually prevents jitter.
>59. What do you want to say by this?
It means what it says. If your grasp of English is too poor to understand it, you shouldn't be arguing with me.
>60. What do you mean by “low DPI”?
>61. What is this statement based on?
History of mouse marketing in America between 2005 and 2010.
>62. Do you know any ICS without such algorithms?
>63. Is there any proof, that there are ICS without such algorithms?
Feel free to dump and reverse engineer the MLT04's firmware to prove the assertion that the MLT04 implements smoothing, angle sanpping, or path correction.
>64. How can real world disagree with its working principle?
It's not a fundamental principle, that should be clear given that the real world disagrees with it.
>65. Why do you deny existence of fundamental limitations?
>66. What is this statement based on?
You are chopping off a sentence.
>67. What is this statement based on?
3366 has higher stats than the 3310, which has higher stats than the MLT04. These stats are being pushed forward. MLT04 was good enough 15 years ago, and yet we're still improving these statistics.
>68. You want to say, that ICS engineers are stupid enough to make mice for “office users”, but at the same time same engineers make sensors for “gamers”?
Engineers don't make management decisions.
>69. Why do you except, that more expensive mice don’t have tons of smoothing, angle snapping and path correction?
I never said that there aren't mice with these things. I said that there are mice without these things. Your understanding of english negatives needs work.
>70. Do you have any proof?
Do you? No. You don't. You're the one enforcing spectacular claims here.