PMW3366 why do you exist? - Overclock.net - An Overclocking Community

View Poll Results: Which image correlation sensor you prefer?
STMicroelectronics OS MLT 04 38 12.14%
Avago A3050 7 2.24%
Avago A3060 2 0.64%
Avago A3090 14 4.47%
Avago A6010 2 0.64%
Avago A9500 3 0.96%
Avago A9800 2 0.64%
Avago S3888 2 0.64%
Avago S3988 18 5.75%
Avago AM010 10 3.19%
Pixart PMW3366 167 53.35%
Pixart PMW3310 48 15.34%
Pixart PAW3305 0 0%
Voters: 313. You may not vote on this poll

Forum Jump: 

PMW3366 why do you exist?

Reply
 
Thread Tools
post #1 of 125 (permalink) Old 07-18-2015, 09:22 AM - Thread Starter
New to Overclock.net
 
Above8's Avatar
 
Join Date: Dec 2013
Location: Russia, Moscow
Posts: 424
Rep: 40 (Unique: 34)
I’ve heard a lot of recommendations and positive feedback about PMW3366 sensor, some people even call it “the best sensor in the world”, so I got curious and I bought G303 to compare it with other sensors.

Fortunately, I have some experience in how to compare different sensors. Before I was using not only ICS (image correlation sensors), but other technologies as well. So I have an idea about how sensor should behave. This is the list of devices, I used before. Spoiler (Click to show)
A4Tech Bloody V2 (Pixart 3305DK @1000Hz)
A4tech OP-620D (Pixart PAW3504DLY-TEL2)
A4tech OP-720 (Pixart PAN3512DK-TJZA)
A4Tech X-710MK (Avago A3060 @1000Hz)
A4Tech XL-750BK (Avago ADNS-6010 @1000Hz)
A4Tech XL-755BK (Avago ADNS-6010 @1000Hz)
A4Tech XL-750MK (Avago A9500 @1000Hz)
CMStorm Inferno (Philips PLN2031 @1000Hz)
CMStorm Recon (Avago ADNS-3090 @1000Hz)
Cyborg R. A. T. 5 (Philips PLN2032 @1000Hz)
Genius NetScroll 200 Laser (Avago A7700 @125Hz)
Genius XScroll Optical
i-Rocks IM3-WE (Avago ADNS-3090 @1000Hz)
Logitech G100s (Avago AM010 N1301T @500Hz)
Logitech G9X (Avago A9500 @1000Hz)
Logitech LS1 (Avago S7550 @125Hz)
MadCatz R. A. T. 5 (Philips PLN2032 @1000Hz)
MadCatz R. A. T. 7 (Philips PLN2033 @1000Hz)
MadCatz R. A. T. TE (Philips PLN2034 @1000Hz)
MadCatz M. M. O. TE (Philips PLN2034 @1000Hz)
Micosoft Intellimouse 1.1 (ST Microelectronics OSMLT 04 @125Hz)
Mitsumi ECM-S6802 (ball @200 Hz)
Ozone Radon 5K (Philips PLN2032 @1000Hz)
Philips SPM7800 (Avago A7530)
Philips SPM9800 (Philips PLN2022 @88Hz)
Razer Abyssus 3.5G (Avago S3888 @1000Hz)
Razer Imperator 3.5G (Philips PLN2032 @1000Hz)
Razer Lachesis 3G (Philips PLN2031 @1000Hz)
Razer Lachesis 3.5G (Philips PLN2032 @1000Hz)
Razer Naga 2014 (Philips PLN2034 @1000Hz)
Razer Spectre (Philips PLN2032 @1000Hz)
Razer Taipan (Avago S9818 @1000Hz)
Roccat Kova [+] (Pixart PAW3305DK @1000Hz)
SteelSeries Ikari Laser (CYONS1001U @1000Hz)
SteelSeries Kinzu v2 (Pixart 3305DK @1000Hz)
SteelSeries Sensei MLG (Avago A9800 @1000Hz)
SteelSeries Sensei RAW (Avago A9500 @1000Hz)
Verbatim Rapier v2
Wacom Intuos 5 L Pen & Touch
Wacom Mouse KC-100

You may know, what I think about comparison of ICS with other technologies (if you don’t, you should read this), so here I’ll focus on comparing PMW3366 with other ICS.

This is only empirical analysis, so don’t expect to see formulas and graphs. I compared different sensors, their algorithms and peculiarity. I tried my best to find best tracking surface for each sensor, to get used to each sensor, to adapt to its peculiarities and to change my play style depending on sensor.

During this test I compared sensors A3060, A3090, A6010, A9500, AM010, MLT04, 3305, PMW3366 and S3888. I’ll start from worst sensor and will move to best sensor for overall aiming ability in average games.

Avago A9500 (A4Tech XL-750MK)
I have no idea what was it made for. It’s worst ICS I’ve seen so far. But not because of so called “acceleration”. First of all it’s laser based, which makes surface requirements even more strict, than in case of LED based sensors. There is a lot of noise and huge amount of smoothing and path correction. It’s not better than cheap office mouse.

Avago AM010 (Logitech G100s)
Cursor movements are separated from hand movements. It’s usual for Logitech sensors. Smooth mouse movement results cursor speed jumps. I have a feeling that sensor collects data for like half a second and then makes cursor run all this distance very fast. Maybe it makes the end result more accurate, but in price of inability to correct path during movement. Taking into account, that it’s main method of aiming, it makes this sensor unusable, especially for leading moving targets or compensating recoil in games like CS, COD or BF. Also worth mentioning noticeable amount of smoothing.

Avago A3060 (A4Tech X-710MK)
Minimal amount of smoothing I’ve seen in ICS so far. Rough movements seem to be close to hand movements. Problems here are strong angle snapping and inability to perform small movement (pixel walk). All this makes sensor too rough and inaccurate, especially for games.

Avago A3090 (i-Rocks IM3-WE)
Significant amount of smoothing and path correction which is really hard to compensate.

Pixart 3305DK (Roccat Kova [+])
It also has angle snapping and pixel walk, but a little bit less, than Avago A3060. It has more smoothing than A3060, but overall it’s a little bit more controllable.

Pixart PMW3366 (Logitech G303)
Logitech’s weird algorithms again. It seems to be successor for AM010. They still use their trick to increase “numbers on box” like CPI and max tracking speed by sacrificing real world performance and controllability. I can’t notice “smoothing” here in it’s classic form. It has different processing. Sensor ignores small movements. It accumulates them to certain threshold. After this threshold sensor starts reporting them. Seems like it has some kind of “death zone” of small movements. Like some kind of angle snapping, but not related to axis. Small deviation from large movement will always be rejected. Big fast movements may be slightly more accurate, that with non-Logitech sensors. But trajectory correction, leading moving targets and compensating recoil is still nearly impossible. I can’t imagine situation, where I can prefer this sensor, to next ones.

Avago A6010 (A4Tech XL-755BK)
There is noticeable smoothing, but path correction is not significant. Sensor feels more “raw”. Main problem is in laser. On some surfaces it tracks small movements somewhat predictable, but has terrible max tracking speed. On other surfaces it has tolerable max tracking speed but lots of noise. I found no surface which it will work “well” on. At least using it on Manticor, it’s possible to correct errors in process, to lead moving targets and to compensate weapon recoil in games.

Avago S3888 (Razer Abyssus 3.5G)
Lots of smoothing and(!) noise. The thing is smoothing is more predictable and correctable somehow. Small slow correcting movements work just fine. Leading moving targets is not a big of a problem, as well as compensating recoil. But big fast movements are not the strongest part of sensor behavior. If I’ll need to choose between Logitech’s flick shots without ability to correct them and Razer’s fast corrections, I’ll always choose corrections, because Logitech’s flick shots are still not reliable. It’s better than Pixart PMW3366 overall.

STMicroelectronics OS MLT 04 (Microsoft Intellimouse Optical 1.1)
Yes, max speed isn’t great, I reach its speed limit even at high sensitivity. Yes, accuracy isn’t perfect. Yes, it has smoothing. But this sensor is predictable and controllable. Results, which I can get with this sensor, are much higher than anything I can achieve with any other ICS. It’s good for both: flick shots and corrections. It’s the most “raw” ICS I’ve ever used. I seriously don’t have a clue what were all other ICS made for and why they all are so much worse. Any other ICS is not even close.

So if you don’t want to use anything but ICS, my advice will always be MLT 04.
Above8 is offline  
Sponsored Links
Advertisement
 
post #2 of 125 (permalink) Old 07-18-2015, 10:31 AM
Mehehe!
 
CorruptBE's Avatar
 
Join Date: Dec 2011
Location: Belgium
Posts: 2,743
Rep: 86 (Unique: 50)
One of the issues with this is that your basing things of 1 variation of mice using a specific sensor.

Take all of the 3090 variations for instance. Some I could use, some I would literally toss aside within 30 minutes (CM Recon comes to mind, weird delayed feeling on smaller movements). My 3090 Aurora (I somehow got a really good batch lol) on the other hand, I prefer its tracking (feeling wise) over the 2 3310 based mice I've used (FK1/Avior).

Also the Avago S3888 (Razer Abyssus 3.5G), Synapse?? I don't know how newer Razer mice perform with it, but the older gen sensors actually feel worse when it's installed.

CorruptBE is offline  
post #3 of 125 (permalink) Old 07-18-2015, 10:42 AM
Mice is bad civilization
 
Melan's Avatar
 
Join Date: Feb 2015
Location: 0,0
Posts: 3,243
Rep: 111 (Unique: 85)
3366 ignores small movements? lolwhat? This thing picks up everything like crazy. AM010 is bit more tame in this regard but if you can't track a moving target I might have bad news for you. It's not the sensor, it's you.
Melan is offline  
Sponsored Links
Advertisement
 
post #4 of 125 (permalink) Old 07-18-2015, 10:43 AM - Thread Starter
New to Overclock.net
 
Above8's Avatar
 
Join Date: Dec 2013
Location: Russia, Moscow
Posts: 424
Rep: 40 (Unique: 34)
Quote:
Originally Posted by CorruptBE View Post

One of the issues with this is that your basing things of 1 variation of mice using a specific sensor.

Take all of the 3090 variations for instance. Some I could use, some I would literally toss aside within 30 minutes (CM Recon comes to mind, weird delayed feeling on smaller movements). My 3090 Aurora (I somehow got a really good batch lol) on the other hand, I prefer its tracking (feeling wise) over the 2 3310 based mice I've used (FK1/Avior).
I don't have all the mice with all sensors, so I did my best to identify sensor particularities. I understand, that there are a lot of other factors like firmaware, shape, weight, mouse feet, cable, clicks and so on, but they all have different effect. By the way I used all mice with same feet.
Quote:
Originally Posted by CorruptBE View Post

Also the Avago S3888 (Razer Abyssus 3.5G), Synapse?? I don't know how newer Razer mice perform with it, but the older gen sensors actually feel worse when it's installed.
Abyssus 3.5 has no software. Only buttons on the bottom.
Quote:
Originally Posted by Melan View Post

3366 ignores small movements? lolwhat? This thing picks up everything like crazy. AM010 is bit more tame in this regard but if you can't track a moving target I might have bad news for you. It's not the sensor, it's you.
Which mice did you compared it with?
Above8 is offline  
post #5 of 125 (permalink) Old 07-18-2015, 10:53 AM
New to Overclock.net
 
Derp's Avatar
 
Join Date: Aug 2008
Posts: 8,763
Rep: 735 (Unique: 515)
The title doesn't fit the thread. You're comparing pretty much all relevant ICS sensors so why single out the 3366? Also I think it's really hard to be fair when comparing the sensors and voting in a poll like this for many reasons that might change your perception of how well a sensor performs.

  • Shape and weight. (3366 suffers here more than others IMO)
  • Glide.
  • Surface used.
  • Different implementations as CorruptBE described.
  • Not actually having experience with all of the sensors listed.
  • Not testing them at the same time and instead judging from memory which isn't very accurate.

I still think the MLT04 is the best sensor that I have ever used. But that's from inaccurate memory since setting them to 500-1000Hz hasn't been an option for a while now.

Derp is offline  
post #6 of 125 (permalink) Old 07-18-2015, 10:55 AM
Mice is bad civilization
 
Melan's Avatar
 
Join Date: Feb 2015
Location: 0,0
Posts: 3,243
Rep: 111 (Unique: 85)
I have G303, G402, FK1 and MX500 right now. I also borrowed G300S and an old IME3.0 from a friend (don't have those anymore... i mean mice). All of them track fine except only MX500 being a bit unstable @ 1kHz. Even G300S with it's awkward sensor position was good.

Btw, I only use 400 CPI and lowish sensitivity (around 65cm/360). Since G300S was a 3055 I used it's default 1000.
Melan is offline  
post #7 of 125 (permalink) Old 07-18-2015, 11:07 AM
New to Overclock.net
 
detto87's Avatar
 
Join Date: Sep 2011
Posts: 2,039
Rep: 59 (Unique: 46)
I hate such polls.
MOST people will read
"Which image correlation sensor you prefer?"
as
"Which image correlation sensor do you use?"
detto87 is offline  
post #8 of 125 (permalink) Old 07-18-2015, 11:10 AM - Thread Starter
New to Overclock.net
 
Above8's Avatar
 
Join Date: Dec 2013
Location: Russia, Moscow
Posts: 424
Rep: 40 (Unique: 34)
Quote:
Originally Posted by Derp View Post

The title doesn't fit the thread. You're comparing pretty much all relevant ICS sensors so why single out the 3366? Also I think it's really hard to be fair when comparing the sensors and voting in a poll like this for many reasons that might change your perception of how well a sensor performs.

  • Shape and wieght. (3366 suffers here more than others IMO)
  • Glide.
  • Surface used.
  • Different implementations as CorruptBE described.
  • Not actually having experience with all of the sensors listed.
  • Not testing them at the same time and instead judging from memory which isn't very accurate.

I still think the MLT04 is the best sensor that I have ever used. But that's from inaccurate memory since setting them to 500-1000Hz hasn't been an option for a while now.
I use it at 125 Hz. I find difference between MLT04 and other ICS is more important than lag.
Quote:
Originally Posted by Melan View Post

I have G303, G402, FK1 and MX500 right now. I also borrowed G300S and an old IME3.0 from a friend (don't have those anymore... i mean mice). All of them track fine except only MX500 being a bit unstable @ 1kHz. Even G300S with it's awkward sensor position was good.

Btw, I only use 400 CPI and lowish sensitivity (around 65cm/360). Since G300S was a 3055 I used it's default 1000.
I wonder how you tested it an in which game.
Quote:
Originally Posted by detto87 View Post

I hate such polls.
MOST people will read
"Which image correlation sensor you prefer?"
as
"Which image correlation sensor do you use?"
Sadly it's true. I don't usually use ICS.
Above8 is offline  
post #9 of 125 (permalink) Old 07-18-2015, 11:15 AM
New to Overclock.net
 
AnimalK's Avatar
 
Join Date: Jul 2013
Location: Montreal, Canada
Posts: 399
Rep: 20 (Unique: 15)
I have a g303 and I really love the shape. Unfortunately I could not get used to the 3366. It felt inconsistent and was unpredictable for small movements. Small movements would sometimes not register or would go further than expected.

I have many 3310 mice from various manufacturers and they are consistently my favourite as I find them responsive, consistent and predictable. The cursor always lands where I expect it.

AnimalK is offline  
post #10 of 125 (permalink) Old 07-18-2015, 11:22 AM
Mehehe!
 
CorruptBE's Avatar
 
Join Date: Dec 2011
Location: Belgium
Posts: 2,743
Rep: 86 (Unique: 50)
Quote:
Originally Posted by AnimalK View Post

I have a g303 and I really love the shape. Unfortunately I could not get used to the 3366. It felt inconsistent and was unpredictable for small movements. Small movements would sometimes not register or would go further than expected.

Might be the lens issue they're discussing in the G303 thread atm.

CorruptBE is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off