Made a video showing 95% of this.

https://www.youtube.com/watch?v=evInNvENf7o

fun fact: the rescue ranger doesn't delay the ability to pull buildings while firing (not shown in video)

Sorry for using something easier to understand when you're asking on a forum dedicated to competitive TF2 about the performance on mouse sensors A vs B down to a level of microseconds of difference. I automatically assumed you wanted someone to make it easier to understand. What I'm curious about though, is if you're going to be such an ass about specific details of performance why don't you look it up yourself, or do you really need someone to hold you hand and walk you through it?

forewarning you'll probably never look at a mouse the same way if you actually look through all this. double warning if you love using zowie mice.

useful PMW3366 info
https://www.overclock.net/forum/375-mice/1561041-reverse-engineering-3366-a.html several posts in this thread contain a lot of useful info about the performance of the PMW3366 mostly by the user "qsxcv"

click latency info
https://www.overclock.net/forum/375-mice/1607990-click-latencies-compiled.html check most of the posts by "uaokkkkkkkk" and "qsxcv"

mouse reviews
https://www.overclock.net/forum/search.php?searchid=6517376 this user "Ino" does a lot of really good mouse reviews and will give you a pretty good idea of what each mouse is like and 99.9% of the info you could possibly want on the mouse. He's also a user here http://www.teamfortress.tv/user/ino

mercury csg sensor (this is similar but NOT the hero sensor)
https://www.overclock.net/forum/375-mice/1615853-g102-mercury-sensor-info.html again "qsxcv"

info on how to spot significant smoothing
https://www.overclock.net/forum/375-mice/1554228-visualizing-smoothing-mousetester.html again "qsxcv"

other threads/posts by qsxcv
https://www.overclock.net/forum/375-mice/1566490-analysis-mousetester-plots.html
https://www.overclock.net/forum/375-mice/1560877-smoothest-mousetester-plots-ever.html
https://www.overclock.net/forum/375-mice/1613050-razer-deathadder-elite-pmw3389-t3qu-internal-pics-sensor-info-etc.html PMW3389 info as well
https://www.overclock.net/forum/375-mice/1588408-teensy-mod-firmware-g100s.html
https://www.overclock.net/forum/375-mice/1563813-somewhat-objectively-evaluating-sensor-responsiveness.html
https://www.overclock.net/photopost/data/1380344/f/f6/f6fff350_9NIU9UJ.png this proves that the 3366 does have smoothing despite logitech claims of absolutely 0 smoothing not a particularly big deal it was expected.
https://www.overclock.net/forum/375-mice/1580918-sensor-question-3310-vs-3988-a.html both skylit and qsxcv have good posts

This guy used to have a website with a TON of info https://www.youtube.com/user/recklessyuki
https://www.youtube.com/watch?v=wO22BuqLnYc video comparing sensor responsiveness of the G403 vs Rival 300 vs EC1-A
https://youtu.be/KHLsmIB8osQ?t=97 here you can see some of his website that is no longer up

there are a ton of great posts on overclock.net by "Skylit" mouse reviewer in the past still has alot of valid inputs "Ino" current mouse reviewer + a lot of valid inputs. "qsxcv" dude hardcore tries to figure out in minute detail how alot of peripheral equipment works I also regularly see him on the testufo forums for monitors. hell he helped design some of the firmware for the ninox astrum."CPate" Logitech hardware Rep basically a direct line to logitech engineers for a lot of his posts.

So many questions lol...

What point in 2008 is this version of the game?

Things added in 2008

Tournament mode
Ability to disable crits
Sniper headshot scope delay (200ms)
Medigun autoheal
Multiple significant particle optimizations
cp_badlands
dropped weapons counting as ammo
Rocket Launcher, Sticky Bomb Launcher, and Grenade launche all saw a signficant reduction in ammo pool
Spy can desguise as teammates
Uber draining faster when ubering multiple players
Blutsager, Kritz, Ubersaw

there's numerous others but it's kinda important to know lol

Thank god Tery was ever vigilant in watching the times/days these guys were trying to schedule matches.

"Obviously" you specifically mentioned the 3360 which isn't the 3366...

Now if you really want an idea of performance between the hero sensor and the PMW3366 in terms of smoothing below 2000DPI will be the same i.e. 2 frames. Beyond that it's incredibly hard to say for sure as I have yet to see someone do extremely intensive testing on the hero sensor.

In terms of raw performance this is Logitech's own graph on it

https://i.imgur.com/5zvVdcq.png

basically the Hero Sensor is 99.9% of the performance for ~20-25% of the power usage. Which is still significantly better than the 3310 in zowie mice, and at minimum on par if not better than the 3360, 3389 and TM3 sensors

Yes.

"noticeable smoothing"
Alot of things here.

What is noticeable to some isn't noticeable to others.

so to begin ALL OPTICAL MICE HAVE SMOOTHING period even the beloved WMO and the "flawless" PMW3366 in the G502. You need at least 2 frames of smoothing otherwise any kind of visual noise on the sensor will cause it to jitter even if you're not moving the mouse.

Generally speaking modern mice run anywhere from ~6000fps (when in use) to ~12000fps (when in use). so 2 frames = 0.33ms at 6000fps and 0.16ms at 12000fps.

so all optical mice will have at least 0.16ms of smoothing. most optical mice when in the low DPI ranges have 2-4 frames of smoothing.

G400s at sub 2000dpi (maybe 2400dpi) is 2 frames.
G100s is 4 frames below 2000dpi
DeathAdder Elite is 2 frames at 1800 DPI or less. ~32 frames at 1900+ DPI
G502 is supposedly 2 frames at all DPIs (for sure is 2 frames below 2000DPI)
Zowie PMW3310 line up is ~20 frames of smoothing at all available DPI levels*
PMW3360 mice all have at least 2 frames of smoothing at sub 2000DPI. ~32 frames of smoothing above 2000DPI

So when someone says "noticeable smoothing" take it with a MASSIVE grain of salt. 9/10 it isn't smoothing they're noticing or they're using a really high DPI.

I haven't seen any official tests with the hero sensor but it's safe to say that below 2000DPI will be 2 or 4 frames of smoothing max. Worst case scenario is ~0.33ms of smoothing when actively using the mouse. If you stop moving the mouse for like 5s and then start moving it again it might be just above 1ms of smoothing but it'll go back down once the sensor first detects movement.

* I question the validity of the statement of 20frames and how resolute that is with zowie mice, but I couldn't find anything better. I'm pretty sure the X axis has less smoothing than the Y axis but I doubt anyone actually teensy tested that to come up with exact numbers.

I mean comparing the Nixeus to the old mx mice it's definitely an upgrade, no angle snapping can do more than 125Hz without having to fiddle with shit, but the click latency of the MX518 (2005) vs the Nixeus Revel (2016) is kind of an interesting point. Comparing the MX518 click latency at 125Hz to the click latency of the Nixeus Revel at 1000Hz would suggest that the MX518 should be about 7ms slower on average than the Nixeus Revel it's actually about ~2.4ms. Force 500Hz on the MX518 and it's about 1ms faster lol.

So why isn't corsa on cafe black then, or did he just not want to play?
How the fuck does botmode get fucked out of the roster he was trying to save?

Literally every player that I was really excited to watch this season is out now.
no duwatna
no delpo
no yomps
no botmode
no corsa

:(

The issue you described is related to the lift off distance of the mouse sensor and/or the max tracking speed of the sensor. If you have a non-solid black cloth mousepad that's clean a lot of sensors don't track properly and will cause either what you have or a lot of noticeable jumping. Another part of what could be going on is that your lifting and putting your mouse back down at a slight angle causing the sensor to suddenly see a moving image with part of the image being closer/further away then the sensor expects a even surface relative to it and registering A LOT more movement than what is actually occurring.

Possible Fixes
1. Clean your mousepad
2. Get a black cloth mousepad
3. Clean the mouse sensor (qtip dipped in water or rubbing alcohol)
4. break the solid black piece of plastic that covers up part of the sensor area if you do this and fuck up that's on you
5. Get a non-zowie mouse

Highly doubtful, you likely have thicker mouse skates than before. Thankfully Logitech makes this a pretty easy fix. Download the "Logitech Gaming Software" and go to the Surface Tuning stuff and Tune your mouse to your mousepad.

It happens to all Zowie mice I've ever seen/used, and if given extreme enough examples basically every mouse will do this. More common with mice with "ultra low" lift off distance and are near the point where the sensor can barely see the surface it's supposed to be tracking.

some other clips from that match...

https://clips.twitch.tv/RepleteRoughPicklesPJSugar panther's POV
https://clips.twitch.tv/DoubtfulBreakableBobaFailFish my POV

https://clips.twitch.tv/DelightfulPiercingMonitorKeepo mid-air airshot exchange

not sure what you're referring to here.

I think they scale the target movement with the framerate rofl

my top 5 all happened with the dumbest settings like 60Hz or instead of a 480fps cap a 240fps cap or 120fps cap

anyway 92.7%

so I've done a fair amount of testing with this bug.

1. Bug still exists

2. Bug occurs when switching weapons and during melee attacks.

1. Bug "doesn't" occur if you switch to the grenade launcher within the first 300ms of firing the sticky. this is because the switch speed is fast enough to complete before the sticky can be det
   
2. Bug is far more noticeable with demo "sword" melees since they have longer switch times and attack intervals
   
3. stickies that are already setup in a trap are unaffected by switching to a different slot BUT are affected by the melee attack delay. given this nature the slot switching and melee attack sticky delays are likely separate bugs

Item Switch sticky delay

If you switch weapons in a time period that would affect the earliest start point of the sticky detonation, and switch back to the stickybomb launcher the sticky detonates immediately so long it's 800+ms after the sticky is fired.

To break it down

Fire sticky -> 700ms later switch to grenade launcher (which takes 500ms to complete) -> 100ms later stop weapon switch with slot2 bind -> sticky detonates 0-15.2ms later without needing to fully switch to the stickybomb launcher

which is better than Fire sticky -> 700ms later switch to grenade launcher -> 500ms later sticky can be detonated after 1.2s

Again items with longer switch speeds makes this worse (unless you switch back to detonate)

Melee attack sticky delay

Name is exactly what it suggests, noteably worse with demo "sword" items due to longer attack intervals. Usually only effects sticky traps, but is technically capable of delaying new sticky detonation.
ex.
Fire Sticky -> during first 299ms switch to non-sword melee (the bottle/paintrain/etc) -> melee attack immediately -> sticky dets at the end of the melee attack

Items affected in one way or another by the above delays
Stickybomb launcher
Quickiebomb launcher
Scottish Resistance
Detonator (pyro secondary)

• this is actually particularly dumb switching from the detonator makes it so you can no longer detonate the flare even if you switch back to the detonator**

Sandman (melee attack blocks secondary)
Wrap-Assassin (melee attack blocks secondary)
Spy-cloak (melee attack blocks cloaking until the end of the stab)

Items that are surprisingly unaffected by the above delays
All demo shields are capable of charging during melee attacks

The reason Valve hasn't fixed either issue (for sticky bombs at least)

source

They think they fixed it last year. (granted they might have fixed stickies that have existed for over 700ms being able to detonate while switching weapons as that definitely no longer occurs)see edit below

-edit-

Stickybomb arm time is ~0.8s. Updated numbers accordingly.

So in December 2017 Valve fixed part of the issue.

https://youtu.be/AtPeAzYuXaE?t=25s
this now behaves in the same manor as the grenade launcher as in the switch can cause a delay if the end of the switch is after the sticky arm time BUT it will detonate. Unable to verify if switching to melee also delayed sticky trap detonation, can confirm that it's definitely not an issue now anyway.

-edit2-
**You can detonate the flare after switching back you just have to wait the full amount of time to deploy the detonator again to be able to do so... Missed in testing due to the flare dropping behind stuff and me assuming the explosion would be more obvious.