the 3955's sensor center offset implementation is NOT doing what you think it is and this 100% proves it by comparing it to my prototype with an actual virtual sensor offset. It seems like it's just an x-axis DPI scaling to VAGUELY approximate the effect of a different sensor... position (but it's way off).show more

Yet another useless feature that brands will no doubt spin as the next game changing feature 💀

my worry is that people will then start thinking "virtual sensor offset == useless" which isn't true, it's just that this ISN'T a virtual sensor offset :(

That's fair. Personally I can't see a single use-case where it'd be useful, I'd rather just use a sensor-tilt imo 😅

I give some examples here of cases where I think being able to use a lower sensor position is beneficial, so a virtual sensor position allowing you to lower it for specific games for example is useful, but when done properly there's a bunch more that can be done that I don't want to spoil :P

That's fair. I assume it'll have some niche use-cases. I'm more interested to see how many brands will actually enable the feature, I feel like this is something a lot of brands will try and steer away from just because of its complexity, same situation as sensor tilting.

I mean, this specific implementation is a joke, you can do the same by just changing x dpi to be lower than y which every mouse has been able to do for 12ys, it's crazy that it's being called virtual sensor

I totally agree with that. It's also kind of wild how many mice don't support split X/ Y-axis DPIs, which is so confusing as it's such a simple thing to implement on 3395/3950s.

for whoever asked me for the dpi tool, it's a basic vibecoded win32 app that registers for rawinput and draws to screen, you can use any other mouse tester app, but I still uploaded it here if you want it

It's a shame that it doesn't work as we hoped but at least we have one product with virtual sensor offset in the market now. Can't wait to see more moving forward - thank a bunch for your content!

Thanks for sharing this, really disappointing to see

yep, I’m definitely very disappointed with it

I mean yeah??? Like if the sensor return you a 2d motion vector, how the f you can get a rotational part of it from ONE sensor... You need 2 sensors for it And okay, maybe not everyone should know basic 7th grade math&geometry, but seeing this from atk devs is disappointing

since I think this is a first party offering from pixart, my guess was (had it worked) that they were either using a much more sophisticated optical flow algorithm to try to estimate rotation from the features in the frames (which realistically is almost impossible given the small size of the sensor and the distance to the pivot), or that maybe they had integrated something like a gyroscope and were trying to use that to identify real-time yaw rotation to scale dpi dynamically, because yes by default you can't do it with a single sensor! turns out it's neither hah

Your skills are crazy, one of best aimers and also computing skills. Is this mouse btw done fully by you or have you had some help?

ty! the mouse is a completely solo project, I did the electronics/firmware/cad all on my own

Nice to know that there is always someone in community who knows in and outs and can explain it decently, to add to it how does brand big as atk fuck this up, if it can be done by solo dev, qc?

you kinda need two optical sensors to do it, or an sensor and some other type of sensor to help figure out if the mouse is rotating, so when I saw it was done with a single sensor I had my doubts. I am not sure if the issue with the branding of the feature here is on ATK or Pixart.

Sounds like nightmare for qc, unless someone puts it on market the one sensor implementation will only exist. Possibly only niche brands could be brave enough to do first move.

Great video brother.

Just dove into this vid with no introduction from your previous posts, but what i can understand is that virtual sensor offset (y) shouldnt have any resulting delta along x or y for that matter, but have a radial effect where the turn radius changes depending on the offset right?

yes, the length of the arc drawn by the sensor around the wrist pivot is directly proportional to the distance between the sensor and the wrist, while straight lines drawn by the sensor by moving the mouse rigidly aren't (they're just translations). I explain why it can be meaningful here:

Thanks boss, and keep up the good work, will take a gander

Whoever want to deep dive to sensor and math

Gonna be interesting to see a real virtual sensor offset, and not... Whatever this is.

Thanks god Bardoz for actually good testing, every reviewer of that mouse should just do X and Y testing like you, great job for showing it to community

pixart next sensor will improve this

You showed Donks sensor shifting more back. Lets talk now about shifting sensor back and left a little. Moving to right side more limited than to left side. Only then angles on your pic will have same angle. I can talk only with you about that because.... you will understood

Damn, I was afraid that was the case.

You know what's crazy... The x-axis scaling makes this sooooo good for me. I get to keep the benefits of the high sensor (pencil like micro adjustments) and not lose out on too much agility for 180s etc. On Lycan I am running -32 and somehow feeling VERY consistent.

my issue with this is that this has been feasible with any modern mouse for like 15 years where you can change x/y sensitivity individually, even some games themselves allow for x/y sensitivity adjustment independently, it has nothing to do with a sensor offset

Yeah honestly this way of packaging it just brought it to light for me but fully agree that it's nothing new. Just a fake "feature" that will be used to sell more products. Capitalism!!

It's weird af, I wanted to test it on atk a9 mini+ that I've got, I play 47cm/360 in cs and if I put even -1 in virtual sensor offset I can't even do 180, while on 0 I can do easily 360 which doesn't make any sense.

it’s just changing dpi, +100 compared to -100 is the same as going from 1 sens to 1.5 sens

On my mouse it's kinda different On 1600 DPI 0 - 1600 dpi (can do 360) -1 +- 800dpi -100 +- 1600dpi +100 +- 2400 dpi Like the minus values are reversed, + kinda works, but still doesn't matter since it doesn't even do what it was supposed to in the first place.

oof, that’s terrible

Isn’t it obvious that proper sensor offset simulation is not possible with only one sensor? I tried the same via a windows driver years back but then also realized the obvious constraint here. Not sure but GWolves could in theory add it to their 2 sensor mice?

umm i dont use this feature much but what does it mean

Can you give the Viper v4 pro a try on this? It has similar LOD / sensor problems

@g_wolves @review_wolves The Vuk seems like a good candidate for REAL virtual sensor offset.