Total new guy here, when it comes to flying quads, but absolutely when it comes to tuning quads...

What is drifting on all three axes at the release of a throttle punch indicative of?  I'm using the factory tune on my Diatone GT-M3, and as I gain experience with, not only it, but quads in general,  I'm noticing dynamics that I would've have never noticed just a month ago.  Case in point, drift noticeable on all three axes when taking the throttle to zero after a punch (of any length).  It is visible in the videos at the end.

The clip is just of the throttle punch and a few quick snap rolls afterwards.  Three videos of the same maneuver but one with motor overlays, yaw PID loop parameters (yaw P - red, yaw I - blue, and yaw error - orange), and RC control inputs (to show the throttle and control inputs during the maneuver).

You can see how upon throttle release, it pitches, rolls, AND yaws, but I've only included the yaw parameters for reference that something is happening.  The motors show that the back left motor shuts off completely upon throttle release, whereas the others have non-zero values.  I'm not sure what's going on with that, but it's entirely consistent after every throttle punch/release.

Any ideas?

I am not quite sure what you mean by drifting. When you cut throttle, the quad will continue in the same direction of the punch out. That is normal.

Do you see how when the throttle is chopped, there is movement in the yaw, roll, and pitch axes? A large error develops in the yaw axis, for example, which can be seen on the yaw PID overlay.

The same thing happens for the other two axes (I'm flying in rate mode), but I only included the yaw axis overlays as an example.

Ok, I see it. Have you tried increasing your "I" gain. If your I gain is getting too high, I would increase anti-gravity. Anyway, here is a good explanation.
https://fpvfrenzy.com/betaflight-anti-gravity/

Thanks, voodoo! I will try that...I am literally at step zero in learning PID tuning on these things (quads). I've not touched a single loop parameter ever, but have only spent time reading about it.

I just ran a pack down, and noticed that depending on how long the throttle punch is (that is, effectively, the magnitude of external forces acting to destabilize the quad after the throttle is released), the magnitude of the yaw/roll/pitch drift upon release is directly proportional. For example, if I go full throttle for half a second and do a sharp release, the drift is much less noticeable than if I go full throttle for two full seconds. I would assume that this would be because the quad isn't going as fast after the shorter punch, and thus doesn't have as high external forces acting upon it as when punching for two full seconds then releasing.

This could relate to low I-gain? That seems to make sense.

Definitely low I gain. If your quad is flying the way you like it to be, I would just raise anti-gravity. If you want a more robotic feel, then increase I gain. If you increase I too much, your quad will not want to move.

The magnitude of the punchout makes sense. Longer punchout means more force.

Ok. I'm not far enough along into flying to really have a "feel" that I'm married to. I fly an Acrobee, a GT-M3, and an iFlight iX2, and besides rates, they all have different PIDs and drastically different feels, none of which I set.

Will try the anti_gravity, but may take a look at the I-gain just to see if it is unduly high or low.

Anti gravity gain was set to 3 already, so I didn't want to go higher than that (pitch/roll/yaw I-gains set to around 35ish). Maybe I could though? Unsure what the symptoms are of too high of an I-gain.

Also, the I-gain throttle threshold was set to 35%. I was super to see it so low, as I would imagine that most quads are going to give around that point, no? That's where my 2.5" and 3" hover, or thereabouts. I was thinking about bumping that up to 40% or maybe a little higher... Is that a bad idea?

I would bump I gain to 40 or even a little higher to see.

Too high I gain does make the quad very sluggish.

Since I-gain is accumulating error, will the quad respond to control inputs then bounce back towards the original orientation before the control input, or something?

I will bump them up 5 or 10 tomorrow and see what happens (while leaving antigravity where it is).

Thank you, voodoo, again.

I just read the sticky for the required info! I am sorry for not putting the proper info with this thread. I will include BF dump and potentially some BB log links tomorrow when I get opportunity.

Here is a great video with regards to I term. Although he never gets completely rid of drift, you will get a better understanding. Also remember, this video was made prior to Betaflight adding anti-gravity. That was the main purpose of anti-gravity. To keep I term flying the way you want but also get rid of the drift.

The sad reality is, you might never get a perfect flying quad no matter how much you tune. And you just have to live with good enough.

Would you say that the amount of stuff evident in my video up there is probably normal or on the border of what is acceptable to most ppl then?

It is personal. For some, what you have doesn't bother them. For others, it is something they will try to tune out. I tend to do moves after cutting throttle, so I don't notice drift at all. I can't recall why I would punchout and do nothing after except for tuning purposes.

Lets just say that I barely noticed it.

Ah ok. I guess it isn't that noticeable. But I didn't know if that was indicative of something wrong with the tune or what.