Smoothing on curve editor in edit page

I noticed that when working on the clip curve editor, if i activate keyframe smoothing on zoom curve the curve gets effectively smoothed out while doing the same thing in x/y position the smooth handles appears but the curve is not automatically smoothed, it keeps being linear and a manual intervention is needed for actual smoothing.
I wonder why this behaviour and not the same of the zoom curve which i think is the desired one when activating smoothing handles.
Moreover, the editor is generally screen space limited and it's difficult to place the smoothing handles precisely.
This applies to all Resolve versions that support this function, as far as i can recall.

I can't duplicate this in Studio 19b2 for Windows. Seems to be working fine.

i'm referring to the curve editor that unfolds pressing the curve button on the clip, not the one on the viewer.
Create two keyframes for a clip in two different positions. Open the curve editor then select X position or Y position (whatever you moved the clip on), select the first keyframe then select the ease in button. The smoothing handle appears but it's aligned to the original linear movement segment.

That's exactly how I tested. Worked fine.

This is what happens on my machine (Windows 10)

The position is not automatically smoothed when activated on the node while the zoom is (which is desirable imo).

What happens if you select both key frames (start and end) and then run smoothing on it ?

i'm referring to the curve editor that unfolds pressing the curve button on the clip, not the one on the viewer.
Create two keyframes for a clip in two different positions. Open the curve editor then select X position or Y position (whatever you moved the clip on), select the first keyframe then select the ease in button. The smoothing handle appears but it's aligned to the original linear movement segment.

Yes - this is how it works as far as I can see. Can't add the 'smoothing' automatically in the way you are thinking of it (per dimension) because doing so would move the object off its path. The whole XY position key framing is a bit of a problem more generally since smoothing is not really performed on a per-dimension basis but rather in two dimensions over time. That's probably why there are additional acceleration handles in the transform tools overlay for position changes. These should be used rather than keyframe manipulation in the keyframe graph for position.

I should note that when you keyframe XY and set ease in / out for the keyframes in the inspector, Resolve does smooth the velocity, just not as a curve in X or Y curve editor graphs.

Add - Playing with this a bit I noticed that the curve graph does not accurately depict the acceleration options selected in the inspector (making the curve graph useless for manipulating X,Y position keyframes, really). So again, it looks like your best bet is to stick with the transform tools overlay rather than the curve graph for X,Y position. Note that on the transform tools overlay you can also change the keyframe to 'smooth' by right-clicking on it.

Yes, i get the point. Anyway, the overlay interface is somewhat a clue to use, i cannot wrap my mind around it.
Hope that something can do in future on this aspect, being able to use curves like in fusion could be useful.

might sound like annoying advice, but i never use the keyframes/curves in the edit page. Opening the fusion page might feel cumbersome at first but once you get the hang of it, it is just as quick or maybe quicker.

i never use the keyframes/curves in the edit page.

Same. If the right click option on the keyframes in the Inspector don't produce satisfactory results, it's time for Fusion.

Relevant video & a handy “all in the Inspector” fix:

Sent from my iPad using Tapatalk

That bit about using the anchor point is a gem

Add - Just don't try to ease-in/out the anchor point - at least not on 19b2 - I just tried it (using the anchor point to create a smooth pan on a fixed zoomed-in image) and it went totally bonkers. Linear works fine.

Looks like the math is incorrect when interpolating between anchor points such that the x value goes full negative and then wraps around to the final positive position. The y interpolation seems fine. Very odd.