Magic Mask on Still image/No Tracking

Can't figure out how to get the Magic Mask to cover entire length of a 'STILL' image without tracking the whole (non video) still image. I used the Magic Mask to get the person in the still image selected by the Magic Mask but it's only for the one frame the playhead is sitting on. To get the mask to cover the entire images length in the timeline I have to track it.
The mask doesn't change throughout because it's a still but I still have to track it to get the mask to cover the whole image across the Timeline.
There's gotta be a way to tell it to use that frames mask across the whole length of the image.

That's just how the feature works.

You might be better off doing this in a photo editor like Affinity or Photoshop.

+1 on this annoyance. Would like to quickly mask part of a still image across the length of the clip. Magic mask works well on... 1 frame of the still.

What about 'Render in Place', so it's seen as a video?

@Uli, granted, and I appreciate the spot-on advice.

I also typically use Render in Place soon after magic mask operations on original video as well, to avoid issues with losing the mask or getting quirky results in the color page node tree during further work (even with a compound clip wrapper, I experience intermittent issues). I don't like having lost easy visibility to the original node trees, for tweaking the grading nodes and/or other fusion work. Render in Place is the available option for a more reliable workflow so I do it, but I don't like it.

And I feel badly believing that people have to struggle and then search to learn a workaround to enjoy such a wonderful masking tool, rather than the tool just working naturally.

And I feel badly believing that people have to struggle and then search to learn a workaround to enjoy such a wonderful masking tool, rather than the tool just working naturally.

Well, a tool like this wasn't designed for a 1-frame image. If you render it in place as (say) a 5-second image, then it'll work.

If it were me, I'd just draw a fairly precise mask by hand as a power window and use that. And use multiple windows if necessary to cut out the image and do what I needed to do. It's not that hard -- it just takes time.

Well, a tool like this wasn't designed for a 1-frame image.

Most monocular depth estimation models (and most probably the one in Resolve too) work on single frame, not sequence, they don't take adjascent frames into account, result for specific frame doesn't depend on those. So they are actually designed exactly for 1-frame images.

Another workaround is to put the image with the magic mask inside a compound clip and freeze the first frame of the compound.

I found one workaround for this.

Render in Place did not work for me. Still Magic Mask only did a good job in the first frame and lost some of the masking data further on.

In Fusion, I put a TimeSpeed -node right after the MagicMask -node and freezed it to the first frame (0). That way the initial mask continued through the whole clip duration.

Hope this helps someone

I found one workaround for this.
In Fusion, I put a TimeSpeed -node right after the MagicMask -node and freezed it to the first frame (0). That way the initial mask continued through the whole clip duration.

Thanks for sharing your workaround for this annoying behavior.

I found one workaround for this.

Render in Place did not work for me. Still Magic Mask only did a good job in the first frame and lost some of the masking data further on.

In Fusion, I put a TimeSpeed -node right after the MagicMask -node and freezed it to the first frame (0). That way the initial mask continued through the whole clip duration.

Hope this helps someone

It did indeed. Many thanks.