Layer line annealing in place during print

I have an idea I plan to test out for which I would like some feedback in case already tried...

I want to apply localized heat around the perimeter of each layer to cause the layer lines to smooth out from surface tension. Obviously there is lots of residual stress in parts so a big heat gun does not work so well, but I think this might work if the heat application is sufficiently fast and localized that the melt zone cannot penetrate deep into the part. So maybe can use an extension of the precisely controlled print head carriage for this purpose. At the end of each layer, run the source along the outer perimeter with the appropriate offset/speed.

I did some rough calcs and I do not think conduction will work, even a tiny air gap between your soldering iron tip/whatever seems like it poses too much thermal resistance, and contact would just be a mess. Have to apply heat fast and get out, without touching.

Do not really like laser either, energy comes from the top so you're going to melt a lot to melt what you want. Plus they scare me.

The idea I've settled on is a downward pointed mini hot air jet like from a rework station, operated at a low flow rate. At near laminar conditions, no-slip condition, and especially with the buoyancy of the hot air, I think it will roll around the nozzle tip and stay very closely confined to just below the nozzle and around its outer perimeter. Basically a version of the soldering tip that works, but potentially the nozzle tip can even sit slightly above the extrusion nozzle, so it doesn't have to be pushed up and down. This would rely on being able to control the depth of the stagnation point at which the hot air turns around and flows back up.

Modeling that doesn't seem worth the time vs simply testing. My current plan is to take a v6 hot end, open up the nozzle and press in a spiral spring pin to get the sharp torch I want. The heat sink and heat break of course are needed so I can use 3d printed parts for the fan holder and duct above.

This idea seems backwards, mainly because too much heat is associated with curling, lack of detail and a whole other bunch of printing problems.

That said, maybe it could work? Why is it different to a heated chamber?

Curling comes from hot plastic contracting on top of cold plastic, repeated over and over. I think this would not do anything one way or the other. The idea is that the added heat is localized to melt a small area but is not sufficient to allow softening once diffused beyond it.

Re. lack of detail: This would of course reduce detail, it'd be a kind of anti-aliasing. The idea is to remove the artifact of the process while eliminating as little of the underlying information as possible.

It is different from a heated chamber because the imposed temperature change would be beyond the melting point, but only for a very short time on any given area so that only the layer line closing occurs (or at least changes the cusps to something less noticeable).