No, I'm against such an interpolation when my models (I barely print anything found online) are at least as accurate already :)

That's why we should choose the result, which has slight "over extrusion" in the new test

That's my finding too, my point is that this choice is all but obvious, since in the past (and the wiki) instructs to pick the smoothest surface.
We should clearly state "smoothest spiral surface with at least some bulge appearing".

When you select flow, which makes the surface flat on the old style calibration, in reality you are over extruding to achieve that flatness

Isn't it because you exactly fill the gaps, while the math expects a rounded extrusion profile?

Try to print some cube with concentric top pattern, the result will not be nice :)

Isn't this a very specific case? overall prints calibrated with the old method never showed any sign of overextrusion in my case. And their dimension accuracy was really good too.

About the gaps in the corners, check the sliced model carefully

Where did I mention such an issue? I was referring to gaps in the spiral, or between spiral and outer portion (basically, where the bulge would appear).

So you must choose between precision, or slight over extrusion to mask this imperfections of FDM printing with chance, that on some places the over extrusion will be visible.

Ok, but how should I choose for each alternative? it's not as obvious as it was in the past and it's not explained anywhere.

Maybe better propose some text in the comments for #9370

At least put a link to this thread in the wiki linked from the calibration menu--perhaps where it currently says TBA. Would have been very helpful just to read this dialogue two weeks ago.

At least put a link to this thread in the wiki linked from the calibration menu--perhaps where it currently says TBA. Would have been very helpful just to read this dialogue two weeks ago.

It was updated as a work in progress a few days ago. Let's not forget that this option is in the nightly build which is not considerable stable yet.
Even so, in the next wiki update I am trying to make it clear what this calibration is looking for.
Any recommendations that you consider necessary to clarify or specify? All help is welcome.
#10369

As someone without much printing experience, I would most appreciate just a checklist style laying things out super unambiguously. Something along the lines of a paragraph explaining the intent of the test and why it works, then visual and textual descriptions of low, good and high flows. Pointing out how much of a difference putting a light at a low angle to highlight the texture. If possible a set of shots of full outputs with different filaments, with a clear indication of what value you should be choosing based on each picture.

I hadn't realized that these were exclusive to the dev build, maybe that could be conveyed on the popup after selecting the test. Along those lines, could you consider making sure that the old calibration method is still doable--I was trying to contrast the spirals and the old method and it was hard to get back. Maybe even just a radio button to let you choose the geometry.

Are you sure you have correctly calibrated E steps (rotation distance) for the extruder?

I thought the same thing, so just to be sure, I measured 100 mm of extrusion again and it was perfect.

I also printed a benchy with the value I got after recalibration and it clearly has under-extrusion.

Interesting. I once had similar problem (the flow test showed lower values then expected and the prints were then under extruded). I found that the problem was in the nozzle (or more precise in the heat brake), which was partially clogged...
Is interesting that the floor inside the benchy is pretty nice, but then there is heavy under extrusion in the walls in the small cube. Very similar to what I experienced with the clogged nozzle.

the floor inside the benchy is pretty nice

Not so nice :(

@ianalexis can you check what volume flow is used for the top surface of test in your case?

Was reading about that, will take a look.

This, using it for a few weeks and it's so much easier to read, there's a very clear sign. I'm not sure it even matters if users are 1% under extruding through not reading the instructions given how much many commercial printers under extrude out of the box in order to be very forgiving for new users.

Very good points, @pi-squared-studio.
@igiannakas I can confirm that the outer "half-moon" lines on either side of the circle do print in alternating directions with each line. When I use the test, I found the best results came from choosing the chip where the union line of the circle can first be felt on both sides of the circle, not just one side (then adding 0.01 to its value). I imagine this mechanics issue likely has something to do with that, but the solution - at least for a 0.4mm nozzle - seems simple enough following the process I just stated, unless we can rework the test to be more universal for nozzle sizes, and remove that level of "user error" out of the equation (empirical results are far better than subjective ones).

So at the end the best way should be in theory use just simple lines in rectangle, but print these lines always from the same side (not zig-zag). Print one half from up and then second half from bottom. This should end with one line space in the middle for the last line and the "walls" are all built in the same direction.
Another question can be, if it's better to use line aligned with one of the axis, or 45˚ lines.

The solution really is to change the pattern code (it’s bespoke anyway) to ensure the circles print in the same print orientation (counter clockwise). Keep the outside in part and inside out part.

Also the pattern is a good idea as it eliminates PA based tuning artefacts which lines suffer from

This works really well for me with a 0.6, I took the time to check this with a 0.6 and a 0.25 as i have a revo on my V0 and i've just put it back together. Readings are identical to what I got with a 0.4. It's really hard to read with a 0.25, but the ring method is easier to feel than using a monotonic surface. That said I prefer to calibrate filaments with a 0.4 then swap.

You will get awful results with a 0.6 or bigger if you're too close to the max vol flow and that's really easy to do if you don't have a high flow setup. The test looks to be generating chips with a flow of 15mm3/s with a 0.6, that's going to be in the inconstent behaviour territory of a lot hotend setups, outrunning a V6 or standard flow revo, and in the under extrusion region of PETG with a CHT on a V6 or a revo HF unless temps are very hot. Filament pictured is eSun bone PLA+ which i'm mostly using for minis, it's good for about 14mm3/s with a v6/revoSF and 0.4 or 0.6 and about 24mm3/s with a revo HF or TZ-V6, and I think the latter is similar enough to a bambu hotend that I don't think it's anything to worry about.

Maybe the wiki needs to have a warning about outrunning the hotend whenusing a big nozzle and the order in which to do calibrations.

Thanks for the report. Good point about the volume flow, this should be mentioned in the manual to check.

If you look at the test results, then you only need to look at the central circle of this test, it will give more accurate results than filling with diagonal lines. Another point is that the printing direction of the spiral track is important. If it is from the center, then this is the ideal result, if it is "to the center", then the plastic will turn out to be loose when printing, which will tend to turn towards the center.
Orca, like other slicers, seems to have a print optimization algorithm that will always rotate the spiral from the outside to the center.

Have you actually run this test? It is programmed to turn the circle from inside to out, and yes, everything outside the circle is to be ignored.

I encountered a similar effect when I was working on the above PR. I did several tests, and the direction of the spiral gave too different results in printing. For convenience, I'll copy the picture here.:
On the left, lines are printed in different directions in the corners, and the center spot is from the edges to the center. However, this is the underside printed on smooth glass. As you can see, the coils on one side of the spot have ripples, which means that such printing conditions are not optimal. Plastic, gradually solidifying, tends to curl inward without having a "support" in the form of an adjacent coil. Yes, these two samples were printed together on one plate.

The flow rate test use custom generator, so it ensure the right order of the each test: #9145

Yeah, too many irrelevant variables in this example; not a good comparison. Like Benik3 said, the test uses a custom generator, it is not as simple as just changing the solid infill pattern. This additionally renders the bottom layer irrelevant for both that reason and the fact that it's the bottom, which is influenced by factors from the bed, like elephant's foot, and inconsistent bed mesh and z-offset, which could be the actual cause of the rippling you see. This is why the test purposefully has several layers of sparse infill to decouple the top from being influenced by the bottom, which has been the case with the old/current test too for years.

Not really. The main goal is to "separate" the testing layers from the bottom one. If you do it thinner, the test print will be affected by the first layer and overall flatness of the bed, nozzle height calibration...
I tried it already and the effect is strong.

Thank you for explaining!