And if you do not use -B0 you should see no frame.

But things are not completely right (and Claude warned me about it). The title is not rotated.

But things are not completely right (and Claude warned me about it). The title is not rotated.

Right. I didn't pay attention to that.

The title angle is now fixed.

This PR now causes test/grdmix/grdmix.sh test to fail. But I think the grdmix.ps baseline is wrong because it misses the frame that is implied by that script using -B0. So, it turns out that apparently we killed two bugs with one stone.

The title angle is now fixed.

I'm affraid that I try it and I am getting the same figure.

Maybe you didn't get the updated branch? Can't retest right now as I'm debugging something else in another branch.

I checked in another computer and this is working fine. Will merge it but test test/grdmix/grdmix.sh will start to fail because baseline PS is wrong.

Maybe you didn't get the updated branch? Can't retest right now as I'm debugging something else in another branch.

Yes, I think I had an issue with this. Now, I have this version from today (6.7.0_f6474f9_2026.05.15) and I am getting the same figure that I posted before.

Not sure I got you you mean. You should get this:

No, I am getting this figure:

Now I tested it on my laptop with windows and I got always the same figure.

This is the GMT version: 6.7.0_f6474f9_2026.05.15

Now, I managed to get the title rotated with the script (without -B0):

gmt image @needle.jpg -Dx0/0+w7c -JM12c -JZ14c -p200/30 -B+w1p+gred3+t"title" -png plot2

Here are both commands that I used to compare:

gmt image @needle.jpg -Dx0/0+w7c -JM12c -JZ14c -p200/30 -B+w1p+gred3+t"title" -B0  -png plot
gmt image @needle.jpg -Dx0/0+w7c -JM12c -JZ14c -p200/30 -B+w1p+gred3+t"title"      -png plot2

I looked at the PostScript file generated by this command and I have to say I do not like how it turned out. From what I learned while creating perspective text, when the -p option is used (with 2 values), this translates into the PS applying a transformation matrix that is later reset at the end.

What I see in the PS is that the title was left outside the block affected by the transformation matrix, and instead a rotation was applied directly:

V 10.3141048156 R (title) bc Z U
I suspect that, since a rotation is being applied instead of using the same perspective transformation matrix, the title placement or orientation may not look correct for some parameter values or viewing angles.

I suspect that, since a rotation is being applied instead of using the same perspective transformation matrix, the title placement or orientation may not look correct for some parameter values or viewing angles.

For example, I got this figure with this command:
gmt image @needle.jpg -Dx0/0+w7c -JM12c -JZ14c -p190/5 -B+w1p+gred3+t"title" -png plot3

Hmmm, maybe right but this also

gmt basemap -R0/10/0/10 -JX12c -JZ14c -p190/5 -B0 -B+t"title" -png lixo