Slavishness to topology is silly. A few triangles on a model are meaningless to worry about, especially since they subdivide into quads.
Triangles? I hope I can fall asleep tonight after reading that word. 
The six-pole should be fairly straight forward to fix programmatically just by inserting an edge, while detecting the directionality of the surrounding quads.
Fun fact: triangles and five-sided ngons are not only used widely at Pixar but are actively encouraged for controlling poly flow on their organic models.

I have to admit… that… I’ve been guilty of using triangles and ngons as well, in a distant past. 
But seriously: of course triangles and ngons are no problem as long as they aid your modeling goal. Personally I’m a bit of an old school quad purist, which can become a self-inflicted obstacle at times. 
Everyone please encourage Autoremesher developer develop long as it wil surpass that closed QuadRemesher. I just now discovered that very very fine remesing is needed in 3d printing field when used correctly(not suitable for accurate parts but for art style things)
Autoremeshers progress since the 3rd Alpha:
Original Mesh:
Alpha 3 (2:42 min):
Alpha 4 (2:22 min):
Beta 1 (0:37 min), MIQ got replaced with QuadCover and simple quad extractor implemented:
Beta 2 (0:45 min):
Beta 3 (3:27 min):
Once more, sorry for little half-offtopic but i remeshed it little bit and fed to 3d printer. Model is not post processed.
So what happened between beta 2 and 3?
3 is much slower and has many topology hiccups. What are the advantages?
An extra step was added for the parameterization of thin surfaces and the quad extractor got some changes
Has anyone alse looked at this method of removing spiraling edge loops (q-helix), it would be nice if this could be added to Quadriflow and Autoremesher.

Paper: https://www.graphics.rwth-aachen.de/media/papers/bommes_2011_eg1.pdf
Nice! Did you also write this to Jeremy Hu on Twitter?
No, will do next time he tweets something reagrding Autoremesher
The developer of Autoremesher has started a new project to automatically add motion to cetain rigs, so the delopment of Autoremesher will probably slow down.
I would suggest to check it out, it looks quite interesting.
Blender really needs a QuadRemesher quality level intelligent remeshing library that is open, lets hope development is not so slow that there will be another 2-4 years before getting even little bit closer to QuadRemesher.
Hello, I just built a Blender Addon for Autoremesher by huxingyi.
You can find it here : https://github.com/Nicolas-Constanty/AutoRemesher-Blender/releases
Save your work before using it can crash depending of the mesh you are using (see Note…).
It is only working on Windows 10 - x64 at the moment, but I will probably add support on other plateforms if there is interest.

Note:
- I did not implemented the loading bar inside Blender at the moment, so when the computation is started, Blender do not respond until the end of the generation.
- The generation is pretty unstable (not working on lowpoly) at the moment, so you might encounter crashes depending of the mesh that you are using. Blender will be killed since I did not handle this kind of errors.
Same behaviour happen on the Qt GUI of the original AutoRemesher tool, so I did not handle it (because the author is aware of this issue, and is working toward to fix it).
Nice, thanks! 
The link at the Github site isn’t working for me.
Hi @Safetyman, it’s still working for me :
Here is direct download link if you realy can’t access the github link :
https://github.com/Nicolas-Constanty/AutoRemesher-Blender/releases/download/1.0.0-beta.3/mesh_autoremesher.zip
Got it… thanks.








