(NB: I've tried to tune the phase, too, with the Paragraphic Phase EQ, but the phase is too messy and I've never succeeded to get good results, so I give up)
Try the second tab called "Filters Linearization" and play with the numbers -- it may significantly reduce the number of filters needed in the paragraphic phase EQ.
Exporting the impulse file (65536 taps, the right sample rate) and uploading into the convolver
Maybe that's a bit too many taps. Try 16,384 if your EQ'ing down low in the bass. The correction would be "fuzzier" the lower you go, but that extra bit of precision may not be so important down there anyways... I only use 2048 taps now primarily for phase correction and and still am getting as good an "alignment" as possible -- with the "textbook" step responses now as well.
All other PEQs duties are simply relegated to another EQ program elsewhere. Figuring out how to best fiddle with the time domain, esp. with multichannel setups
manually can be a rather complicated undertaking. I would relegate it as something that can be done
much, much later if bored out of your mind.
But you have to learn how to use
frequency dependent windowing (FDW) to clean and smooth out the phase responses. In an untreated living room -- and with multi-way speakers -- I know it can get pretty messy there... It's in the first button beside "dB SPL 83" of REW. As well as export (may be better from average data) the phase and/or excess phase into rePhase.
... Sometimes the phase can look wonky because of the unavoidable room and placement -- after some EQ it goes back to looking normal -- don't exactly know why this happens, but I suffer from something similar and the phenomenon is very
measurable and
repeatable.
See the anomaly of the HF:
It's very odd, but it is also very consistent/repeatable. Only after applying my HF shelving EQ's seems to do the trick (match with the opposite channel) for whatever reason.