Hot air lead-free BGA reflow

I'm experimenting with trying to reflow a 0.8mm BGA myself for prototypes/repairs.

I've solved the main problem: alignment. I've used my 3D printer to print custom board guides that align on large through-hole pads and guide the BGAs into place. So far, I'm 100% on alignment for the 4 BGAs I've done so far.

Now, I'm trying to reflow it with my hot air system. I've got a BGA tip that fits over the part, and I have followed some general advice from a tech who told me to set it to 450 degC and put the BGA head down just about on the PCB totally covering the BGA for no more than 20 seconds.

I'm noticing that even the outer row of pads (All I can examine) are not shiny. They've reflowed beautifully, but all the flux has not burned off. I notice that paste pads that have been under the hot air long enough go completely shiny, and many other pads never quite burn off the flux (I assume) and remain somewhat dull. Is this fine, or does that indicate it didn't reflow properly?

How well does the air from these hot air BGA heads penetrate to the inner pads of a medium sized (14mm square) BGA? Should I bump up to my maximum 480 degC and/or heat it longer... ?

After noticing the dull pads on my first board, I attempted to reflow it for a longer period (40 sec) and pretty much burned the board. So, I know how far NOT to take it.

Any advice on the actual reflow of these?

EDIT: I forgot to mention I am using an IR preheating unit. (I actually use the hot air gun with the board on the preheating unit) I leave the preheating unit at about 250 degC for a couple minutes before starting.

I am seeing some videos online showing the use of a hot air gun and sort of following a real reflow oven profile at lower temperatures. Should I be doing that? I was thinking the hot air was not going to be as efficient and needed to be somewhat higher.

EDIT 2: I've tried following a profile used on a Youtube video for reflowing an XBox chipset, and even the paste on the surface of the board doesn't reflow. I guess if I really want to do this right I need to get some thermocouples and measure this directly. I re-ran the Youtube profile with 460C instead of 360C, and it looks good visually. I'll complete that board and see if it works.

EDIT 3: It works! An i.MX283 ARM processor running full Linux @ 454Mhz. 0.5-1 watt w/Ethernet. I'll get these 6 BGAs (3 boards, 2 BGAs each) X-rayed and see how I did.