Keith's waterfall is shown with the default settings in REW with the exception of the time range extended to 450ms (from 300ms):
The noise floor has traditionally been approx. 40dB with "legacy" REW kits but we've found with the USB mics that people are seeing anywhere from 50-55dB in a "quiet" room. For this reason, we've tried to increase the dB of the measurements to at least 30dB above the noise floor (so closer to 85dB measurements) as well as adjust the noise floor up to the minimum threshold reported by USB mic users (50 dB). As a compromise (and knowing Keith, also to dispel any concerns about an even better looking waterfall being the result of using a higher noise floor) when presenting his graphs, Keith has elected to show down to 45dB instead of what he actually measures (which I believe is closer to 50dB with his UMM-6). This gives the appearance of presenting his waterfall in a more conservative light than using 50dB as the noise floor. My last RTA actually showed closer to 56 dB as my noise floor!
The only other presentation difference I can think of could be related to the width of the image capture or the preview size used to display it in the forum. I always use 900px as the capture width (I think 871 is the default) and I preview my waterfalls with the 500x1000px large option. If you instead use a url address to present photos (instead of uploading them directly) then I'm not sure if you're presented with the same preview sizes or not which might explain why some users have larger previews than others.
EDIT: Remember that Keith's graph is unsmoothed but after EQ and his room is heavily treated so you'd really want to see his response/waterfall prior to any room treatments or EQ to see the benefits each is having on his response and decay times. It actually looks to me to be more like 83-92 dB range (peak between 55-60Hz and null around 250Hz?) from 15-300Hz for a +/- closer to 4.5.
Edited by jkasanic - 10/15/13 at 11:18am