I made it to an observating location near Driggs, Idaho, specifically at 43deg 39' 42.3" N and 111deg 10' 32.6" W. Skies were essentially cloudless, except for thin clouds on the distant southern horizon around 75 miles distant. There was light-moderate smoke (maybe 15-20% opacity / optical depth). Some thicker smoke (mixed in with clouds) on the distant northern horizon around 75 miles away. I think my strategy paid off of doing observations 100% naked eye, without distractions from taking pictures and such. To help illustrate what I saw, below are visually realistic panoramic views of the eclipse. Ray-tracing is used and things like atmospheric scattering are accounted for, allowing us to see the moon shadow's appearance in the atmosphere and on the ground.
All-sky simulation of solar eclipse (click above image for larger view) looking up at the sky hemisphere from the ground, during the partial phases. South is in the center and north is on the edges. West is left of center and east is right of center. The location is SW of Driggs Idaho where I was observing so we see my view of the hills and mountains on the horizon. The closer hills in the west range up to about 6 degrees eleveation angle. In the east Grand Teton spikes up above the nearer foothills. About 7 minutes before totality (2nd contact) I stopped my viewing of the crescent with the eclipse glasses to gaze around the landscape.
As totality approached I could see the moon's shadow start to become evident in the sky, though the mountains to the west prevented me from seeing good color underneath the approaching shadow as 2nd contact approached. What I noticed most instead was looking above my own shadow at the NNW horizon. Here there was a noticeable brightness gradient darker to the left and brighter to the right of my shadow. This was a signature of the approaching moon shadow, though the distant smoke to the NE complicated the signature a bit. The above image shows this view as it is centered on the antisolar point above where my shadow on the ground is. We can see near the horizon (just above the hilltops) a subtle sky gradient darker to the left of center vs brighter to the right, a signature of the approaching shadow.
Between about 5 and 2 minutes before totality things seemed eerily frozen in time with hardly any noticeable changes in the brightness of the sky or landscape. The shadow gradient persisted in the NNW and seemed like it was in a steady state. At around 2 minutes before 2nd contact I could see some diffuse shadow bands showing up on a letter sized sheet of white paper I had set on the ground. After thinking that things would stay in this frozen state for some time I could suddenly see the light levels darken dramatically. This is consistent with calculations that have the greatest proportional rate of darkening about 20 seconds before 2nd contact.
More on the Earthly side I could see dragon flies swarming around me just above the barley field and gnats congregating a few feet off the ground, as might usually be noticed at sunset. I later heard a report of birds congretating at the Teton River about 1/2 mile to my east. As the light levels plunged the shadow bands became quite a bit sharper. As mentioned the greatest proportional rate of (zenithal sky as measured in the simulation) darkening actually happens about 20 seconds before 2nd contact. I kept looking at the landscape as I wasn't sure exactly when it would be safe to look at the sun at 2nd contact (in retrospect the time to look would be when shadows cease to become visible, since unlike the diamond ring the corona is insufficiently bright to cast much of a shadow). Then I glanced up high to the SE to behold the corona in totality.
In this frame at mid-totality (click above image for larger view) we see varying amounts of yellow or orange along the horizon, depending on the varying angular height of the topography at different azimuths. This matches my visual view (looking away from the corona to scan the 360 degree horizon) as I saw the best reddish colors to the north and more yellow to the south (center) with some high clouds also appearing relatively white. There was less color east and west.
I was reinforced that photos fail to fully capture seeing the eclipse, and I can see some ideas to try to improve upon my 1991 processed version of D di C and Gary Emerson's photos. I was impressed to see the inner corona looked more like the composite brightness view I had made without the enhancement of details. In other words, the streamers naked eye were much more visible in the middle to outer corona than in the inner, and it was neat just to see the overall brightness gradient between outer, middle, and inner. Streamers could be seen rather far away, especially to the SE, surrounded by the more uniform F corona. Elongation wasn't seen this time around as the outer corona becomes the zodiacal light (they are one and the same). Possibly the sky was brighter this time compared with 1991 when I believe I did see this. The simulation for the 2017 eclipse also doesn't show the elongated F corona being bright enough to see against the sky background. The smoke in the sky may have played a role as well.
There was extra brightness in the inner corona due to the small moon too. Some brightenings had looked like prominences and the chromosphere, though I think I didn't actually see them based on the location and lack of color. Maybe I should try sunglasses next time? This was actually my first good naked eye view of the diamond ring, at 3rd contact, so much a brighter pinpoint than even the bright corona. It was suprisingly higher than I expected between 1 and 2'oclock position on the moon's limb. Here is a time lapse summary of the 6 minutes surrounding totality sped up by a factor of 20. North is in the center and we can see how the moon's shadow produces changing gradients in sky brightness.