Liabe Brave
Member
Each time I do counts I check multiple edges, on multiple images, before I write up the results. I only put one or two in each post because including all of them would make the thread unreadably repetitive.
Of course, but JPEG compression doesn't actually ruin edges very fast (it hits color information before luminosity). The mosquito noise it introduces will interfere with counting of subpixel elements since the noise is on the same scale, but solid geometry is more robust. For example, here's an uncompressed camera PNG, followed by a JPEG that is almost ten times smaller.
If you zoom in the compression is definitely visible, but it doesn't much affect contrasted edges. Only when you get to about 15-20x JPEG compression does the disruption overwhelm counting.
But even barring that, my Ryse counts have never actually been of JPEGs. Earlier posts did include JPEG versions because the image host Imgur automatically converts large files. (I'm now using another image host and my newer posts contain PNGs.) But the counts I did always used the original PNGs. Here's one, in case you want to check my work.
You're right that the upscaling isn't causing the shimmer...but it's making it worse. A game that requires upscale can fill fewer pixels to represent tiny elements, and this lower accuracy is magnified by upscaling. For example, here's Battlefield 4 on PS4 and One. (Ryse runs at the res of the PS4 version, so won't look as bad as the One version; this is just to show that lower resolution results in more artifacts.)
The strong bokeh in Ryse--though pretty--actually increases the shimmer as well, since it turns distant small highlights into larger iris shapes.
Comparison of what? The story trailer cutscenes I'm uncertain about aren't in any of the recent gameplay videos, including the palace one. Nevertheless, despite having no second objective measure available, there are strong indicators of widely variable quality across the two videos. For example, do these two shots really strike you as having precisely the same rendering techniques and resolution as each other?
Keep in mind that compression isn't doing anything here. The (confirmed 900p) lower shot, which appears far less sharp and detailed to me, comes from a video with more than triple the bitrate of the upper shot!
But you don't even have to compare videos to see differences. The story trailer alone contains scenes with very different quality. For example, here's two crops of Vitallion's armor from separate points in the story trailer. Note that I didn't zoom or affect either one of these; these are native-size crops.
To my eye, this comparison strongly suggests multiple rendering resolutions.
