Prores Faceoff
Prores Faceoff
Just how lossy is Prores LT compared to Prores HQ? On the surface level, the images look almost indistinguishable to the human eye. Here’s two. Click the images to get the high-res if you don’t believe me.
With the help of Chris Noth, we’ll see how degraded these images become after extracting several keys that one might be asked to do in a color grading session. Sorry Chris, but it’s in the interest of science.
I took the same Canon 7D clip and converted it to both HQ and LT from the “raw” H264 the camera natively shoots to. I pulled a key on the HQ footage and then applied that key to the LT footage to compare how each image fared. Obviously, the HQ would give us better results as it runs at a higher bitrate, but I wanted to see how different the LT would be.
I first selected Chris’ stylish leather jacket. I didn’t take any time refining this key, I just wanted to see what would happen when I pulled some of the data with Resolve’s color picker. I knew this one would be pretty dirty and inexact as there’s a lot of black in the picture already. Sure enough, his black hair and his dark shirt went with it. Click the images to see the full grunge.
Next I went for a skin tone extraction, something I do many times on pretty much every job. If this wasn’t a test I would refine the key and then blur it, and in this case likely vignette and track his face and hands if I didn’t want to affect the bricks.
Here’s a closer view of both Chris’ face and the floor, the latter of which you’ll particularly notice the blockiness of the LT compression.
I liken the Canon workflow as being similar to dubbing a VHS to a DVD; that is, the DVD won’t make the VHS image quality look better, but there is no reason not to use the highest bitrate when encoding that DVD, as loss can still be introduced, even if your starting image is shot on a compressed format to begin with.
With a DSLR workflow, an assistant typically converts once to a flavor of Prores at the start of the job and that stands as the final Quicktime format for grading. Converting to LT saves hard drive space but makes my job harder when it’s time to really delve into the image past primary corrections.
I was surprised that, under the hood, the images from both codecs would be so different since they looked identical when converted from the original. Is Prores LT a total dealbreaker? Not necessarily. I could probably refine the skintone key in LT and get it looking pretty good, especially when blurring to include more values that the key didn’t get, but why not work with the best quality possible? Hard drives are cheap.










