Canon 7D- Real World Sharpness Test

S

shoqman

Member
Ok, so on the left here is a full res still, downscaled to 1080p, then compressed to high quality H264 to see what that throws into the mix.

On the right is footage straight from the 7D- sharpness set to +1. Post sharpening +0 footage looks even better, but I wanted to do this straight from the camera.

Enjoy.

-shoqman
 

Attachments

  • 7d_Compare_2.jpg
    7d_Compare_2.jpg
    213.5 KB · Views: 0
Thanks for posting. Here are some observations:

The foreground tree leaves and grass looks sharper on the video (right image) with more detail. On the left image , the grass looks blurry, much like inadequate bitrate, poor compression, or high deblock settings for h.264 (like GH1 "mud"). I would have predicted the image on the left should look much sharper if it was derived from a still, even if you used "high quality" h.264 settings. Would you mind posting the unaltered still source image? And what settings did you use to compress the still ?

But the overall sharpness changes for the background elements like the hedges and shrubbery near the house, which now appear sharper on the left image. There are entire horizontal panels missing from the right image (near the garage window) . I wonder why this would be?

At sharpness of only +1 for the video (right image) there already seems to be some faint ringing seen on the vertical poles. I'm not surprised by your comments that using a better quality sharpener in post looks better
 
Sure, here is the still, just downscaled, vs the footage.

I just think it's absurd to compare the 5k still directly to the footage. I think this test would compare quite favorably to actual video cameras.
 

Attachments

  • 7d_Compare_3.jpg
    7d_Compare_3.jpg
    217.9 KB · Views: 0
Thanks for the look dude. One thing, both image seem to be focused differently. If you look on the left picture the house, trees and hedges are basically more in focus than the flagpole(?) and branch right above it in the foreground.....while the picture on the right shows the hedges out of focus but the flagpole(?) and branch right above it are more in focus. Anyways...they are definitely comparable. So the in cam method of sizing down is actually doing a good job. As I mentioned here and in other threads all camera's stills look much sharper than their video footage. That's just a given.

Barry's test shows us how much of a difference there is from the RAW image to the video. Your test shows us we will get comparable results by sizing down ourselves in post (if we had that option) and that the camera is actually doing a decent job.
 
If the focus was different it was unintentional- I shot through the EOS Utility- one button press for video and one for a photo so I don't know how that happened.

This test is actually identical to Barry's, except for adding one notch of sharpness.
 
And then to knock off the deadness of digital let's grain the first one to match a common 35 mm cinema film stock; Kodak Vision 250D. (Done in After Effects; Film Grain plug-in, Intensity 0.75, Color Saturation 0.75, Shadow Application 0.2)

PS: We'll see how long my ISP lets me link this image

7d_Compare_2grain.jpg
 
shoqman said:
On the right is footage straight from the 7D- sharpness set to +1. Post sharpening +0 footage looks even better, but I wanted to do this straight from the camera.
Click to expand...

Please NEVER use the camera to sharpen. Set the sharpness as low as you can. Cameras suck at sharpening.
Look at the images below. By using incorrect sharpening settings I can give your still image the same yucky double-contoured look as your video image*

*If you're unsure what to look for, check out these:
-left part of the left pole
-right part of the right pole
-leaves at the top of the image

By using good sharpening settings, everything is fine. Of course it brought out the JPG artifacts, you must have used low quality settings when saving for the web.

 
Yeah if you read my original post it states that a better image can be obtained by leaving sharpening off and doing it in post, but as per Barry's test I wanted to see
what could be done in camera.
 
You must log in or register to reply here.
Back
Top