Pixel peeping is such a fallacy, because close inspection of footage on a computer does DSLRs such an injustice. It is not how video is meant to be viewed.
It’s misleading because often people view still frame grabs of video on forums, but still frames change the characteristic of noise, how it fizzes over the image and becomes more or less noticeable with motion. Resolution changes as well, because of the spatial characteristics of footage when it’s in motion. Also, 1:1 crops of a 1080p image magnify any flaws in the footage.
Add to that, web compression is a big problem. On Vimeo your clip is downscaled to 720p and reduced to something like 4Mbit. On many blogs and forums, web compression of JPEGs smudge fine detail out and again change the characteristics of noise and banding even more. So it really does do footage a disservice.
It doesn’t stop there. Most laptop screens have TFT panels which are very poor at colour reproduction. They have improved over the last few years but footage looks so much better on a proper display, even a mid-priced Plasma screen. Because for one, you are not sat a few inches away eyeballing the screen!
Contrast is also a weak point of computer flat panel displays, and again they have improved but they still don’t deliver true inky blacks.
Stills guys on DPReview.com and places like that make the same mistake. They critique photos based on scientific testing, viewing photos under a microscope when they’re supposed to be viewed printed and hung on a wall where there whole characteristic changes.
This is why I always try to go off feeling and unscientific hands-on experiences rather than test charts.
Yet in some ways pixel peeping does have a valid purpose and that’s to separate one camera from another. It gives you an idea ultimately about what the camera is producing under the surface.
Above is a short Panasonic GH2 versus hacked GH1 and Canon 60D clip. It’s quite revealing just how good the GH2 is in terms of natural colour and detail – however you won’t be able to tell from playing it above, even full screen.
Vimeo’s compression is very high unfortunately and it’s default setting is to playback videos at a downscaled 720p. It reduces the bitrate dramatically in order to stream the clip as quickly as possible, even on slower computers – for the masses. Downloading the original file is an option but that takes longer and it is not instant gratification – also, you need to be a Plus subscriber to do so.
Obviously it is harder to judge the merits of a camera when the footage is so highly compressed.
The other flip-side of Vimeo’s compression is that sometimes it even manages to hide the shortcomings of DSLRs. I recently tried to demonstrate the difference a good noise reduction plug-in makes to high ISO footage, but on Vimeo I can hardly tell the difference! To my eye in the NLE on a good screen, it was light & day.
Web compression removes so much detail that even noise at ISO 3200 is less noticeable. One guy comments on my video here that he didn’t know what the split screen was showing, even though one side is a 5D Mark II at ISO 3200 and the other side was the same footage with heavy noise reduction applied with Neat Video. One side has a green mask of noise excitably bubbling away, but on Vimeo even that isn’t as noticeable when it’s downscaled and compressed so heavily.
Try playing back the part of the video which shows the GH1 at ISO 1600 in Vimeo compared to the original downloaded file – you will notice a huge difference. That kind of noise and banding is also a lot more visible under motion than it is from a still frame.
Unfortunately these are the realities we currently face when it comes judging the merits of images via the medium of the web. Until bandwidth and processing power increases for the vast majority, there is not a lot Vimeo can do about it!
Switching the clip in Vimeo to playback in 1080p helps a lot, but it is still highly compressed and 1080p clips rarely playback smoothly slower machines, so it wouldn’t be fair on a good few people if I changed all my clips to stream in 1080p.
YouTube 1080p clips seem to playback better thanks to less CPU intensive compression, but YouTube is plagued by a barrage of advertising and an absolutely terrible trollish community.
As for Exposure Room, I have never been a member but it would be interesting to gauge the opinion of those who use it.
A lot of the criticisms people make about footage is all down to pixel peeping it in highly compressed form on a laptop screen. It doesn’t compare – just doesn’t – to having it projected or watching it back on a large HD TV in glorious high bitrates.
If you have never watched DSLR footage on the big screen before or projected, seek out the opportunity as best you can and I promise you, you will be amazed.
