What is this softness? (and can we trust what we see on the web?)

P

plnelson

Member
I've been considering the XA20/25 for a couple of projects. It has the advantage of being really cheap and also very small and light with pretty good IS so you can shoot handheld without a tripod or some sort of rig. It has the disadvantage of being really cheap so I'm trying to decide if I can live with it.

I've been looking at a lot of XA20/25 sample video on the web. Here is a video made by our very own Aar5 and posted on Vimeo to test/demo the XA25.
http://vimeo.com/76918863

During the first 30 seconds or so he's panning through some autumn trees and, to me it looks really really soft. So I'm trying to understand what I'm seeing. If I freeze it around 00:29 I notice two things:
1. It really is soft - the edges of the leaves and trunks are not very sharp.
2. The colors are very blocky - if I look at the trunks of the trees I see mostly one color and then an abrupt switch to another color as though there just aren't enough bits of precision to represent the desired values.

How many of these problems are an inevitable result of the XA25? For example the softness: It has a 20X zoom. Making a 20X zoom, especially at a bargain-basement price involves a lot of compromise. Are we seeing limitations of the lens here? Or are we seeing motion-blur because he's shooting a shaded subject while panning with a ND filter on, so this necessitates a longish shutter speed? Or what?

And the blockiness? Is this a compression artifact? If he had used more bandwidth or a different encoding would it be less? Or is it "baked in" to the XA's codec, or a limitation of the number of bits per sample of the XA20's sensor? Or is it a Vimeo artifact? Or what?

Finally, I tried viewing the same Vimeo video on Chrome and Firefox and there's a noticeable difference - the blockiness is not as bad on Firefox, but it's even softer. So when we're evaluating test videos on the web can we even trust what we're seeing and if not, what should we do?

Thanks in advance!
 
You haven't said what your projects are, but in general you get what you pay for with diminishing returns. If you spend twice as much on an XF300 or C100 you'll possibly get better pictures, but nowhere near twice as good. At the price point, I don't think you can beat the XA20/XA25/G30. The softness in the example to me looks mainly like soft focus, which is what you'd demand from using an ND filter, i.e. it's deliberate. In any case, you can't tell how good a lens is if you stick a cheap ND filter in front of it. Nor can you tell about blockiness form a video that's been recompressed by Vime/YouTube. The 28Mbps these cameras can shoot at is already overkill, there won't be any blockiness in the original footage. I don't think anyone buying this camera has experienced the slightest buyer's remorse (here's some of my footage from this camera ).
 
Rainer said:
The softness in the example to me looks mainly like soft focus, which is what you'd demand from using an ND filter, i.e. it's deliberate.
Click to expand...

I can't parse this at all - can you elaborate?

The purpose of an ND filter is to allow you to shoot with a wider aperture in order to have a shallower depth of field. A neutral density filter would have to be astoundingly bad to noticably soften an image. Now, of course many lenses perform more poorly wide open, and a 20X zoom is going to be full of compromises but it's still hard to believe the Canon lens is that bad wide open, so I'm skeptical that that's the explanation.
 
plnelson said:
I can't parse this at all - can you elaborate?
Click to expand...
Yes, that's what the ND is for. So that you can use a wide aperture in daylight. Wide aperture means shallow dof. Shallow dof means only one plane of the image is in focus. It's impossible for closer up foliage to all be in focus on a 1/3" sensor at widest aperture. Overall, it's all going to look soft except for the one twig in focus. Add any distortion from the two cheap plane of glass filter and the motion, and you have no idea whatsoever what the lens is like.
 
My guess is the softness is a combination of the vari-ND filter, the method of creating the slo-motion as well as poor compression technique (or low bitrate when uploaded to Vimeo). Basically it has a lot going against it. If you took all of that away and just filmed viewing in normal motion with the camera alone, I bet it would look better.
 
My Canon hv20 (or even the Sony HC1) from years back shoots better stuff than what I just saw. I have absolutely no idea how anyone can destroy an image like that but somehow they do.
 
The XA-20 is capable of taking fairly good footage,i have the similar HF-G30,online footage often looks inferior to the origional.
 
You must log in or register to reply here.
Back
Top