More HD Lens Tests
 

I've just shot a couple more lens comparisons, this time with news print (which seems to be a preferred method for some).
One of the lenses is a Fujinon HA22x7.8, the other a Canon J14x8.5, both at 40mm, shot first at f4 then at f8.
Thoughts, guesses?!
Steve

No idea which is which, but I'd say lens 1 looks sharper.

I see little difference! Lens 1 may be slightly sharper.

The J14 isn't exactly new either! How old is it? I wonder how a J17 or my A18 would compare.

Lens 1 is the SD lens. On the landscape shots I did there was a - fairly - noticeable difference but as you both say there's little to choose between them here.
Steve

From that test and the limitations of the monitor I am looking at these, there is certainly not a lot in it -

Lens 1 does look like it has the edge, but in a moving presentation there would be no perceivable difference.

Come on Steve.... I can't wait to hear the details....

Regard Stu:
www.studioscotland.com

Beat me to it....

In your opinion, what sort of money are you talking about, for a high quality general purpose decent wide to telephoto SD lens that will give you the best results before you need to move into the HD glass range?

Regard Stu:
www.studioscotland.com

Well, that's the question isn't it! According to this test it's £500 for a secondhand old SD lens. But as you say, perhaps in other situations the difference would show.
I've no experience with the "cheaper" HD lenses, there's a Canon KH20 I think that's about £4k. I've actually only used a few HD lenses, mostly because the rental houses, BBC etc. tend to just use the same ones, so HJ11x4.7, HJ18x28 and HJ40, plus the Fujinon I have here and that's about all I ever get sent out with really.
The more input we can have from users the better I think.
Steve

The cheap Canon 2/3" HD lens is the KJ20x8.5. I've used one and it really is nothing special. There is no extender, but at 20x it offers a good zoom range. I found it to be very sharp in the center, falling off a little at the edges. I would rather buy a £4k low cost HD lens than take a gamble on a £4k "good" SD lens.

I'm not sure you'd be wise to do that Alister. Aren't some of these "budget" HD lenses designed to work in conjunction with these CAC functions that are appearing now - basically allowing a crappy lens design to be fixed at shooting rather than editing stage. I wou;dn't be at all surpised to find a fairly recent top range SD lens being better than one of these low-end HD lenses.
Did you find edge sharpness falling off at all apertures or just wide-ish? No 2x extender sounds like a drawback, but to be honest if you're looking at top quality results (which surely is what HD is all about) then you shouldn't be using a 2x anyway. Must say I tried 2x on the F350 on a Canon HJ lens and it really didn't look good - but I assume a lot of this was because it was also going through the 1/2" - 2/3" adapter which also has an optic in it.
Steve

Like most lenses the KJ20 worked best between F4 and F8. CA was reasonably controlled, there was some and it was worse towards the edges, certainly as with any lens it will benefit from electronic CA reduction, which the 700 should get in the next firmware release.

I've used top end HD zoom lenses on many cameras, including the 700 and they have all exhibited CA, every single one of them. Even primes exhibit some CA (although a lot less). Some of the CA is caused by the prism in the optical block.

The problem is finding that good SD lens. In every HD/SD lens test I have done the HD lenses have been better than SD lenses. It has never been by much and in most real world situations the difference would often go un-noticed, but they have always been better.

2/3" lenses on 1/2" cameras never work well. The smaller photo sites and shorter prism paths of the 1/2" cameras are not well matched to 2/3" lenses. 2x extenders and HD are not a good combination with any lens or camera, but a 2/3" lens on a 1/2" camera has to be the worst possible combination.

Buying used lenses is a lottery. It only takes a knock on the side of a lens barrel to put the collimation out, this can lead to focus issues that may not immediately be obvious and may only show up at certain focal lengths. I stand by what I said, if I was going to spend £4k I would buy the new HD lens and not take a gamble on an unknown SD lens. Having said that if I found a really good SD lens for £1k then I would quite possibly buy that and then hire in lenses as required.

Alister, you have a good point, but the HD lens used in the comparison was a ~$20,000 lens, I was shocked to see how little a difference there was.

I would think this story would change if the lenses were wide open or f2, but maybe not.

But still, for the price to be so far apart for generally the same perception is a nod to going the SD lens route.

And it's a fairly old SD lens too, you might expect better from a later generation (I think it's about 15 years old?) Also I did think about wide open performance, that's why I shot at different apertures, the f4 shot was nearer f3.5 so as wide as I'd tend to want to shoot, as to be honest no video lenses apart from Digiprimes and the like are any good wide open.
And I don't think it's down to taking "a gamble on an unknown SD lens" either, as I think the ranges of SD lenses are pretty well known (there was always budget like the YJ, and industry standard like the J series), and the SD lenses I have here both perform about the same (the other is an even older Fujinon 8.5x5.5, good but flare-prone).
I don't know about any of you engineer-types, but I assume that viewing a 100% crop from a still image is going to tell you as much or more than viewing even on a 50" HD TV, so that these tests do mean something?
Steve

OK, here's a couple more. It's funny, but I can definitely see more of difference on the wide, landscape shots. It's still pretty damn close though.
Which is which this time?
Steve

I would guess lens #2 is the HD lens.

In researching the problems I've had with CA in some of my images, I learned that the HIGHER the resolving power a lens has, the more CA becomes an issue. It's a real bear for lens designers to tackle.

So IMHO, CA correction is not something to overcome the handicap of 'crappy' lenses.

-gb-

My guess is that lens 1 is the HD lens. Lens 2 is very soft at the edges and in the corners. Bottom left is positively blurred!

It is very difficult to tell with that image as it is pretty low contrast with lots of haze acting as a natural diffuser.

Further to my post above, I still think lens 1 is the HD lens. I've spent a bit of time looking at the images and the difference is so small that for a while I did start to wonder if Steve was pulling a fast one by posting the same image twice :). However what I see is Lens 1 is certainly sharper at the edges, while lens 2 is soft at the edges getting shaper towards the center where it appears to have very similar sharpness to lens 1.

What I did notice when I got both images up side by side on a big monitor was that lens 1 appears to have quite a bit more contrast than lens 2. The exposure looks similar so I think this is actually the lens. Perhaps lens 2 is an older lens and the coatings are not as fresh or the optics are a little dirty. Also the colours are slightly different between the lenses.

I saw similar differences when comparing 1/2" HD and SD lenses on my F350. The HD lenses had better contrast and the SD lenses tended to be soft at the edges.

Thanks for posting these Steve, they are remarkably similar. If you get a chance on a bright sunny day I'de love to see a similar comparison.

I agree with Greg that CA correction is not something to be passed off as a gimmick or trick, but a useful tool to help improve image quality. There is a whole bunch of processing going on in 3 chip video cameras already to help reduce flare and other optical effects created by the prism block, CAC is just a new tool to help achieve a realistic image. Of course it's only a good tool if it's operation is in effect invisible.

Shall be checking back to hear which is which.

Spot on on all counts Alister, I'm very impressed!
I've always noticed that HD lenses seem to win in the contrast stakes on all comparisons I've done (including vs. say Canon 300mm prime), and my cynical side ha always wondered if this is boosted on purpose to increase apparent sharpness to make it HDish?
As you say, they are very close, and if you're nit-picking on 100% crops side-by-side I'd say it's definitely not worth spending the extra £15k!
Steve

Great post.

I purposefully made a quick decision to get a viewer's feel and it shows to me what Steve said about the need to spend the extra money.

If contrast is the main difference then that is not that big of an issue as this can be addressed with camera settings and or software in post.

I thought the higher resolution chips of the PDW-700 would show the weakness of an SD lens more so than a camera like the HPX-500.

Makes the whole game look needlessly expensive.

Bumper load SD vs HD lens tests!
 

Right, here are some more tests shot on PDW700.
"5" is an old Fujinon 8.5x5.5
"90" is a Tamron 90 macro
"300" is a Canon 300mm f2.8
"HA" is a Fujinon HD lens
"HA90" is the Fujinon HD lens at 90mm
"HA300" is the FUjinon HD lens at 300mm
"J14" is an old Canon J14

Haven't yet had a good look, be interesting to see what you think.

Steve

Here's my quick analysis:

Looks to me on the wide shot that the 5.5 Fujinon is actually the best, then the HD lens, then the J14. The 300 prime looks a lot better than HD lens (BUT the HD lens did have to have the 2x extender on to get to 300, but aperture was f5.6 or so, so not as though it was wide open, and if it's made with an extender then it should bloody work!). The 90 macro looks pretty similar to the HD lens to me, tricky one to judge because being a CU the depth of field comes into play and bit of one look sharper than the other, and then vice versa on another part of the image.

All these were mid apertures-ish, between f5.6-f8 ish.
Hmmm!

Steve

It would be better to test wide open or even f2.8, any differences will start to show through clearly. At f5.6 to f8 even poorer lenses tend to show up well on tests.

When I did my previous tests I did some at f3.5ish, and same at f8ish, and the differences showed in both (ie the poorer lens looked poorer at both apertures and by the same amount). Personally I try to avoid wide open on any lens (except a Leica 105-280 I had a few years back that was sensational wide open), as they all suffer unless you stop down about 1 stop or more, and then crappy again of couse beyond f11ish.
Steve

F2.8 is a pretty commonly used stop, you've got that one stop improvement on most zooms. Although, I do like to have a lens that you can actually use wide open, it's handy in low light situations or if you want a shallower DOF.

Contrast is very important. While you can tweak the camera setting to make the image look more contrasty, if the shadow detail has been lost because the lens is bleeding light into the dark areas you will never get it back.

It could be that in this and the other tests I have seen and done that the comparison has been between fresh HD lenses and older SD lenses. Over time the oils and lubricants used in lenses evaporates and some of the vapor ends up on the lens elements causing some fogging and loss of contrast. Or it could be that the HD lenses are using better coatings or better glass and thus can produce more contrast.

Fogging like this is one of the reasons why I don't like spending lots of money on used lenses.

As you say Steve it is not a £15k difference.

You just won't get the performance wide open, all video lenses tend to be soft wide open, some dreadfully so. Also, I think what I and others are looking for here is not whether an SD lens can compete with an HD lens at the extremes and under worst-case conditions, but whether for a fraction of the cost you can get an image of very similar quality when the lens is used sensibly (ie medium apertures, no extender).
Looking again, that Fuji 8.5x5.5 does look a lot better than the HD Fujinon. Same for the 300 Canon (though again not 100% fair as the HD lens had extender on).

Steve

The SD wide angle zooms are excellent, so I wouldn't be too surprised that they hold up well against a HD standard zoom.

Medium apertures are OK for day exteriors, but for interiors and night exteriors the wider stops are commonly used, not only on docs, but also on drama work. It's how well the zooms actually do hold up at these wide stops, especially under extreme lighting conditions that's important, using only the medium apertures is extremely limiting creatively. I'd push the lenses further than shooting the day exteriors.

I'd also check for CA, which is an issue even on the HD lenses.

In general CA seems to be less of an issue with SD lenses, and certainly is with this wide angle vs the HD lens.
If I get time I'll try to do a couple of wide open tests, but I expect them to have the same relative performance, and for all to look a bit crap! When you're talking about dramas, using wide apertures really does require primes, that's why they've always been favoured for serious creative work, and still are.
Steve

OK, here's a couple of print, with the 5.5 SD lens and the HD 22x Fujinon, both wide open. Close I'd say, but DoF again is a problem.
Steve

Dramas commonly use zooms, especially the Wide Angles, because of the breathing issue with the other zooms and they tend to be better quality. Of course, primes are also used, but zooms tend to be the workhorses on many dramas, especially for TV. There are some high quality HD zooms intended for drama type work.

:: CARL ZEISS DIGIPRIME & DIGIZOOM LENSES ::
Cooke S4 HD Zoom
Plus the Panavision zooms.

However, the Canon or Fuji are more likely on quite a few productions. Often using f2.8 because stopping down does improve the performance, but you've got a shallower DOF which is desirable (if not a 35mm DOF). It also keeps the heat from the lighting down and has lower power requirements.

Perhaps the test really says that if you want to shoot HD using wide apertures you need better lenses than these particular zooms.

BTW Cooke had a SD zoom video lens for 2/3", an impressive performer.

Thats' the thing though, Brian, the sort of zooms that do have a good rep are going to be seriously expensive even by HD lens standards. The first HD lens I ever used was a Canon CineStyle, and I must admit it was excellent, but even with that I'm sure I remember that it wanted to be stopped down about a stop for optimal performance.
What do you think to the wide open tests I've just done?
Steve

Assuming the same f stop, my first thought is that the 5jpeg doesn't seem to have the transmission losses of the ha jpeg lens. The 5jpeg seems to sharper at the edges and perhaps a slight edge in the centre.

Yes, stopping down the Canon a stop does really improve the performance.