ZEISS Light Weight Zoom LWZ-1 15.5 – 45 mm T2.6 – CA and lp/mm test

[Andrew M.]

ZEISS Light Weight Zoom LWZ-1 15.5 – 45 mm T2.6 – CA and lp/mm test

Brand new from the box. Weights only 2kg (4.4lb)
Very smooth focus going almost 360 deg and very steady focus while zooming.
Well, see yourself, CA unacceptable for 4K, big disappointment here.
It is equal to 3 Misterium pixels. Good for 2K though.
However I can take this CA all out in the single frame using software.
If we could only get CA cancellation option in REDCINE.

Very sharp lenses though, with linear behavior from 15.5 to 40mm
Changing the aperture from 2.6 to 5.6 and 11 didn’t make much difference as CA and lp/mm is concern. Good linearity in all aspects.

Bar patterns shown here are the E ones, just next to the S35 vertical line.
First picture of bar patterns is done at 2.6T and 15.5mm
The second at 5.6 and 15.5mm
40mm setting was giving me similar results.

I was expecting much better performance here, well it is light weight alright, in the pocket as well.

Andrew

[chuck colburn]

I wouldn't be too quick to blame that fringing (all or part of it) strictly on the lens. A light source on any type of projector can influence the quality of the projected image. I know that sounds wierd but I learned this to be true. When I ran the tech opps. for the UCLA film school the projectionist in the Melnitz Hall theater (always considered one of the best screening facilites in L.A.) came to me with a complaint of softness and color fringing on one of his projectors. Now one would think it was either a mechanical alignment error in the gate or a optical problem in the lens, but as it turned out they had recently upgraded the lamp housing to a new and higher powerd xenon unit. So I dicked around with the reflector in the lamp housing and was able to eliminate both problems. Then I made a mechanical bore scope that could be mounted in the gate which allows him to sight in the the center of the reflector to the film gate. After that all he had to do when changing bulbs was twiddle the reflector tilt screws while running some SMPTE projector alignment test film. So I'm not saying that there is a problem with that particular thru the lens projector you used, but it's always good to eliminate possible problems. Hence there is no subsitution for a "film" test as the final word.

Which line on the image of the full test target is the S35 one? Because if it is the one on the outside I would say the lens is doing quite well.

[Andrew M.]

Chuck, look on my previous post, I was testing 65mm Master Prime on the same projector the same area near the S35 frame edge. (Second to the outside last)
http://www.reduser.net/forum/showthread.php?t=1475
This is the best lens so far I could find out of many.

I wish this was the problem with this zoom, I was counting on the near 65mm performance since it is very new design.
I guess they went for the weight this time.

I am using nothing else but one of the best projectors, ARRI recommended.
One thing that I was pleasantly surprised is the wide range of the same performance regardless of aperture or zoom setting.
But isn’t it what zoom supposed to be best in? but not for 4k material acquisition not this time. Will stick to 65mm for now.

[Bruce Allen]

Andrew, thanks for the great review. Be aware you can decrease CA in video using most post software (After Effects, Shake, etc)... just split out the r,g,b channels using a channel shift filter, apply different lens distortion amounts to the different channels, then re-combine. You can also add it back too - I do that sometimes to make CG rendered stuff look more real ;)

Bruce Allen
www.boacinema.com

[brickwall]

Excellent test, Andrew. Your result with the 65mm MP was also a real eye-opener for me. I'm curious, who makes the projector you are using, and where can you get one? I need one of those for some projects I am working on.

Brian
(lens designer near Richmond, VA)

[Sanjin Jukic]

Could be a product from

http://www.goetsch.ch/

[brickwall]

Could be a product from

http://www.goetsch.ch/

Thanks, Sanjin

Goetsch appears to make surveying equipment, but I'll ask my German friend if he can find an optical projector on that site.

In the past I've used Pearl projectors from Japan when I worked in Asia, but I can't find a U.S. importer for these.

Brian

[Sanjin Jukic]

Also I could not find it on the website but they could do a custom measuring
part that can be added to some projectors for the lens tests.
It's a Swiss Made product and all is possible.

[Andrew M.]

Excellent test, Andrew. Your result with the 65mm MP was also a real eye-opener for me. I'm curious, who makes the projector you are using, and where can you get one? I need one of those for some projects I am working on.

Brian
(lens designer near Richmond, VA)
Brian, projector is made by Götschmann in München, it is G67P
http://www.goetschmann.de/products/g67p.htm
Sorry English version here:
http://www.robertwhite.co.uk/product.asp?P_ID=1691&PT_ID=335

I was told that one of the new Optimos is better then this Zeiss one.
Maybe I should test it next.
I am not a guy that will count all Misterium pixels, if all are there after shooting the frame but somehow I can't figure it out how you can play symphonic orchestra music through the rail station speakers.

Also Misterium anti-aliasing filter brings this 5 micron pixel spacing to more like 10+ micron or so (Graeme correct me if I am wrong) thus, the true 4K acquisition is still posing many challenges as you all see.

So for now we are playing 96kHz CD symphony piece with Treble knob turned all the way down and we are trying to feed it in to the Grandma's speakers.

Still better then playing music from the tape recorder, guys do you remember those ones:-)

[T. Glen Phelps]

Brand new from the box.

If I may be so bold, were did you get it and approximately how much did you pay for it?

[Andrew M.]

I rented it:-)

[Andrew M.]

Andrew, thanks for the great review. Be aware you can decrease CA in video using most post software (After Effects, Shake, etc)... just split out the r,g,b channels using a channel shift filter, apply different lens distortion amounts to the different channels, then re-combine. You can also add it back too - I do that sometimes to make CG rendered stuff look more real ;)

Bruce Allen
www.boacinema.com

Bruce, how long it takes for 1080p material to process it for CA in Shake or in After Effects?
Do both application support CA corrections?

[Sanjin Jukic]

Brian, projector is made by Götschmann in München, it is G67P
http://www.goetschmann.de/products/g67p.htm
Sorry English version here:
http://www.robertwhite.co.uk/product.asp?P_ID=1691&PT_ID=335

I was told that one of the new Optimos is better then this Zeiss one.
Maybe I should test it next.
I am not a guy that will count all Misterium pixels, if all are there after shooting the frame but somehow I can't figure it out how you can play symphonic orchestra music through the rail station speakers.

Also Misterium anti-aliasing filter brings this 5 micron pixel spacing to more like 10+ micron or so (Graeme correct me if I am wrong) thus, the true 4K acquisition is still posing many challenges as you all see.

So for now we are playing 96kHz CD symphony piece with Treble knob turned all the way down and we are trying to feed it in to the Grandma's speakers.

Still better then playing music from the tape recorder, guys do you remember those ones:-)

Thanks for the link and clearing up the "Goetsch" word root mess!

[Bruce Allen]

Bruce, how long it takes for 1080p material to process it for CA in Shake or in After Effects?

Pretty slow. Someone should write a GPU-optimized version. I'll benchmark when I have a chance.



Do both application support CA corrections?

Yes. Not automatic though - you have to tweak it by hand a bit. But it's pretty simple to do on any program as long as you have a lens distort / undistort filter (nothing fancy, just simple first-order barrel / pincushion distortion fixing will help a lot). You could probably do it straight in Avid or FCP if you found the right filter.

All you do is:
1. make 3 copies of your footage - one with only Red channel, one only Green, one only Blue
2. apply lens distort / undistort filter to R and B footage. Use different values for R and B (one is usually positive, the other negative) and tweak until it seems okaay.
3. combine the footage again.

You can also fix overall barrel / pincushion distortion at the same time - then you'd want to tweak the Green as well.

Someone should write this as a RedCine plugin. Also, would be a good plugin for finishing tools (scratch, lustre, etc, even Apple Color). I'm sure you can do it in smoke already because they have pretty much a Flame toolset.

Ideally you'd have a preset database so you could select the lens and f-stop used and it'd figure out the CA from that. Or it could read the lens data from the Cooke /i metadata. Also, you could very probably write an "auto-calibrate" function - the computer just tweaks settings until the R,G,B channels line up nicely.

It's not hard to do. I'd do it (I used to be a programmer) except coding sucks 'cause everyone pirates your stuff and complains when there are bugs on the obscure system they're using... also, not as much fun as directing and pays less than post-production for me.

On a modern computer, should be fast. Especially if you use GPU. Also, you could optimize it - what I described above is not efficient because you're wasting time distorting completely black channels (when tweaking the blue layer, there's no need to distort the R and G channels because they're totally black and you're only concerned about the B channel).

Going further down the line you could do the same thing to fix vignette. It's very easy - I just boost curves and mask it so that it the boost only changes the edges.

You could also add sharpening to try to correct for sharpness fall-off at the edges. And while you're at it, you could try to correct for the color casts of various lenses. You could do some kind of highlight bloom color fringe reduction but that's tricky.

All of this has been done in the DSLR world with programs such as DxO Optics though. Although they make it seem like frikkin' rocket science or something - not true! Lens distortion functions are widely published. Doing this seems kinda obvious IMHO.

Cheers

Bruce Allen
www.boacinema.com

[Andrew M.]

I agree, GPU support for this is a must to speed up the process.
Some kind of hardware like dedicated DSP, build in to the camera correcting on the spot would be perfect though.