Underwater Bubble Blowin' Users Group

[Michael Hastings]

Zeiss, RED, Cooke, etc. vs. off the shelf Canon and Nikon lenses

Zeiss, RED, Cooke, etc. vs. off the shelf Canon and Nikon lenses

It's true that posters may be shot on larger film. But 266 pixels per inch is considered a minimum for magazine photos and of course the other elements are typically 600 to 1200 dpi and photo quality prints from even very cheap printers are much higher res. Consider that the mysterium image blown up to just 60 inches is only 75 pixels per inch - the difference is the likely viewing distance.

My sort of intuitive point here is to say that both SLR and cine people have many years of shooting to the same medium - 35mm film - and regardless of the direction of the frame and frame size - pros have always wanted/insisted on getting the most they can from that film - and are willing to pay for it. I think it is generally agreed that 2K clearly isn't as sharp as film and 4K is probably where you need to be to be comparable with film.

I would suggest to you that the price difference between a Canon EF 135mm f/2.0L at $1000 each compared to a Zeiss Ultra prime T1.9 135mm at $10,000 each has more to do with the economies of scale in making tens of thousands of the Canon lens as opposed to tens or maybe a few hundred of the Zeiss Ultra Prime pl mount lens, than it does with the optical quality. Again, there is almost certainly a little bit of improvement with the Zeiss based on slightly better design, components, and stringent quality control, but I would suggest that the difference in quality is not on the order of the difference in price, and in fact is fairly small.

There is a rule of thumb in manufacturing which says the price will drop in half when you can double the sales volume.

Mike, I just noticed the L (Cannon) after I did post my answer.
I do not have any experience with L and just by reading about it, it looks like much better lenses, though the newer EF-S should be much “more modern” manufactured by new standards and using new materials. Who knows what to think about it?
Don’t forget though that 135 format is two times the surface of S35 and the pictures taken on the 135 format film, usually are to be shown in very small picture formats, journal size in the most. Posters and large displays are heavily post processed and if not, then are shot on the 120 format film (56X118mm – 2.1 to 1 ratio)
So I will argue that Misterium has much better resolution then 135, even if you take the whole surface of 135 frame film versus S35 Misterium.
First the film is not completely flat, less particles in 135 frame then pixels in S35 frame.
Second Misteriumm in most cases will be stretched to 40 to 60 inches picture size to watch if not cinema size screen. The film is stretched to the newspaper size. (we are talking still lenses here, remember)
So when you design still lenses you have more slag then the movie lenses.

 

[Michael Hastings]

RE: canon mount

RE: canon mount

Erik:

Have you been working with RED on the mount - i.e. will you have by NAB or shortly thereafter the actual physical mount to the RED. If not any idea how long that will take. Also will you be pulling power from the B4 connector on the RED or where will you get it.

We will have done more testing by NAB... Which is when a (big) announcement will be made... Don't mean to be elusive. We want to do some more testing and evaluation before we announce whether we have a great product or a really great product; and at least be able to offer an initial indication as to which lenses we will be recomending, and possibly more importantly, recomending against.

 

[Andrew M.]

It's true that posters may be shot on larger film. But 266 pixels per inch is considered a minimum for magazine photos and of course the other elements are typically 600 to 1200 dpi and photo quality prints from even very cheap printers are much higher res. Consider that the mysterium image blown up to just 60 inches is only 75 pixels per inch - the difference is the likely viewing distance.

My sort of intuitive point here is to say that both SLR and cine people have many years of shooting to the same medium - 35mm film - and regardless of the direction of the frame and frame size - pros have always wanted/insisted on getting the most they can from that film - and are willing to pay for it. I think it is generally agreed that 2K clearly isn't as sharp as film and 4K is probably where you need to be to be comparable with film.

I would suggest to you that the price difference between a Canon EF 135mm f/2.0L at $1000 each compared to a Zeiss Ultra prime T1.9 135mm at $10,000 each has more to do with the economies of scale in making tens of thousands of the Canon lens as opposed to tens or maybe a few hundred of the Zeiss Ultra Prime pl mount lens, than it does with the optical quality. Again, there is almost certainly a little bit of improvement with the Zeiss based on slightly better design, components, and stringent quality control, but I would suggest that the difference in quality is not on the order of the difference in price, and in fact is fairly small.

Jim promised to demystify the lenses once the RED is out and RED ZOOM and RED primes are in. I hope that he and his people are doing homework on this right now.
Manufactures will do a minimum to make customer happy, so I agree the Zeiss you pay for the experience and name the first, economy of scale second.
Lenses for the special use are much much better then what we see for cine or HD lenses.
But still photography lenses or cine lenses will not be manufactured much above the precision that the customer needs it for. Still photography has economy of scale so it will be probably the best bang for the buck. Waiting for the RED lenses will probably open up the new revolution again in this field.
Remember, you do want to preserve the investment in the lenses once Misterium will be going 24MP or 8K so I hope the RED lenses will be manufactured with the future in mind and the precision that will go beyond the 12MP 4K sensor. Going in to 24MP in S35 size it calls for military specs.

 

[Mike the beginner]

On the question of red working with third parties it was indicated by someone at red team that "they are working with thirrd parties" but they have both signed a non-disclosure agreement. Hopefully Birger Engineering will be one of those those parties.

Keep up the great work you guys this is really interesting stuff!

Mike the beginner

 

[Ken Corben]

Future proofing?

Future proofing?

Looks like Mike and Birger may have a brilliant solution with the Canon 10-22mm for underwater apps if the product is broadcast HDTV 1080i that is also economical in scale. This is not trivial - this is a major breakthrough in size and cost for underwater filmmakers!

24P Redcode RAW Underwater:
I would also speculate that the educated guesses are spot on regarding cost to benefit of Zeiss v Canon for underwater apps. I am skeptical that 2K and 4K theatrical projection of underwater images with the Canon lens will compare to the Zeiss Prime to a DP's eye. That said, will the audience really care? I think not.

The speed "might" be an issue for the Canon at night or in caves etc?

Perhaps a moot point for many given the price difference. The proof will be in the testing, heh Mike?

 

[Andrew M.]

If anybody wants the test picture from 10-22 Canaon send me the private message with your e-mail address and I will e-mail it back to you.
Too big to post.
Then look at it on your 40"-60" screen or stretch it on your computer screen to the same virtual size.
(2MP RGB at 10 FL wide screen)

 

[Steven Parker]

DCS Seminar

DCS Seminar

Hey Bubble gang--

This may be old news for you all but the Digital Cinema Society has posted their Underwater Cinematography Seminar - Vince Pace is one of the speakers.

Go to www.digitalcinemasociety.org

and click on New Streaming Content. There's also a 4K seminar and other fun shooter stuff

Enjoy!

 

[Andrew M.]

Mike, here it is. When looking on it at 40-50" virtual screen size it looks very soft.
I would give it 1MP look. Put it on your screen and let me know your impression.
Picture ''wide1' was taken with FL 10 on EF-S, 10-22, F14, 1/500 slightly underexposed
Picture "wide2" was taken with FL 17 on EF-S 17-85, F13, 1/320, slightly overexposed


Canon 20D 8MP CMOS sensor S35 size cropped and saved with the pixel count close to 2MP RGB as a jpg.

Andrew

 

[Mark Thorpe]

So can we get some links?

Mark.

 

[Andrew M.]

So can we get some links?

Mark.

Sent to your e-mail address...

Check the ski rack at the left and right. You see color fringing and collimation problem, 2-3 pixels wide.
Find the focus problems if you can.
Looks like only 55% of the field is acceptable.

 

[Mark Thorpe]

Sent to your e-mail address...
Check the ski rack at the left and right. You see color fringing and collimation problem, 2-3 pixels wide.
Find the focus problems if you can.
Looks like only 55% of the field is acceptable.

OK, Got it. Yeah, I can see the 2 to 3 pixel blurring of the colors you mentioned. I would hazard a guess to say that the subjects face (family member?) is also soft. The fact that she's is wearing a beanie gives rise to shadows etc but it looks kinda soft.

The main thing I did notice is that it looks damn cold, way too cold for this warm weather, tropical island livin', underwater swimmin' British chap.

Cheers,
Mark.

 

[Andrew M.]

10-22 EF-S sample picture

10-22 EF-S sample picture

Mark, you got it right, EF zoom Canon lenses are soft.
I do not have much experience with no zoom lenses from Canaon so I can’t comment.
10-22 is more like 14-22 I wouldn’t open it to 10 even for 8X10 picture size, you will see color fringing and other distortions. And for 1 to 2.35 screen format you need these wide angle panoramic shots.
As to the colors, well it is Canada’s Winter! Just to cool you off out there Snow reflections and trees dead colors, I have some warm colors samples as well, out of these lenses. But good point, have to check it.

Why users do not demand test pattern pictures when they buy lenses?
Especially when you do spend more then $1,000 on it.
I think I will get myself a few good test patterns material and test all before I buy. 1000 and 2000 lines plus something for aberration testing.
When you spend a few $$K on the lenses they should give you individual lenses results TIFF files showing performance of the test patterns for this particular lenses or in minimum general test pattern with the tolerances statement. Otherwise it is like buying black magic.

Below I attached the fragment of the picture located ~85% to the horizontal side of the picture from EF-S 10-22 with the setting of 10. Pixel to pixel mapping of 2MP RGB with 1 to 2.35 screen format. It looks much better at setting of 14 but even then it is far from being useful for 1000 lines resolution.

 

[Michael Hastings]

Why would you crop it? The sensor is already almost identical to the RED, not to mention you want the highest res you can get to determine the lens sharpness. The RED images on the website are 4096x2034 3.6 Mb, and your file is 3504x1497 1.6MB. If you keep zooming in on the RED test stills you can see some color fringing even on the folds of the black and white (i.e. no color) polka dot hat, definitely not as much but the only images they have posted are from a $25K Cooke 65mm portrait focal length lens - which should be about the best possible, and from the 300mm - I haven't seen anything from a lens that would even be considered normal much less a wide or extreme wide angle. I'm not saying you're wrong I just don't think we've scientifically proved anything here. You need to grab a raw frame from the 10-22 and for an actual comparison of EF lens quality from an L series 50 or 85mm lens shot on both a 20D and EOS-1Ds Mark II full frame camera as well to get a real indication of the lens capability.

Canon USA Price : $ 1,399.95 Normal EF 50mm f/1.2L USM Autofocus Lens
Canon USA Price : $ 1,789.95 Telephoto EF 85mm f/1.2L II USM Autofocus Lens

These lenses would give you an overall better comparison of the expected quality of L series Canon lenses.

Any Superwide lens (remember the 10mm is 107 degrees - almost absurdly wide) is going to have some of those issues on the edges - and is to be expected even more so in an $700 SuperWide. The issue will be if the extra angle of view is worth the trade off over the 14mm L series listed below.

The next best option for a wide angle would be the L series 14mm

Canon USA Price : $ 1,800.00 Super Wide Angle EF 14mm f/2.8L USM Autofocus Lens

You would expect this one to be significantly better quality and it would be the alternative for underwater shooting - However it is about 90 degrees instead of 107 - still very wide - quite close to the old standard 15mm Nikonos.

The other lens I will be evaluating when the time comes is the 1,699.95 L series Zoom Telephoto EF 70-200mm f/2.8L IS $ 1,699.95
It is an L series with image stabilization.

The test I would really like to see would be between the Canon L Series 14mm and one of the Zeiss Ultra Prime 14s to see how much more you get for the extra $13,000.

The point of using Canon lenses is not to get the absolute cheapest they make - it is to get the best that they make - which are still about a tenth of the equivalent Zeiss or Cooke lenses - and in the telephoto side also include image stabilization which the Cine lenses don't have.

However, with that said - I think even some of the less expensive lenses may work well for those primarily looking to output 1080P high def. I personally want to shoot all of my underwater and nature footage at 4K, so will buy the L series as since I come from the broadcast video world where a cheap lens is $6K the L series actually seem cheap to me.

Mike, here it is. When looking on it at 40-50" virtual screen size it looks very soft.
I would give it 1MP look. Put it on your screen and let me know your impression.
Picture ''wide1' was taken with FL 10 on EF-S, 10-22, F14, 1/500 slightly underexposed
Picture "wide2" was taken with FL 17 on EF-S 17-85, F13, 1/320, slightly overexposed


Canon 20D 8MP CMOS sensor S35 size cropped and saved with the pixel count close to 2MP RGB as a jpg.

Andrew

 

[Erik Widding]

Have you been working with RED on the mount - i.e. will you have by NAB or shortly thereafter the actual physical mount to the RED. If not any idea how long that will take. Also will you be pulling power from the B4 connector on the RED or where will you get it.

Answers to these questions and more, will be available on the first day of NAB... As for the question of timeline, all I can say until NAB - I expect you won't be disappointed with the schedule.

 

[Michael Hastings]

Test charts

Test charts

Check out some of the DSC test charts - they are in Canada - should be able to google them. Or bring your camera to NAB

I think I will get myself a few good test patterns material and test all before I buy. 1000 and 2000 lines plus something for aberration testing.
When you spend a few $$K on the lenses they should give you individual lenses results TIFF files showing performance of the test patterns for this particular lenses or in minimum general test pattern with the tolerances statement. Otherwise it is like buying black magic.

 

[Michael Hastings]

Test charts

Test charts

Check out some of the DSC test charts - they are in Canada - should be able to google them. Or bring your camera to NAB

I think I will get myself a few good test patterns material and test all before I buy. 1000 and 2000 lines plus something for aberration testing.
When you spend a few $$K on the lenses they should give you individual lenses results TIFF files showing performance of the test patterns for this particular lenses or in minimum general test pattern with the tolerances statement. Otherwise it is like buying black magic.

 

[Michael Hastings]

Thanks, Erik I look forward to seeing you.

Thanks, Erik I look forward to seeing you.

Thanks, Erik I look forward to seeing you.

Answers to these questions and more, will be available on the first day of NAB... As for the question of timeline, all I can say until NAB - I expect you won't be disappointed with the schedule.

 

[Andrew M.]

Check out some of the DSC test charts - they are in Canada - should be able to google them. Or bring your camera to NAB

Thanks! Mike. I will get some test charts. I also want to test Misterium on these charts if it really pulls 2000 lines with REDCODE RAW.
As to cropping, remember 20D Canon sensor is 3:4 ratio, so you have to crop it to get 1 to 2.35 ratio. I did crop only top and bottom. Sides I didn’t touch.
Also 8MP Canon CMOS is delivering about 2.5MP RGB so after cropping for 2.35 screen ratio it will give you about 1.85MP RGB so it is almost as much pixels as you have on the HDTV 1080p screen.

Also I did check my lenses shooting Canon RAW, I see the same distortions.
I think comparing L with some cine lenses will be worth to do it.
I do not have any experience with L and I wanted to get some zoom lenses though.
Remember that we need something about FL10 since on S35 the neutral FL there is only FL28. On full size 35 (135) the neutral FL is 50.
When shooting wide screen 1 to 2.35 or 1 to 2.4 it is so good to have FL10. Gives you smooth panoramic picture. If I can’t find good zoom near 10 I will buy some good prime 8 or 10 and buy RED 18-85. As to FL 100 and up something bit sharper then Canon will be great.

 

[Ken Corben]

ND filters

ND filters

Any clever ideas for ND filtration in the housing?

Figure ASA 320 at surface to 10 feet in Fiji daylight is gonna be near or above F22?

Issues are screw on filter vignetting on the wide primes?
Interchangeability in the housing (on/off)?

 

[Michael Hastings]

Explain RGB

Explain RGB

I'm not sure I understand why you are using RGB numbers.

I think for absolute best results in abovewater shooting you are going to be better off with the Canon L type 14mm or you are going to have to bite the bullet and buy one those ultra primes.

Underwater, even if the EF-S is a little softer than the 14mm we may be better off with the wider angle because you are shooting through less water.

Thanks! Mike. I will get some test charts. I also want to test Misterium on these charts if it really pulls 2000 lines with REDCODE RAW.
As to cropping, remember 20D Canon sensor is 3:4 ratio, so you have to crop it to get 1 to 2.35 ratio. I did crop only top and bottom. Sides I didn’t touch.
Also 8MP Canon CMOS is delivering about 2.5MP RGB so after cropping for 2.35 screen ratio it will give you about 1.85MP RGB so it is almost as much pixels as you have on the HDTV 1080p screen.

Also I did check my lenses shooting Canon RAW, I see the same distortions.
I think comparing L with some cine lenses will be worth to do it.
I do not have any experience with L and I wanted to get some zoom lenses though.
Remember that we need something about FL10 since on S35 the neutral FL there is only FL28. On full size 35 (135) the neutral FL is 50.
When shooting wide screen 1 to 2.35 or 1 to 2.4 it is so good to have FL10. Gives you smooth panoramic picture. If I can’t find good zoom near 10 I will buy some good prime 8 or 10 and buy RED 18-85. As to FL 100 and up something bit sharper then Canon will be great.