Why (formerly not) Canon EOS lenses?

[Michael Hastings]

Canon control versus Nikon

Canon control versus Nikon

Erik:

Isn't the Canon iris electric, as opposed to a mechanical pin on the Nikon that makes it harder to make an accurate controller.?

It seems to me that one of the confusion points for people on the forum is that the Canon and Nikon mounts listed from RED are just physical mounts for the older mechanical style lenses.

Yet the newer lenses - and particular the Canon EF-S and Nikon DX Lenses which, because they are designed for the digital sensor which is smaller - like the RED sensor - give us the SuperWide lenses like the 10-22 that just aren't available in any of the old mechanical mounts - as they wouldn't have covered the full 24x36mm frame of standard 35mm Still cameras. The EF-S and DX lenses operate via electric motors - which is really more like what we are familiar with on video cameras.

RED may at some point come up with controllers for these and for iris control on B4 video lenses, but it is a whole lot more complicated. One of the reasons RED could achieve a camera in such a short time is that they really didn't have to deal with exposure detection and autoiris - since they were dealing with the motion picture paradigm where exposure is always manually controlled.

RED's electronic engineering task really was just integrating the chip and then a bunch of computer interfaces. Exposure is the user's problem, and since they are essentially feeding out some form of RAW data (even if compressed) they didn't have to deal much with the knee circuits, colorimetry, etc. that really complicates things in a video camera. Not to trivialize the RED engineering, but simply to point out that comparing a RED to a broadcast type HD camera is not entirely apples to apples.

Erik

You mentioned in another thread about Nikon lenses performing very well. I know it is early days yet but do you have a preference for Nikon or Canon lenses, from a control point of view?
M

 

[Erik Widding]

Isn't the Canon iris electric, as opposed to a mechanical pin on the Nikon that makes it harder to make an accurate controller.?

The Canon EF control is a preference for us, over the Nikon F, because we already have it as a tested and shipping product. It has been shipping in various forms for about five years. We give the Canon lens digital numbers to command it to do things, and we get similar feedback. This way we know if the lens did what we commanded it to, did almost what we commanded it to, or just errored out. We chose to do the Canon first because it was all electronic.

As a contrast, there is a mechanical component to the iris control on the Nikon lenses. THe Nikon D and G type lenses tell us what they are capable of doing, but a mechanical actuator is still required for the iris. In the one Nikon lens that I have taken apart (a very cheap one) I was surprised to find an encoder that allowed for the backlash in the gearing system to be accounted for in focus. It suggested to me that the cheap Nikon lenses may produce a more accurate result than the cheap Canon lenses. The cheap Canon lenses don't have this. It makes sense. Until recently all Nikon lenses had to have a gear train. Even the high end ones. Canon had the luxury within the framework of their system to have direct drive.

 

[Erik Widding]

ERIK: I am just as happy to use your controller. All we really want to be able to do is put a 12" or so cable so we can have the knobs from the outside located in a handy position. In other words one method would be to just unsolder your pots and attach a short wire and then attach the pot to the sealed shaft that penetrates the wall of the housing. Obviously a much nicer solution would to either install a connector ourselves to parallel the control knobs, or ideally to have you have a connector(s) (I assume they are just 3 leg pots, so a six pin or two three pin) on your control unit that allowed us just tap into or parallel the pots to our pots and knobs. (We have pots that we use with broadcast lenses, that come with 1/4" stainless steel long shafts that can be sealed directly with our standard control seal.) It seems to me this would be useful for anyone that wants to do remote from a jib or whatever. (I realize that situation may be handled by just moving the controller and having a longer RS232 cable - but it seems like the connector would be pretty cheap and even allow the possibility of alternative pots, rockers, splitting the iris and focus knobs apart for two handed use on a tripod or separate people having responsibility for focus and iris) The connector could parallel or just be switched back and forth if there is an electronic conflict.

This gives me a bunch of things to think about...

 

[Michael Hastings]

Canon EF mount iris

Canon EF mount iris

My point wasn't so much about accuracy, it was simply that the Canon is all electronic - the Nikon mount would have to have some kind of motors and such to move the pin on the Nikon lens - creating the need for a significant amount of electromechanical design. But it is interesting to note on the focus. How much difference are we talking in practical terms, and do the commands actually equal a specific distance for the focus or is it all relative?

As a contrast, there is a mechanical component to the iris control on the Nikon lenses.

 

[Erik Widding]

My point wasn't so much about accuracy, it was simply that the Canon is all electronic - the Nikon mount would have to have some kind of motors and such to move the pin on the Nikon lens - creating the need for a significant amount of electromechanical design.

We are presently doing this work. And I will acknowledge "significant" is a reasonable descriptor. When we can, we substitute electronics, software and sensors for mechanical stuff. The tooling costs are so much less expensive. Little tiny motors are not the easiest thing in the world to buy. There isn't so much in the way of standard prodcts to buy.

But it is interesting to note on the focus. How much difference are we talking in practical terms, and do the commands actually equal a specific distance for the focus or is it all relative?

Good question, but unfortunately I will have to answer later. We are finishing up the servo mode on the Canon controller, and do know that we will make improvements. It works great with some lenses, and needs a little more tuning with some others. With the Nikon lenses we just don't have a large enough sample size to know yet.

 

[Jon Armstrong]

Being a long time Canon user my favourite lens is a very old f2.8 20-35. none of the later lenses can match it since the later ones exhibit a lot of spherical aberation.

Many of the old Canon FD lenses are extremely good. They may not meet the optical standards of the new high tech lenses but they are well coated very sharp and have little or no distortion

The major advantage of the EF lenses is the enormous exit pupil upon which the efficiency of the optics depends

Jon

 

[Michael Hastings]

Re: focus accuracy and other issues

Re: focus accuracy and other issues

Erik:

My question about accuracy was really about the Canon lenses you have been working with.

BTW: There seem to be a fair number of people that are hung up on using their existing Nikon lenses, and that 's OK, but to me - coming from the video world where the cheapest B4 wide angles (not superwides) are about $7K and the Superwides that would be equivalent to the 10-22 start at about $25K - a precision 10-22 lens for $700 and even if your adapter is $2K - is to me almost laughably cheap. So to have to purchase a new Canon lens if that is what works better doesn't seem to me much of a sacrifice.

PS Just a decent wide angle converter for say an HVX200 is about $1200.
PPS - It still would be nice if your adapter and controller came in at $1500 or so.

My original question: How much difference are we talking in practical terms, and do the commands actually equal a specific distance for the focus or is it all relative?

Good question, but unfortunately I will have to answer later. We are finishing up the servo mode on the Canon controller, and do know that we will make improvements. It works great with some lenses, and needs a little more tuning with some others. With the Nikon lenses we just don't have a large enough sample size to know yet.

 

[Erik Widding]

My question about accuracy was really about the Canon lenses you have been working with.

With lenses that have ring-type USM motors we are seeing repeatability on the order of a few steps. The step counts for some of these lenses:
85mm f/1.2: 4200 steps
14mm f/2.8: 1400
70-200mm f/2.8: 2500
200mm f/1.8: 3750
200mm f/2.8: 2100
35mm f/1.4: 2500
60mm f/2.8: 2000
100mm f/2.8: 2500

On all of these lenses we have repeatability on the order of 0.1%. Steps correspond to more of a log than a linear scale of focus distance. Half of the steps might be used for the first meter or two of focus on some lenses. Steps are linear in terms of the rotation of the mechanism.

The 10-22 f/3.5-4.3 has about 1200 steps. We have played with this lens for about five minutes. It is not one of the ones that works really well with servo mode, yet. We will add it to our set. Give us a few weeks... It is extremely repeatable with our normal mode of operation, move stop move stop etc. Servo mode is really new.

Basically the lenses are more accurate in their positioning than a typical human being is capable of by simply turning the dial manually. This offers some really interesting possibilities if Graeme's radical new focus assist were to be the basis for an auto-interactive focus mode for ENG users. We can all guess what the focus assist is going to offer.

But if this is even just a texture based quality of focus (maximized high frequency component means better focus) it has enough information to keep a "talking head" in focus as the 3D path of the head relative to the focus plane can be understood. If best focus is near the middle of the object (think nose) focus is too close. If best focus is near the outer edges, focus is too far away. Blob analyze the scene... blob analyze the focus quality... apply Newton's second law (i.e. a PID loop)... voila... auto-interactive focusing...

I call this auto-interactive focus, as the camera operator just has to pick which blob to start tracking.

BTW: There seem to be a fair number of people that are hung up on using their existing Nikon lenses, and that 's OK, but to me - coming from the video world where the cheapest B4 wide angles (not superwides) are about $7K and the Superwides that would be equivalent to the 10-22 start at about $25K - a precision 10-22 lens for $700 and even if your adapter is $2K - is to me almost laughably cheap. So to have to purchase a new Canon lens if that is what works better doesn't seem to me much of a sacrifice.

Many of our industrial customers have the same hangup. Surprisingly even after their economics are accounted for. You wouldn't believe the per minute costs to run some test facilities, from the testing of operating jet engines to the crash testing of automobiles. Just saving 15 minutes on the setup of the imaging system can pay for the controller and the lens. That being said, "the customer is always right"; we are coming out with a nikon product; and we expect it will mostly just help with the credibility/attention for our canon product.

PPS - It still would be nice if your adapter and controller came in at $1500 or so.

I think you will not be disappointed...

 

[LighthouseMEdia]

Idea guys....
What if you could also program several focus points, and have the lens transition between them?

 

[Michael Hastings]

Pricing on EF adapter

Pricing on EF adapter

Erik - your the man!!!!

----PPS - It still would be nice if your adapter and controller came in at $1500 or so.----

I think you will not be disappointed...

 

[Michael Hastings]

Re: Ef-s 10-22

Re: Ef-s 10-22

Erik: Do you have one of the 10-22 or do you have to borrow one? I know it is tough to do development trying to work around a loaner schedule and would be willing to buy one of the 10-22s and leave it with you for a while so you could work on it at your leisure. The 10-22 is important for us as it seems to me the only real solution for high quality. low cost superwide on the RED.

Also wondering if you could explain a little more what is servo mode?

The 10-22 f/3.5-4.3 has about 1200 steps. We have played with this lens for about five minutes. It is not one of the ones that works really well with servo mode, yet. We will add it to our set. Give us a few weeks... It is extremely repeatable with our normal mode of operation, move stop move stop etc. Servo mode is really new.

 

[Erik Widding]

Erik: Do you have one of the 10-22 or do you have to borrow one? I know it is tough to do development trying to work around a loaner schedule and would be willing to buy one of the 10-22s and leave it with you for a while so you could work on it at your leisure. The 10-22 is important for us as it seems to me the only real solution for high quality. low cost superwide on the RED.

A rental house we are working with in Montreal has one. I will probably add one to our test set in a couple of weeks. This lens will definitely become a permanent member of the test set when we release the motorized zoom version of our product, possibly late this year. But if you wanted to loan us one of these for a few weeks I certainly won't complain. ;-)

Also wondering if you could explain a little more what is servo mode?

Servo mode is a term that we stole from the SLR guys. When using the camera in "servo" mode the focus will track a moving object while the photographrer choses when to actually shoot one or a series of pictures. All the great sports shots that you have seen were taken using the camera in this mode.

So, in servo mode, we command the lens to adjust its trajectory and or destination while the lens is moving. This allows us to make the lens movement look smooth so adjustments can be made while acquiring imagery. The trajectory that we are following is in the short term a moving knob, or a series of waypoints from a moco system, for example.

Another way of putting this is our normal static mode will move the lens as commanded as quickly as possible, and then tell the controlling system that it has completed this move. This lag does not allow for continuous motion. The servo mode allows a time (speed) component to be included so that the lens will be where you want it when you want it there.

 

[Vincent Rice]

Can I just put vote in for Olympus/Leica/Sigma 4/3 system lenses? Its always struck me that the 4/3 aperture size is perfect for 35mm motion picture use. The units are very compact and Olympus 'exotic glass' lenses are generally very good.

 

[Tom Lowe]

I've got a 10-22 if someone in the LA/OC area wants to test it.

 

[Jeremy Hughes]

http://www.red.com/news/view/78

Yay!

 

[Jon McCoy]

Am I correct in surmising then, that the real difference between a 35mm cine lens, and a 35mm photographic lens is superficial? Immediately to mind - geared/toothed focus ring for follow focus apparatus (okay, not "superficial", but can be overcome).

I've come from a video background, so have always bought lenses flogged by the manufacturers for the specific cameras or systems - a combination of vendor obedience and "compatibility".

Given that the Red camera is effectively a Digital 35mm camera, it'd seem that the laws of 35mm seem to apply.

With that in mind, clearly if you can get the focal points to line up, then 35mm film is no different from 35mm film. (Not to mention, if a lens can shoot 16M pixels, then it's bound to be happy shooting 4K without distortion).

Simply, I'm trying to get it straight, ensure I'm not getting high off the fumes from the local industrial estate, that I can get the quality and the flexibility of shooting 4K on a Red One camera, using an EOS (or even 4/3rds) lens?

:help:

 

[DAServices]

Erik,

Have you tested the 16-35mm f2.8/L? Or plan on supporting it? And if so how repeatable is it? I plan on buying your mount and currently own 70-200mm f2.8/L (Which you support), 85mm f1.2/L (Which you support), I just wanted to know about the 16-35mm f2.8/L, 100-400mm f4.5-5.6/L, 50mm f1.8, and the Tamron 28-105mm f2.8. Have you tested and Canon 3rd party EF lens?

 

[Erik Widding]

Have you tested the 16-35mm f2.8/L? Or plan on supporting it? And if so how repeatable is it? I plan on buying your mount and currently own 70-200mm f2.8/L (Which you support), 85mm f1.2/L (Which you support), I just wanted to know about the 16-35mm f2.8/L, 100-400mm f4.5-5.6/L, 50mm f1.8, and the Tamron 28-105mm f2.8. Have you tested and Canon 3rd party EF lens?

We won't be testing third party lenses on our mount. If you are interested in Sigma lenses they will be supported on our soon to be released SigmaSA and FourThirds mounts. This is not to say they won't work. We just won't be testing or supporting them. Five minutes with the lens, mount, and controller of your choice (i.e. SuperGrip) will give you a feeling as to whether or not you will like a lens for a given setup and application.

The 100-400mm lens might be a bit of a pain to use, given the large change in length when zooming, and the push-pull mechanism. Rotary zooms can be easily motorized. Push-pull not quite so easy.

The 50mm f/1.8 is a pretty crapily constructed lens (from a mechanical perspective). It makes nice images, but the mechanism has a tendency to wind up when it hits the stop, and absolute position reference will be lost. If this lens is used, it will be fine with a "rocker" control (i.e. ENG grip), but almost certainly not with a "knob" (i.e. cinema follow focus) or in a moco setup.

 

[DAServices]

Yeah, I really didn't plan on using the 100-400. I figured if I did it would be a pain, because of the push/pull, The 50mm I almost never use once I bought the 85mm f1.2. The 70-200mm f2.8 (my favorite on the 1 DS mkII) but the one I'm very interested in also, is the superwide 16-35 f2.8/L have you done any testing yet?,

 

[kgelner]

One thing about the earlier post suggestion 4/3 format lenses - what is the sensor size on the Red? If it's really 35mm then it wouldn't be able to make use of the lenses build for 1.5x crops or below... if the sensor were small enough, the Sigma 10-20mm EX would be an interesting alternative.

For those asking after the quality of the Sigma 14mm and 15mm primes earlier, I can vouch from personal use that both of these are extremely sharp - the 15mm is a fish though, just to warn you it would be somewhat curvy. Some example images from each lens can be found at PBase in the Sigmadslr user lens sample galleries:

http://www.pbase.com/sigmadslr/14_mm_ex_

http://www.pbase.com/sigmadslr/15_fish

Now what you really want on the end of the Red is a Sigma 200-500 f2.8...

http://kigiphoto.smugmug.com/gallery/2582507#136064090