Edited by Ram Shani, 29 June 2010 - 03:28 PM.
Jump to content
new canon 4k!!
17 replies to this topic
Posted 29 June 2010 - 04:53 PM
I second that Canon is developing something in this direction. No wonder.
Posted 29 June 2010 - 05:16 PM
That's nice and all but they clearly haven't learned anything from having their DSLRs used for video work.
First problem is that it will still be EF mount. They are incredibly foolish to not offer a PL mount. If they offered it, they could at least do it at an upcharge and take some of the money that's been going to clairmont, keslow, birns and sawyer, etc. to slap PLs on canon DSLRs.
Second, electronic follow focus? That's just silly. It means every lens will be more expensive than it needs to be and there still won't be repeatable focus marks because the motoring and gearing inside isn't made for it. One must also factor in the problems I foresee with the electronics going down all the time because they won't put a durable professional grade connector on the thing.
SDI out seems to be the only thing they're planning on getting right, assuming it's an out in 1080P. The link is obviously a rumor and doesn't specify SD or HD.
Posted 29 June 2010 - 07:48 PM
I agree . . . an interchangeable or PL lens mount would be better than EF only. But it makes sense in the "corporation" mode, (while making no sense to the rest of us, of course): no self respecting lens and camera manufacturer would diss their lenses (or lens mount) for someone else's, even if they are not a traditional cine lens manufacturer. Unfortunately for us, Canon did not do it for the XL series, and it doesn't look like they will right now.
While Jannard is still dithering on the Scarlet series, Canon and others are going to beat him to the punch. So Jim, do yourself a favor and release a sub $3K, non (pseudo) RAW DSMC 3K camera with included on-board monitoring, interchangeable lens mounts, 4:4:4 HD SDI out and other goodies before your company's sales peter out once and for all. This industry is crazy about cost-savings. Few will buy $20K body-only cameras when they can get similar results with $5K camera bodies, RAW or no RAW. You can offer both, of course, but the time is NOW!
Posted 29 June 2010 - 08:37 PM
To me buying a line skipping camera is like buying an interlace camera because I need interlacing aliasing artifacts like I need a hole in my head. However with this new Canon 4K sensor which promises no line skipping it might be all over for Jim Jannard especially if the Canon can shoot at high frame rates.
Posted 29 June 2010 - 09:38 PM
Does this mean we get yet another revolution in filmmaking? Awesome!
Posted 29 June 2010 - 09:41 PM
Electronic follow focus makes perfect sense using EF lenses with their built-in focus motor. My experience experimenting with Canon's remote control software and my 7D and two EF-S lenses is that there's a bit of backlash when returning to the same focus point. Canon will have to improve the motor mechanism to attain really good electronic FF. Has anyone here played around with the remote software and an "L" series lens?
Posted 30 June 2010 - 01:46 AM
Having played very briefly with driving an EF lens electronically, in my experience they do not necessarily nail focus marks with a great degree of repeatability unless you return the lens to an end-stop every so often. Canon EF cameras visibly do this during extended periods of hunting for focus. I may have been driving it wrong, of course; the protocol is not public and everyone who's tried to reverse-engineer it has had compatibility issues.
Posted 30 June 2010 - 02:32 AM
Isn't this what the Birger mount for the RED One is supposed to do? However, this mount no longer has the RED seal of approval.
Posted 30 June 2010 - 04:49 AM
This may be a factor that's specific to some particular lenses, but I think the problem is that the Canon EF lens system just isn't designed to do what people are trying to make it do with those controllers. Perhaps I can explain this best by example. Everything from here down is prefixed with "As far as I know" and suffixed with "...but I might be completely wrong", but I'm reasonably confident in it.
The fundamental problem is that there is often no need for an EF lens to know what distance it's focussed at, and not all of them support reporting that (perhaps none do, but I'm not sure). The focus detection systems in (D)SLRs tend to work on a "farther" or "closer" basis, not an absolute distance measurement. As such there are as far as I know four instructions in the EF lens command set which have to do with focussing:
- Go to infinity
- Go to MOD
- Read current encoder position
- Go to encoder position (x)
So, it supports "farther" (a higher encoder position) or "nearer" (a lower encoder position) but it does not intrinsically support "go to six feet".
There are further strangenesses. The encoder position is not locked to the actual barrel position, especially if you take the lens out of AF and move the barrel manually. The only way to find out where you are at any given moment is to return to an end stop, note what the reported encoder position is, and work from there. The end stop will not always be at the same encoder position.
You might think it would be possible to characterise a particular lens such that you knew that "six feet" was (x) encoder positions up from zero. This does work pretty well. The system uses two bytes to indicate lens position so an EF lens is theoretically capable of handling any one of just over 65,000 discrete focus positions, although I never saw a value that suggested the entire range is in use.
The problem with this is that moving from place to place within the range is subject to quite a lot of drift. If you decide that six feet is 5,000 encoder positions higher than MOD, that's fine, you'll nail that with some reliability. However if you then decide that seven feet is 500 encoder positions higher, go to seven feet, then attempt to return to six, the system is subject to really rather a lot of drift and sloppiness. This is a reflection of the fact that it is not designed to do this, it is designed to hunt around until an external feedback system informs it that it is in focus. It isn't designed to nail absolute positions with absolute reliability.
Of course it is possible for Canon to engineer around this by producing a lens range that does have its encoders locked accurately to barrel position, but this means that what we're likely to see is an "EF Movie" lens range which will, of course, be considerably more expensive than normal EF stills lenses.
Posted 30 June 2010 - 07:44 AM
Did the specs listed recording format? 4K RAW? 1080P? Compressed (likely, if to CF cards)?
Posted 30 June 2010 - 08:18 AM
Which lens? I'm really curious to know if the "L" lenses are any different with respect to the backlash issues. Though with 16 bit drive it should be possible to characterize a given lens and build a LUT so that the drive software knew exactly how to get the lens back to a given focus.
Posted 30 June 2010 - 09:05 AM
Probably whichever was the kit lens with the EOS-1000F. The 35-80? I bought something cheap from ebay on the basis that I might inadvertently destroy it with my tinkering. Therein may lie a problem as I am informed that the protocol has been altered since it was developed in the mid to late 80s, although I have not come across a reported case of Canon/Canon incompatibility.
I fear not, at least with some lenses. They just aren't repeatable. You can set the thing to encoder position X, set it to encoder position X+1000, then set it to X again - and it will not be in the same place it was originally. The system appears to be designed to provide for "up a bit" or "down a bit" with a degree of proportionality, rather than absolute indexing. I don't think certain lenses will ever be controllable in the way that's being proposed here. You can get to a given focus but only by winding it back down to the end stop and going from there.
Reportedly some lenses (possibly Ls) do have nominated distance stops at places other than infinity and MOD, which is presumably a measure to allow it to reindex and find focus more quickly, but I'm not sure it helps us here much.
Sorry, this is turning into a thread hijack.
Posted 30 June 2010 - 12:30 PM
That is impressive control of the lens' motors by the electronics / software, with a caveat for real world use. I watched the first video of the series only and it looks a very "visual" method of focus pulling. No mention that I could find of the more traditional method of pulling focus off the lens barrel and tape measure "by the numbers." How many camera assistants have a full sized high res monitor in ideal viewing situations available on set to them at all times to pull focus off of ? . .
Posted 30 June 2010 - 12:41 PM
I've never used the system, but it does seem rather like using a remote follow focus, where you do make your own distance marks on the remote. Perhaps that could involve measuring and making the marks yourself for each lens, but as I said I haven't used it, although I got the impression, once set up, it would line up with the lens focus marks
Posted 30 June 2010 - 03:49 PM
Well, yes -but just to play devil's advocate- for focus pullers to go through every lens and make marks could be rather cumbersome. There are proven focus pulling technologies where that is not necessary, so I doubt any focus puller would want to embrace a technology that makes his on-set life harder. Maybe I am not getting the full picture of the product's capabilities from the video tho.
Edited by Saul Rodgar, 30 June 2010 - 03:50 PM.
Posted 30 June 2010 - 04:59 PM
I suspect it was aimed at the lower budget end of the RED market, so perhaps they would be more willing to put up with issues like that. If you had the budget, you'd just get the real thing lens wise.