Something I find frustrating is when a solution to a particular problem, through popularity, becomes the de facto standard among the mouth-breathing masses.
Case in point here: Nikon lenses on Canon DSLRs for video. Now before anyone decides to flame (if anyone is reading this), hear me out.
When the 5D mark II first came out, it had this awesome 1080P mode that everyone wanted to use, but no manual exposure control. And all the Canon EF lenses have electronic aperture controls, so it's really quite a bad combination for people who want to set their exposure exactly how they want it.
Some clever people realised that older (and some current) Nikon lenses still have manual aperture rings, and used adapters to mount them on the 5D. Problem solved. Good lenses, manual aperture control. Bingo. A good hack.
But this is now not necessary as Canon have released firmware that lets you set the exposure manually. It was never necessary on the 7D. Or the 550D. Hack not required any more.
But, still so many people shoot with Nikon lenses on these cameras. And aside from the popularity argument, which is never a good reason to do ANYTHING, I can't think of a single reason why...
Now, before the Nikon fanboys flame me about the superior quality of their lenses, let me finish. Canon and Nikon both have fantastic lenses. The high end of both marques are very good indeed. Nikon do currently have some amazing NEW lenses out that are super sharp. They are putting Canon's lenses to shame. Examples would be the new Nikon 24mm 1.4, and the Nikon 14-24mm zoom. They are drool-worthy lenses. Specifically the 24mm 1.4... it makes me wish Nikon would make a good video DSLR that does 1080P. I cannot stress enough how good this new Nikon lens is. It's almost indescribably good. It makes me Canon 24mm 1.4L (version 1) lens look really poor at the very wide apertures.
Problem is, it's one of Nikon's new lens designs with no manual aperture ring. Same for all the really diamond-cuttingly sharp new Nikon lenses. The ones with manual aperture rings are all really old optical designs. Good, but nothing compared to the new ones. So yes, Nikon has a bunch of truly spectacular lenses right now, but you can't use them on a 5D with an adapter.
So, it boils down to this. You can use old optical designs on a 5D with an adapter and bathe in the associated cool-factor lent to them by the mega-mastery of the new lenses that you can't use. Or you could just use Canon lenses.
In short, I'm saying it makes no sense from an optical point of view. It can make a difference if you consider price, because a lot of those old Nikon lenses are cheap on ebay. And still good enough to be usable. But you cannot even compare them to some of Nikon's new lenses, which cost sometimes 10x as much.
Monday 26 April 2010
Friday 23 April 2010
A Spring video with the new 550D
Ok, so it's not so new anymore. But I shot this about a month ago, and it's doing really well on Vimeo. Have a look. I wrote a bunch of details about it too.
This is a short that evolved out of a camera test. The whole thing was a collaboration between Creativesunshine (a production company that specializes in movement and dance on-screen) and myself (eager to test out the new 550D). Our lovely actress: Nikita Mitchell nikitamitchell.com
So, I brought along a few nice lenses (Canon 50mm 1.4, Canon 24mm 1.4, Sigma 12-24mm). I also had a 100mm 2.8, but never used it because the other lenses gave us the coverage we needed.
I don’t have a Z-finder or an LCDVF or anything, so I was quite concerned about pulling focus and even just setting focus, but to be honest, this hadn’t proved to be a problem. Maybe the wider screen of the 550D lets you see just a little better. If I look closely, I can see if something is in/out of focus. The brightness of the screen was no problem either. It wasn’t the brightest sunny day (thin thin veil of cloud sometimes), but it was bright and the screen kept up. That said, a monitor would have been handy for the very low angle shots (we rehearsed the take, but then shot blind as I couldn’t follow the small screen at that speed).
We used ND to keep the aperture wide open… between 1.4 and 2.8 depending on the blur we wanted. We used tiffen 4×4 filters in a matte box which we just attached to the front of the lens – no support rods were necessary on the relatively short lenses. I will be buying some in the near future for flexibility though. We had a .6 and a .9. We had a couple of other filters along for the ride which we didn’t use.
The field of flowers was some park in Wimbledon… I don’t know the name and lost my bearings on the way. The patch of flowers was smaller than it looks though - clever framing makes it look bigger.
Camera movement was handled by a small skateboard dolly and a tripod and a little mini-crane or boom rather. The tracking shot of her twirling around was actually hand-held because we didn’t have time to set up the dolly again for that shot. For the boom, a monitor would have been reeaaally useful – because again, we rehearsed a few times and then shot blind. It’s impressive the results this can yield though, if you’re not too much of a control freak. I find monitors on-set can slow down a production immeasurably. Much faster work can be done with a DoP the director trusts. That’s how it was here, and the whole thing was very fast and the results are fantastic. Occasionally we’d go over a take to check it was the right idea, but generally, it went very fast.
FCP Training
The camera aided this speed - it’s actually really easy to use. After everything is set to manual, all the settings are easily changed with the Q-button, while things like aperture and ISO are easily changed as they have their own buttons. I missed the rear dial control of the 5D and 7D for about 5min. After that, I really didn’t care. I even kind of like the button approach of the 550D. It’s pretty much impossible to nudge a setting like you can do with the rear control wheel.
A kelvin white-balance would be useful, but just using the old video trick of a white-balance on something white works even better, even if it takes a second to switch to photo mode, take the photo, set the WB to that photo, then go back to video. Auto WB is surprisingly good on this camera though (though I didn’t use it on this occasion).
Camera settings wise: I just used the camera’s Neutral colour settings – none of the superflat or custom picture styles. It looked amazing straight out of the camera. I was really tempted to use it ungraded (beautiful soft look), but just added the tiniest hint of saturation, warmth and contrast to make it pop a touch more.
I didn’t run into any of the overheating issues some folks have. I don’t know if it’s because I’m using that fast Transcend 16gb class 10 SD cards… buffering is never a problem. Maybe too much buffering exacerbates the problem? Consequently I have run into a heat issue… but even then, it was surprising after reading how ‘bad’ it was. I was shooting in a dance studio. It was really hot and really humid (after a class of 30 or so had been dancing for two hours). I was shooting almost continuously without turning the camera off for about 30min when I got a heat warning. I simply turned the camera off between takes, and it went away. Perhaps with summer coming, it’ll pop up more… but I think if you just power-down between takes, it’s probably a non-issue. Also, if you work with people who’re used to film and not HD video… they are used to a little downtime between shots.
Training guide
This camera can also roll really quickly - very useful. I can snatch shots (like the final shot) where everything just looks lovely and it wasn’t a planned take.
I have found with the higher-resolution or bigger screen (wider, so more of the 16×9 shot over more pixels), that following focus is really not so tricky. I’ve shot a few things with unpredictable movement in the past week, and just followed it myself (no separate focus puller), by just turning the lens barrel (no follow focus rig). It’s not a perfect solution. A good follow focus rig would be much easier. And a good focus-puller pulling between measured marks with actors hitting those marks would be ideal. If the actor missed the mark, a nice 7-inch monitor would be nice, with peaking, to help the focus puller. But in a non-ideal world where you’re shooting fast and light – it’s actually possible to focus just on this screen.
We knew we would be slowing down a lot of the shots to give it a sort of imaginary, light, lyrical quality (I always need slow-motion to be motivated… otherwise it’s gratuitous). So we shot most of it at 60fps (some at 50fps to see if there was a noticeable difference in the motion. Shutter mostly stuck at 125 for ‘natural’ motion blur (the whole 180 shutter thing). Some of the faster movements we experimented with higher shutter speeds. I can’t remember which ended up in the video. Jello-cam just didn’t really seem to be a problem. So long as the footage is stabilized somehow (even a hand-held stabilizer or a shoulder rig or even a picked up tripod) makes a world of difference. Quick pans or rotational instability causes horrid jello-cam… just don’t use either, and you’ll be fine. I’ve never been a fan of whip-pans anyway.
Editing… I did it before I got the FCP plugin, so I just used an MPEG Streamclip batch to convert the files to Prores 422. Then I conformed them all to 25fps in Cinema Tools, then started editing. Colouring just done in Colour (wish they’d change the name to a less common word – would help in searches. Same as Motion).
That’s about it. If anyone has any questions, I’ll get back to people when I can.
This is a short that evolved out of a camera test. The whole thing was a collaboration between Creativesunshine (a production company that specializes in movement and dance on-screen) and myself (eager to test out the new 550D). Our lovely actress: Nikita Mitchell nikitamitchell.com
So, I brought along a few nice lenses (Canon 50mm 1.4, Canon 24mm 1.4, Sigma 12-24mm). I also had a 100mm 2.8, but never used it because the other lenses gave us the coverage we needed.
I don’t have a Z-finder or an LCDVF or anything, so I was quite concerned about pulling focus and even just setting focus, but to be honest, this hadn’t proved to be a problem. Maybe the wider screen of the 550D lets you see just a little better. If I look closely, I can see if something is in/out of focus. The brightness of the screen was no problem either. It wasn’t the brightest sunny day (thin thin veil of cloud sometimes), but it was bright and the screen kept up. That said, a monitor would have been handy for the very low angle shots (we rehearsed the take, but then shot blind as I couldn’t follow the small screen at that speed).
We used ND to keep the aperture wide open… between 1.4 and 2.8 depending on the blur we wanted. We used tiffen 4×4 filters in a matte box which we just attached to the front of the lens – no support rods were necessary on the relatively short lenses. I will be buying some in the near future for flexibility though. We had a .6 and a .9. We had a couple of other filters along for the ride which we didn’t use.
The field of flowers was some park in Wimbledon… I don’t know the name and lost my bearings on the way. The patch of flowers was smaller than it looks though - clever framing makes it look bigger.
Camera movement was handled by a small skateboard dolly and a tripod and a little mini-crane or boom rather. The tracking shot of her twirling around was actually hand-held because we didn’t have time to set up the dolly again for that shot. For the boom, a monitor would have been reeaaally useful – because again, we rehearsed a few times and then shot blind. It’s impressive the results this can yield though, if you’re not too much of a control freak. I find monitors on-set can slow down a production immeasurably. Much faster work can be done with a DoP the director trusts. That’s how it was here, and the whole thing was very fast and the results are fantastic. Occasionally we’d go over a take to check it was the right idea, but generally, it went very fast.
FCP Training
The camera aided this speed - it’s actually really easy to use. After everything is set to manual, all the settings are easily changed with the Q-button, while things like aperture and ISO are easily changed as they have their own buttons. I missed the rear dial control of the 5D and 7D for about 5min. After that, I really didn’t care. I even kind of like the button approach of the 550D. It’s pretty much impossible to nudge a setting like you can do with the rear control wheel.
A kelvin white-balance would be useful, but just using the old video trick of a white-balance on something white works even better, even if it takes a second to switch to photo mode, take the photo, set the WB to that photo, then go back to video. Auto WB is surprisingly good on this camera though (though I didn’t use it on this occasion).
Camera settings wise: I just used the camera’s Neutral colour settings – none of the superflat or custom picture styles. It looked amazing straight out of the camera. I was really tempted to use it ungraded (beautiful soft look), but just added the tiniest hint of saturation, warmth and contrast to make it pop a touch more.
I didn’t run into any of the overheating issues some folks have. I don’t know if it’s because I’m using that fast Transcend 16gb class 10 SD cards… buffering is never a problem. Maybe too much buffering exacerbates the problem? Consequently I have run into a heat issue… but even then, it was surprising after reading how ‘bad’ it was. I was shooting in a dance studio. It was really hot and really humid (after a class of 30 or so had been dancing for two hours). I was shooting almost continuously without turning the camera off for about 30min when I got a heat warning. I simply turned the camera off between takes, and it went away. Perhaps with summer coming, it’ll pop up more… but I think if you just power-down between takes, it’s probably a non-issue. Also, if you work with people who’re used to film and not HD video… they are used to a little downtime between shots.
Training guide
This camera can also roll really quickly - very useful. I can snatch shots (like the final shot) where everything just looks lovely and it wasn’t a planned take.
I have found with the higher-resolution or bigger screen (wider, so more of the 16×9 shot over more pixels), that following focus is really not so tricky. I’ve shot a few things with unpredictable movement in the past week, and just followed it myself (no separate focus puller), by just turning the lens barrel (no follow focus rig). It’s not a perfect solution. A good follow focus rig would be much easier. And a good focus-puller pulling between measured marks with actors hitting those marks would be ideal. If the actor missed the mark, a nice 7-inch monitor would be nice, with peaking, to help the focus puller. But in a non-ideal world where you’re shooting fast and light – it’s actually possible to focus just on this screen.
We knew we would be slowing down a lot of the shots to give it a sort of imaginary, light, lyrical quality (I always need slow-motion to be motivated… otherwise it’s gratuitous). So we shot most of it at 60fps (some at 50fps to see if there was a noticeable difference in the motion. Shutter mostly stuck at 125 for ‘natural’ motion blur (the whole 180 shutter thing). Some of the faster movements we experimented with higher shutter speeds. I can’t remember which ended up in the video. Jello-cam just didn’t really seem to be a problem. So long as the footage is stabilized somehow (even a hand-held stabilizer or a shoulder rig or even a picked up tripod) makes a world of difference. Quick pans or rotational instability causes horrid jello-cam… just don’t use either, and you’ll be fine. I’ve never been a fan of whip-pans anyway.
Editing… I did it before I got the FCP plugin, so I just used an MPEG Streamclip batch to convert the files to Prores 422. Then I conformed them all to 25fps in Cinema Tools, then started editing. Colouring just done in Colour (wish they’d change the name to a less common word – would help in searches. Same as Motion).
That’s about it. If anyone has any questions, I’ll get back to people when I can.
DSLRs and misinformation
I see there's a lot of misinformation out there regarding DSLR cameras, specifically from video people. Generally they say things like 'the compression is terrible', and 'not broadcast quality' blah blah.
HDV for instance, is not the best quality either, and it gets used on TV (not a lot, I know). Same with footage from things like EX1s and EX3s. These people are essentially spreading stories to make people doubt the quality of these new systems that they tend to not own. They are trying to artificially create a 'pro-quality' distinction between DSLRs and their kit.
5Ds all the way down to 550Ds record with a 47megabit per second data rate. For a basic comparison of how much information there is, HDV records at either 19 or 25Mbps... about half the information! Even EX3s on the 'high quality' setting only record at 35Mbps, meaning DSLRs record on average a 3rd more data than even EX3s. Of course there are tons of cameras that record at much higher bitrates, but then we're looking at really expensive cameras. HDCAM or anything that compresses to a 4:4:4 format, or anything that records a raw image will definitely beat these cameras both in edge detail and number of line-pairs recorded. But they are significantly more expensive for very little perceptual gain in quality.
This 'pro' 'non-broadcast' stuff just annoys the hell out of me because of how blatantly untrue it is. H.264 doesn't make 'square blocks' on a high-resolution monitor. I've checked both on my 24' full HD screen and a £20,000 Cinetal colour grading monitor. It's nonsense.
You might have heard 'House MD' recently shot an entire episode on 5Ds? Don't tell me that the top people working on the top show in the world would shoot on something that was 'not broadcast quality'.
All that misinformation aside; there ARE issues with these cameras. It's too easy to set high shutter speeds, so people tend to slip-up and record jerky motion. It does have a problem with moire - creating strange swimming patterns with thinly-spaced repeating lines (stripy shirts, brick walls etc.). But note that this problem is not limited to DSLRs, all video cameras get it to a point. 10 years ago, it was just a rule... moire happens, shoot around it. Now at least we can throw backgrounds a tiny bit out of focus to avoid it. Aliasing (stepping on diagonal straight lines) is also a problem... but not much of a problem at 1080p. It is a more significant problem at 720p. Buy a cheap camera (550D) and buy an L-lens with the difference you saved on a 7D. When Canon eventually release a video-optimised camera based on a larger sensor, you'll sell your 550D for a fraction, and still have the good lens. If the lens doesn't fit, L-lenses have good resale value. Win win.
Just my thoughts on what's real and what is misinformation. If I'm significantly mistaken about any aspect of this, let me know – mistakes are where we learn.
HDV for instance, is not the best quality either, and it gets used on TV (not a lot, I know). Same with footage from things like EX1s and EX3s. These people are essentially spreading stories to make people doubt the quality of these new systems that they tend to not own. They are trying to artificially create a 'pro-quality' distinction between DSLRs and their kit.
5Ds all the way down to 550Ds record with a 47megabit per second data rate. For a basic comparison of how much information there is, HDV records at either 19 or 25Mbps... about half the information! Even EX3s on the 'high quality' setting only record at 35Mbps, meaning DSLRs record on average a 3rd more data than even EX3s. Of course there are tons of cameras that record at much higher bitrates, but then we're looking at really expensive cameras. HDCAM or anything that compresses to a 4:4:4 format, or anything that records a raw image will definitely beat these cameras both in edge detail and number of line-pairs recorded. But they are significantly more expensive for very little perceptual gain in quality.
This 'pro' 'non-broadcast' stuff just annoys the hell out of me because of how blatantly untrue it is. H.264 doesn't make 'square blocks' on a high-resolution monitor. I've checked both on my 24' full HD screen and a £20,000 Cinetal colour grading monitor. It's nonsense.
You might have heard 'House MD' recently shot an entire episode on 5Ds? Don't tell me that the top people working on the top show in the world would shoot on something that was 'not broadcast quality'.
All that misinformation aside; there ARE issues with these cameras. It's too easy to set high shutter speeds, so people tend to slip-up and record jerky motion. It does have a problem with moire - creating strange swimming patterns with thinly-spaced repeating lines (stripy shirts, brick walls etc.). But note that this problem is not limited to DSLRs, all video cameras get it to a point. 10 years ago, it was just a rule... moire happens, shoot around it. Now at least we can throw backgrounds a tiny bit out of focus to avoid it. Aliasing (stepping on diagonal straight lines) is also a problem... but not much of a problem at 1080p. It is a more significant problem at 720p. Buy a cheap camera (550D) and buy an L-lens with the difference you saved on a 7D. When Canon eventually release a video-optimised camera based on a larger sensor, you'll sell your 550D for a fraction, and still have the good lens. If the lens doesn't fit, L-lenses have good resale value. Win win.
Just my thoughts on what's real and what is misinformation. If I'm significantly mistaken about any aspect of this, let me know – mistakes are where we learn.
Subscribe to:
Posts (Atom)