Ryan Holmes
3
COW Coins
Forum Replies Created
-
[Tom Wille] “Do you not think it has anything to do with the lens? That’s kind of the common denominator to all of the problems, but I’ve never heard of that being an issue for anyone before.”
Tom the Canon 6D is notorious for soft video. The Rokinon 14mm is a great little lens. What you’re likely experiencing is the bi-product of upscaling soft, moire footage. So the way Canon DSLR’s get a 20MP image sensor down to a 2MP HD image is by a process called “line skipping.” Line skipping is sampling various “lines” or pixels in order to achieve the image you see on your screen. The problem is that this approach almost always tends to produce a soft image with a high moire effect (tight patterns appear to crawl).
DSLR’s as a whole don’t shoot “crisp” video pictures because they are either line skipping or binning to get the sensor down to HD resolution. Compared with a dedicated video or cinema camera like the Canon C300, which is taking an 8MP image and downsampling the whole image to HD. So you tend to get very sharp content because it’s not cheating it’s way down to HD. To your DSLR I would expect that you’d see better (i.e. sharper) results in 1080 than in 720 since the line skipping won’t work as hard to get to 1080.
If I can be frank, your lens is fine. It’s the DSLR body that is the problem. Shoot the same scene, same lens with a C100 or C300 (or Sony FS7, Blackmagic URSA, etc) and you won’t see this problem. Canon’s DSLR’s have video added as an afterthought. The technology they use (and Nikon) to create that HD image makes a bunch of sacrifices. Yes, they shoot video. But that video is far from ideal. Do a little exploring on the Internet and you’ll understand more of the downsides to DSLR’s.
https://fstoppers.com/video/canon-6d-showing-hideous-moire-problems-video-4113
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
[Chris Cella] “I’ve also read that Premiere doesn’t like to export H.264 1920×1080 videos. Anyone heard anything about that?”
Never heard anything like that before. I export 1080p24 and 1080p30 sequences every week with no problems.
[Chris Cella] “I would appreciate any suggestions as this has stopped my workflow for over a week now and my superiors are breathing down my neck. Thanks for your time.”
I would try what Joe recommends. You can also try exporting from directly within Premiere and see if that gives you the same error.
Additionally, you can breakup your sequence into smaller portions to see if those “sections” export out. If one of them fails then you probably do have something screwy in there – corrupt file, broken render, etc.
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
[Don Mick] “Ideally, I want the highest Quality .mp4 file from this file so that we may send off to festivals, both for prepping a dvd and for ftp/cloud uploads to festival servers. I would also like a decent filesize.”
So these 2 things are in constant struggle. Quality vs. file size. They aren’t necessarily mutually exclusive, but the bigger the file size generally the better quality your final product.
[Don Mick] “I wanted to export as a high quality mp4, so i matched most settings except putting level at High 4.2. can anyone tell me if this is a good recipe and if not, please recommend ideal settings to get a high quality MP4, codec H264 from a Quicktime, codec ProRes 4444?”
There’s no magic bullet for this stuff. The presets in Media Encoder are a good place to start and you can tweak from there. Match settings is fine, but you’ll want to focus on your data rate. The Mbps is telling the software how many bits to throw at an image. The more bits the better the image, but the larger the file size. The level is a specification dictating what the data rate range can be, and also what devices the file can play on (older devices can’t play the newer, higher levels).
You’ll also want to take a look at CBR vs. VBR – or Constant Bit Rate vs. Variable Bit Rate. In general a CBR will give you the best quality and the biggest file size. VBR typically will give you acceptable quality and a smaller file size. A CBR approach is going to take the data rate you’ve set and indiscriminately through it at the image. So if you specify 5Mbps, all 5Mb’s will be used all the time. In a VBR setup, the software will attempt to use only what is necessary to define the image. So if it can use less, it’ll use less. Tradeoffs for VBR – sometimes the quality isn’t as good, and it takes longer since it’s doing more estimation than a CBR).
These are all very general rules. There are caveats to all this, but hopefully that points you in a direction. Also take a look at YouTube and Vimeo’s compression guides. Vimeo has tutorials on their site for all encoding software. This may give you some additional help:
https://support.google.com/youtube/answer/1722171?hl=en
https://vimeo.com/help/compressionRyan Holmes
http://www.ryanholmes.me
@CutColorPost -
[Blaise Douros] “Ryan, he wants to use the software intervalometer in Magic Lantern.”
Right, ML can act as an intervalometer. For me, I would rather run an actual intervalometer as hardware based approaches tend to yield more consistent results and less battery drain than software approaches. You can pickup an intervalometer on Amazon for under $20.
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
Louis if you want to shoot stills on your 5D2 you don’t need ML to do that. The 5D2 can take RAW or JPEG stills which you can time-lapse together.
Magic Lantern (ML) is meant for those shooting video that were frustrated by Canon’s lack of attentiveness to a quality image. ML hacks the firmware of the camera to enable a user to shoot video in what is essentially a RAW video mode at 24 or 30 frames per second. Shooting RAW video is in no way “economical” as the data rate needed to sustain that quality level chew threw cards quickly (64GB gets you about 12-18 minutes of content).
But if you’re just wanting to do stills your Canon 5D2 can already do RAW stills (or JPEG) and can accept an intervalometer.
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
You need to install the appropriate codecs for Premiere to be able to decode and playback what you captured. As Bill stated, MXF is just a wrapper. The analogy I use is gum. The wrapper that goes around the gum is generic, but the flavor of gum inside could be anything – Juicy Fruit, Big Red, Spearmint, etc.
The MXF file is a wrapper, like MOV. The codec inside is what matters – DNxHD, ProRes, Animation, JPEG200, etc. are just different codecs that go inside of a wrapper. So you need to determine what codec you used to capture your archived footage and then make sure that’s installed on the Premiere system.
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
It made duplicates because you probably already had all the presets installed in Media Encoder. If you’re on a Mac in CC I believe they are bundled in now as soon as you download. So importing the ProRes presets will just create duplicates for you.
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
If he shot Magic Lantern RAW on a 5D3, he should have 14-bit dng’s to work with. The camera is 8-bit when shooting internally under the Canon firmware. ML bypasses all that and gives you the image straight off the sensor. So working in something like a DPX file is completely reasonable (or even a ProRes 4444 XQ codec).
As you note, Dustin, either workflow works. It’s what you and your team is most comfortable with.
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
[Grant Peacock] “I’ve also learned that if the outboard recorder is chosen and set up properly, it will allow for a less compressed version of the material to be captured. “
[Al Bergstein] “Not sure the footage will be better looking than the C300 native, but it will go directly to an SSD in an editable format like Apple’s or Avid. “
The footage internally in the camera is compressed to a 50Mbps, 4:2:2, 8-bit XF-Codec using a MXF wrapper. If you attach an outboard recorded the SDI or HDMI ports will give you uncompressed, 4:2:2, 8-bit video. So you’re bypassing the internal camera compression. If you record to a Blackmagic, Atomos, Convergent Design, etc recorder you’re essentially getting that uncompressed signal and then compressing that into a DNxHD or ProRes codec. Both of those codecs run higher data rates and are more forgiving with editing, color grading, and vfx.
I run C100/C300 and only use the internal cards as a backup recording. All recordings get run to either a Sound Devices PIX240 or a Convergent Design Odysey7. The Convergent Design screens are absolutely gorgeous. Works for focus checking, framing, color matching, and recording. Some of the new Sound Devices recorders look good too like the PIX-E5 and PIX-E7.
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost -
[Grant Peacock] “Some allow for lateral movement during driving the card into the slot that will cause misalignment. One now has a bent pin from regular, careful use. My immediate take away is that the minimum number of card readers to take on the road is 2. Lesson learned, under difficult circumstances!”
This is the inherent nature of CF cards. Since it is a pin based design you’ll always, always, always have the potential for bent pins. There’s no way around it, no matter how careful (even a possibility in the C300!). Best advice, carry multiple readers and always have a credit card on hand. Credit cards are about the perfect size to slide in an attempt to rebend the crushed pin…and if not, you can buy another reader!
Another option is to purchase an external recorder (Atomos, Convergent Design, Sound Devices, etc) and record externally. These devices typically use some form of SSD and are more robust in dealing with the constant in/out nature of recordable media. You may also look at the BlackMagic Video Assist which was just shown at NAB. $500 with a screen and records all versions of ProRes and DNxHD to SD card (I think it ships sometime this summer).
Ryan Holmes
http://www.ryanholmes.me
@CutColorPost