Blaise Douros
Forum Replies Created
-
It’s going to be 8-bit color, very likely either in Rec709 or sRGB.
-
Exaaaaaaactly. A drive full of raw (unedited) footage is very different from footage in RAW format…
-
This is what stabilization is for. Use the point tracking stabilizer (so you have control, vs the warp stabilizer which is automated), get the footage stable, and you can then match those sections up.
-
The missing piece is that your footage isn’t an edit-friendly codec.
MP4 h.264 is a Long-GOP codec: basically, it means that every four or five frames, the codec records a “real” frame (called a key frame), and then the next four frames are just a record of what changed since that frame. So if you’re scrubbing the timeline, if you land on a key frame your machine is happy. But if you land on frame number four, that means that your machine actually has to process five frames: the key frame, plus each successive frame until it gets to the one displayed. Couple that with a 4x increase in the amount of data from HD to UHD, and you’ve got a recipe for your scrubbing to suck.
You can do one of two things: first, you can transcode to a lossless intermediate codec like one of the ProRes, DnX, or Cineform flavors, at full res, and edit and deliver from that. This will increase your storage space, but the files will be easier on your machine to edit.
Second, you can use Premiere’s Proxy workflow to generate lightweight intermediate files (usually in a lower res and also ProRes, DnX, or GoPro Cineform), which you use only during the edit process to speed up your machine. Before your final export, you relink to the original files, and the machine uses your original footage to export the final version.
“But that’s so much transcoding,” I hear you say! Yeah, I know. But that is the price of having a more responsive timeline.
-
Are you talking about an exploded view, where everything comes apart? If so, AE can’t do it natively. You will need to use a 3D app like Cinema 4D or Maya; you would need a 3D model of the digger, with each piece set up to be animated on its own. You can do the final composite in AE, but AE does not have the ability to model or animate 3D objects.
-
I mean…ProResHQ at UHD rez is 707 Mbps per UHD stream, according to Frame.io. According to AJA DataCalc, it’s about 123 MBytes/sec, which would be well within the read/write speed you mentioned.
However, you didn’t specify how fast your connection to the QNAP is, nor how many streams of video you have on your timeline. So if you’re running Gigabit Ethernet, two streams of ProResHQ at UHD would max out your bandwidth, right out of the gate. Dropped frames would be expected. Not so much if you are on 10GbE, unless maybe Jumbo Frames aren’t properly set. If you’re attaching directly through Thunderbolt, sometimes the speeds are not nearly as advertised, either.
Did Bob Zelin set up your QNAP? Have you given him a call, if so?
If you are connected via GbE, try using something like ProResLT or ProRes Proxy, or a lightweight Cineform flavor as Greg suggests. That would cut the data rate, assuming that your network is the bottleneck, here.
-
You’re right, mea culpa. It appears that at least some EF cameras have an indent that don’t allow -S lenses to be mounted.
-
I mean, no one has even brought up the fact that we’re dealing with exponentially more data. Not even as a figure of speech, either: 4K is quadruple the data of HD, which was quadruple the data of SD.
The upshot being: DV had a 25 Mbps data rate at 29.97p for 4:1:1 color, while my FS7 shooting UHD XAVC at 10-bit 4:2:2 with more advanced compression still has a data rate of 300 Mbps, not to mention the strain that decompressing that much larger amount of detail puts on the processor.I think you may be looking back with somewhat rose-colored glasses. When I first started my first real job as a post-production assistant in the 2000s, the performance of FCP was abysmal. Rendering left and right, transcoding to one of only a few formats that really would play back at anything close to real time, STILL having to go to an online house to actually finish a show in HD since FCP wouldn’t really play back DVCProHD in real time so you had to edit in SD and deliver EDLs to an Avid shop to finish. Capturing from frikkin’ TAPES? Don’t get me started on broken timecode capturing. Ugh. It was horrible.
Today, we have file-based workflows, practically automated proxy generation, format-agnostic NLEs, and mind-blowingly incredible camera capabilities for absolute chump change. My company’s FS7 setup cost under $10K, and it is unbelievably more capable than the $100K Varicam setups we shot when I was in my earliest days of real production jobs in the 2000s. And before that, as a kid, I used to tag along to my dad’s office, where they had a 40 BetaSP and DigiBeta deck machine room wired into three hardware edit suites where you had to set in and out points with actual buttons. That’s how I learned to edit–I’m probably the only person my age who remembers that stuff.
All that to say, the gentleman doth protest too much, methinks.
Also, try generating ProRes or DNxHD proxies, if you aren’t already. If you’re trying to decode h.264 in 4K in realtime, your processor is choking under the strain. Premiere will connect and disconnect them with a click of a button, and it’s usually just an overnight render.
-
EF-S lenses can be mounted on an EF mount. However, the -S denotes that it is only designed to cover an APS-C sensor, and will produce a vignetted image if placed on full frame. I don’t know that there is any potential mirror damage, as the EF and EF-S mounts are the same, but it will not provide full sensor coverage.
-
It’s definitely rolling shutter jello. The reason it doesn’t appear on your last clip is because it is shot on a shorter lens–the tiny movements of the camera and body in the wind are amplified greatly when shooting a long focal length, resulting in a “bigger” movement of the image relative to the sensor. When the camera jiggles during the exposure, you get the rolling shutter jello that you’re seeing. In my experience, the lens’ IS is not really designed to cope with this for video, so it doesn’t help much.
Warp stabilizer may be able to help with this–set it to 5-10% stability (rather than the default 50%), enable Detailed Analysis, and see what you get. It’s a bummer to have to apply it to a lot of clips, but it’s better than delivering a product you’re unhappy with.
A beefier tripod (or hanging a weight from the center of a lighter tripod) may help a bit!