Savannah Miller

Oh. I think it was just people's memory of what was better without comparing the two side-by-side. The camera was definitely a huge jump above compressed 5d3 video so people tend to remember it as better than it really was. Maybe it really was that good and I've just forgotten.

5d mark III has around 11 stops of DR. If you use the HDR mode you get a little more than that. It's interesting as you can do both HDR and Raw at the same time. From people that came from DSLR, the BMCC was very hard to use, whereas 5D mark III was a lot easier. But from a professional standpoint, not being able to playback anything and the overall hacky nature of the 5D3 was a bummer. Most people didn't have the skill to make the BMCC look good even though it was a better image.

Yeah. Magic lantern would also be better if canon didn't put slow SD card controllers in their cameras. A lot of the modern Canon cameras that are SD only such as the 6D, etc.have really slow SD card slots much slower than UHS-I standard so a lot of the higher resolutions are not possible. That being said everyone did seem to think ML looked amazing (and it did compared to compressed Canon video) but side by side even with the old RED MX sensor, it clearly lacks a little bit. Magic Lantern raw ONLY looks good because the nice color it has when processed with Adobe camera raw. Otherwise it clearly lacks in dynamic range and overall image latitude compared to even the 10 year old RED MX sensor. So compared to GH5, GH5s, even though it looks "better" I would guess that it's quite far behind.

ML Raw at 100 ISO probably has similar dynamic range to GH5 and it's RAW so it has some quality advantages. Color looks more like what people seem to enjoy working with, so indeed it probably is better, especially how easy it is to use with Adobe Raw interface. V-Log is likely harder to deal with and takes more work to get that Canon 5d3 look. But there's no mistaking this Pocket 4K will be much better.

I wonder if Blackmagic was single-handedly responsible for the huge drop in Cfast prices. You can now get Komputerbay 256gb cards for $250. They should be able to handle 4K 60fps. Otherwise, SSD should be able to do 60fps as well, and they're very affordable. You can get a 1tb Samsung t5 for only $349.99. That's probably the cheapest high speed recording available of any camera on the market. That USB feature is a game-changer and other brands like Z-cam were clearly surprised when that feature was announced and they said it might make it onto their cameras, but I doubt it's a simple firmware fix for them.

Some TV shows shoot prores XQ, so file size is not the issue, it's just that they don't like raw. Lossless raw might even be smaller than Prores XQ but the workflow is more expensive.

The guy who shot it was a member of the Blackmagic facebook group. He removed it hastily after Blackmagic requested it be taken down but others reposted. If it's truly 422 then I see no reason why the clip is only 12fps and contains so many dropped frames. Higher end productions like TV shows are shot usually 4:4:4 1080p. So I could assume 1080p 4:2:2 will likely be used when shots are needed with this camera. For Netflix I assume if this camera is used, maybe we'll see more 4K DCI Raw at 3:1 for VFX work although 4:2:2 Prores HQ I think fits within Netflix recording spec for 4K capture. High Capacity recording is nothing these days as hard drives are getting cheaper and cheaper. Really only reason against raw is the speed of the workflow as RAW is not much different from prores file sizes.

No. Cfast 2.0 can write 4K 60fps RAW at 3:1 compression with no issues. If anything SD cards especially the latest from SONY are getting quite fast too and can almost do 4K 60fps raw as well, but maybe there's a UHS-II limitation. The USB-C was intended for large capacity recording.

I think it does. RED Gemini also loses 0.5 stops of dynamic range when using the "low light" mode, but given they clearly say that the low light mode is for "low light only" and the fact that they generally overclaim their dynamic range, I feel like it's probably 1-2 stops as well.

If you look at the chart from the livestream, there's two ISO modes. Native 400, and 3200. Once you shoot 1600+ the Dynamic range drops to around 10 stops (vs 13) because it switches into the 3200 mode. The guy who shot the test clip confirmed he was using the Sandisk Extreme Pro UHS-1 V90 which is rated at 95 MB/s and can sustain around 80MB/s+ continuous. 4K UHD is about 88MB/s in 24fps mode, so there has to be some other reason for the dropped frames or maybe it's just that once a single frame is dropped, other issues start compounding and it gets worse and worse.

A couple things: The SD card that was used to record the "leaked" clips was actually quite fast. It was the Sandisk Extreme Pro that the shooter used in either his Pocket or Micro cam (not sure) so there was likely some other issue as to why the recording was dropping frames. Or maybe dropped frames just compound and create tons of problems causing the 12 fps mess. There was some obvious tint problems with the colors and settings were hastily changed. The clip was actually shot by changing the different ISO settings, so it's all over the place in terms of exposure and dynamic range. There's only 10 stops in the 1600-25600 range. It looks like the shots where people tended to prefer the exposure balance were around 1600-3200, likely the reason the highlights clip so poorly. Most people grading the shots are just not great at color grading and didn't correct for some of the tint issues, hence why the colors look "off." Plus on top of that, the shots that have more highlight retention are slightly underexposed too. On top of that, color shifts with ISO changes too, and especially when going from standard ISO, to low light ISO mode. So the standard ISO mode is likely a more "finished" color than that of the higher ISO modes causing further problems. If someone had just exposed for 400 or 800 ISO the shots would have likely turned out 10x better.

From the demo footage that was leaked, the audio did sound very good.

All I was saying is Extended video is more advanced than just a simple lut because it's heavily dependent on ISO, Color temperature, etc. whether or not you get the correct color results. From that notion, a simple HLG lut like the one included in the camera's firmware at NAB would work, but it's not as perfect of a solution.

They have an HLG lut included so you can view in the monitor/bake it in. I'm not sure exactly how it works but it's not a true HLG like a dedicated color science.

BMPCC4K does not have HLG. It has an HLG lut for monitoring. I could be entirely wrong on this but I believe that the cameras internal REC 709 output is not exactly as simple as applying a LUT because ISO and color temperature play a huge role in how the cameras internal colors are processed. So the Wide DR setting is more complex and takes all of these things into account when you adjust color temperature/iso to deliver the best colors. Depends on where you order from. If you order in the USA from B&H there are no taxes but your state might charge you some at the end of the year or not. Whether you want to risk not paying them is up to you. Kristian Lam from Blackmagic was using the manfrotto smartphone mount ($10) to secure an SSD to the camera. https://www.manfrotto.us/product/MCLAMP?gdffi=d4782d53a7a849ef83b3597b37ed37ac&gdfms=7A2335F3F7774B6FB20C16EA4002F5F2&gclid=Cj0KCQjw28_XBRDhARIsAEk21Fiz7d3ju_NtRSlNDk1DD09rEFkMiuNHmgSJXTma9e8xJV4LRSklwE4aAnNyEALw_wcB or this: https://www.amazon.com/Ulanzi-Tripod-Mount-Smartphone-Adapter/dp/B01EYFV6FY

Are you sure? The ursa mini 4.6K sensor (at least the one on the Fairchild website) claims it can run at 240fps full sensor readout. Then why does Blackmagic only run it at 15ms? Also, the pocket camera was 17ms and they improved it to around 13 with the micro.

Sony already has problems with overheating so that's probably why they can't run their sensors any faster. Even the latest RED cameras are fairly slow sensor readout even compared to Blackmagic. That's one of the compromises when you build a larger, higher resolution sensor.

Not exactly. Sensors tend to heat up when you scan them faster, so cameras will often run them slower. Not sure if it saves power too but that could also be another reason. They run them at the fastest acceptable speed that they can manage even though a lot of sensors can run much faster. On "The Resident" the micro shots are easy to spot. I don't know if it's just a wider FOV, but someting about them does look off at times. Micro shots match better in color than any other small camera would, but something about them makes them easy to spot. Ursa Mini Pro not so much unless there's an extreme highlight or something that clips a little harder than the Alexa would. But even then it's not like it's offensively clipping or anything, it's just maybe a little more. The strong halo/lightwrap around bright light sources might be a sign of the SLR magic APO primes, and the usage of the camera for me is another clue as to which camera is which. Overall though the look is maintained well across all cams. For the limited uses of aerial footage, if that's all from the DJI x7 then those shots are very impressive. That camera holds up well.

F55 is a really good camera. It's worth 10x the price of the ursa mini 4k in my opinion. I find it to be a 10x better camera. Maybe it's the user, but just watch "The Crown" on netflix. Beautiful stuff.

Dynamic range of the 4K model is pretty weak. If you compare it to they Sony F55 which is also global shutter, the performance is much worse. So I think the 4K global shutter model was a big step backwards in image quality compared to other Blackmagic cams. But because it was 4K global shutter, Aliasing/moire/rolling shutter artifacts were mostly gone so at the time it seemed better. Now looking at Ursa Mini 4.6K images you can see a clear difference in the fairchild sensors vs cmosis. Ursa mini 4k/BMPC/full size ursa are not relevant anymore with the pocket 4K coming out soon.

Yes. Latest Blackmagic releases no longer have rolling shutter problems. Ursa Mini 4.6K/Pro is decently fast. So is the new micro cinema camera. When they switched to 4K+ cameras aliasing/moire largely went away too.

If it's a customized variant of gh5s and terra 4K sensor then rolling shutter should be very fast. GH5s is around 10.5ms and the Terra 4K is 8ms. That's a top-of-the-line performance for any camera of that size, and honestly better than most bigger cinema cams too. Newer RED 8K cams are quite slow sensor readout times and so are the really large Mostro/Vistavision sensors as well. There's actually very few cameras that match that performance. 4K VFX is actually very slow. The way our pipline typically works is we ingest reels of shots directly from the show and then generate the plates ourselves from original camera files and edl/xml data. Shows not shot on Alexa maybe things are different, but this is the general workflow. 4K dpx frames are around 50mb each, so with 20 people all accessing a server requesting these large files, it's noticeably slower. We have insanely fast workstations, but that doesn't make a difference when you're server can't keep up. Only 4K shows that I do are netflix/hulu and occasionally there's a lower budget show that shoots on RED and they tend to do 4K post. But I think they master at 1080. Everything generally takes about 2x longer with 4K, but I have tricks that I can still work fast.

He answered many times on various forums that it's basically smaller and lighter so you can handhold it and get the shots you want. Alexa is a huge camera and hard to wield. Alexa Mini is smaller than Ursa Mini/Pro but it's a nightmare in terms of ergonomics so you have to rig it and then it's larger. So it's more/less a matter of which camera is easier to use to get the shots you need. He does use Olympus EM-1 Mark II as well on the show. Actually sometimes the Ursa Mini Pro is actually a BETTER camera image-wise too. It has much more shadow dynamic range vs the Alexa so it tends to look better in low-light. I'd also say it has slightly more FPN if you have to lift the blacks of your image (which you never do), but in the upper and midrange it's a cleaner image. In the scene he linked of Nevin and Hawkins privately chatting in one of the rooms, the Ursa Mini likely performs better than the Alexa would in the same situation.

Yeah. I work in VFX and every show I've ever worked on is Prores 4:4:4. Never had the luxury of an Arriraw show other than maybe a few Game of Thrones shots (not sure if they were raw) but I don't think it would make much of a difference. The only real difference between RAW and Prores that I see is that you can adjust your exposure using linear sensor values which makes it generally easier to just slide up and down the exposure. With log, it's a bit more tricky and using a color space transform to linearize helps too. But even then, that's not a huge issue unless you expose incorrectly which shouldn't happen. Maybe if you have moire/aliasing you can change how you debayer to help hide some of that too, but I don't work in that area so I wouldn't know how much that helps. Other people at work sometimes blame keying errors on the internal sharpening that some of our shows shoot with, but I haven't found it a huge problem.

Those shots look amazing, but the shot on top of the ambulance, is that done with gopro? Is it some sort of stock footage? That's the only shot that looks different from the rest. If Blackmagic sends only 12-bit, then that's not better than the 12-bit log encoded DNG. Plus a lot of Blackmagic's proprietary color science info is contained within their DNGs which might not work with Prores raw. Long story short the only real benefit of Prores raw over Prores 4:4:4 is the ability to change white balance which is not a huge issue because a competent colorist can change the white balance of prores 4:4:4 quite easily. Secondly, white balance is one of the easiest things to set correctly, so it's not a huge benefit in that area either. There's nothing wrong with shooting Prores and then for shots where you think you might want to change white balance extreme amounts, to shoot Cinema DNG. That leaves very little scenarios where Prores raw is helpful.