I received a loaner BMPC 4K for the weekend and did a few test shots.
It's an interesting machine, not without faults, but quite wonderful if you keep it in its sweet-spot (which is anything under ISO 800)
Jump to content
28 replies to this topic
Posted 25 February 2014 - 07:57 PM
Interesting, the REDs are not Global shutter cameras unless you add the Motion Mount (2-Stops) nor is the Alexa the sensor BMD is using in the 4K is cmos and could be more sensitive if it were run in it's rolling shutter mode instead of the global. It's pretty decent for a $3k "4K" video camera I nitpick but most bayer mask cameras only resolve 2/3 of the overall resolution i.e. this "4k" camera has 2K of green and 1K each of red and blue so goes the marketing....
Posted 25 February 2014 - 08:55 PM
Nice review and video, Adam!
Three minor suggestions:
- You might add that the BMPC-4K's flavor of "4K" is "UHD 3840 x 2160".
- You might add a link to AbelCine's "FOV Comparator", which now includes the BMPC-4K.
- You might add a link to the list of "BMPC-4K Shooting tips from early users" I'm maintaining on BMD's forum.
Thanks again, and cheers!
P.S.: I hope it's OK I've added your video & review links to my blog.
Posted 25 February 2014 - 11:48 PM
I thought the 4K was a CCD?
Alexa does have a shutter similar, in the studio model (no stop loss either, though rather expensive)
I thought it was about 72% for a bayer; but truthfully, are we even counting anymore? If anything, I think we often have far too much resolution anyway-- hence why so many people are gravitating towards "vintage" lenses. Not that there is anything wrong with that.
Posted 26 February 2014 - 02:03 AM
No the BMD 4K is weird, they use a CMOS chip which has the ability to be run in Rolling or Global shutters, they chose to set it to Global only which reduced the sensitivity significantly. I thought the Alexa studio was a "Real" shutter i.e. mechanical and that all of Arri's ALEV chips were cmos without Global though I am most familiar with the Alexa M from my other other job as an Aerial tech.
Posted 26 February 2014 - 02:06 AM
I think you may be right that the rest of the Alexas are also Global CMOS. I could've sworn the 4K was CCD-- but I can't find anything on it a all it seems currently-- though i didn't look too deeply.
Posted 26 February 2014 - 07:10 AM
People have been assuming that anything which has a global shutter is necessarily CCD rather than CMOS, which is not so. Global shutter (on CMOS at least) requires a single extra transistor per pixel, occupying space that could otherwise be pixel area, hence the poorer sensitivity and/or dynamic range on global-shutter cameras. This would be the case regardless of whether the global shutter feature was in use or not, although deactivating it might have other, similar effects that could obscure the issue. At risk of self-plugging, I wrote an article dealing with this in light of recent developments in multi-layered sensor tech which may be of interest.
Also, electronic global shutters are - in theory - slightly imperfect. A transistor can blank a pixel to only about a 50,000:1 ratio, with the result that very bright highlights (the sun, maybe) could punch through an electronic shutter. This is unlikely to be much of an issue in practice, though.
My understanding is that Alexas are all the same sensor and thus a (very fast and largely un-objectionable) rolling shutter, obviated on mechanical shutter models.
I have seen Alexa footage with flash banding, which I view a a critical flaw which should be considered as making the footage unusable.
Posted 26 February 2014 - 09:25 AM
What battery solution were you using in that shoot Adam?
Posted 27 February 2014 - 04:30 PM
Peter: Thank you, good feedback. I'll update my post with your suggestions. Please feel free to use the post however you wish.
Posted 27 February 2014 - 07:10 PM
Sorry, I'm still not impressed. And that is not a knock at the video-makers. But every sample video I've looked at has shown the limitations of the camera. I see very flat images with extremely poor contrast. And the definition in the shadow areas are horrible. Put an image from that camera next to any 35mm film strip and the lush texture of film will blow it out of the water every time.
This does not look anything near 4K and this product is advertised as having a Super 35 sensor. Still looks like low-end video to me.
Posted 27 February 2014 - 07:18 PM
Harsh. I still have this one here and will until Monday; do cite some stuff you've seen that looks iffy and I'll see if I can go shoot something and figure out what's going on.
Posted 27 February 2014 - 08:35 PM
It may be harsh, but I'm not happy with a product that's being advertised as having a Super 35mm sensor when it produces images that are hardly even up to a 2K scan of 16mm.
One video to look at is Adam's posted in this thread http://vimeo.com/87615576 Take a look at the color saturation & contrast (or lack thereof.) Sorry Adam, but in my view the issue is with the camera, not you. Another one to look at is http://vimeo.com/86359257 (also posted in this forum.)
The one that I've liked is http://vimeo.com/86776681 posted by Moises Perez (another one of our members.) It has very nice images, but they were all exteriors. What I'm seeing is that this sensor functions far better under natural lighting than under artificial lighting.
Phil, can you shoot some exteriors under natural light & then some interiors under incandescent light so that we can compare the results?
Edited by Bill DiPietra, 27 February 2014 - 08:36 PM.
Posted 27 February 2014 - 08:40 PM
Why not just do the same INT lit by the primary instruments, HMI, Tungsten, Kinoflo, and LED. Perhaps set up so you can "mix" them in the shot and roll through the white balances as well as as a Key on the talent .
Posted 27 February 2014 - 08:47 PM
Bill, I don't disagree with you. It was a quick test and I didn't do any light sculpting, it was really an editorial of what was happening.
Had I properly lit it, there would be more to see, but this was straight prores out of the camera. I wish RAW was enabled (coming soon, with a firmware update...) so all I did was mostly an 'auto correct' in Resolve and that's it. I wanted people to see what came straight from the camera.
When the RAW is enabled, I'll properly light some scenes and let's see what that does....
Posted 27 February 2014 - 08:51 PM
It only has 12 stops of dynamic range, and while I didn't test to prove that, the results suggest to me that 12 is a bit of a stretch. Shadows get crushy really fast. That situation is exemplified by the fact that it's hard to expose for maximum range due to the lack of realitime or onboard histograms. There's just the zebra warnings to work with, not really that idea whatsoever.
Posted 27 February 2014 - 11:19 PM
In any case, thanks for doing the test. Since I've been considering venturing into the digital realm, it's exactly what I waned to know about the camera's capabilities.
Posted 01 March 2014 - 01:25 AM
Chances are the BMPC-4K footage you've seen that looks "flat with extremely poor contrast" was either graded to look that way, or was shot using the cam's log-like mode and left that way. A major feature of the camera in its current form is its mastering-quality ProRes HQ 10-bit 4:2:2 recording format, which is not known for being limited to "flat with extremely poor contrast". ProRes HQ allows enormous flexibility in how it's graded in post.
(BMD will release a free firmware update to add 12-bit compressed RAW recording to the BMPC-4K "soon").
As for "natural light" and "artificial light": Some artificial light sources (such as quartz-tungsten) can be color-corrected on set to almost duplicate natural sunlight, and HMI sources are close enough to be considered "daylight". Most modern camera sensors perform best with "daylight" ~5600K color temp lighting (not tungsten ~3200K), whether it's real sunlight or a high-quality artificial source.
Concerning sensor size, most camera manufacturers refer to sensors that are within a few millimeters of S35 16:9 film format to be "S35" size. Blackmagic is not the only one.
I don't think anyone expects a $3K video camera to meet the capabilities of film. The BMPC-4K is one of the few "S35" video cameras available at any price that has a global shutter (free of rolling shutter artifacts), relatively clean high-resolution (UHD 3840 x 2160 and HD 1920 x 1080) with far less aliasing & moire compared to similarly-priced cams, very pleasing color science, unusually high-quality recording formats, includes the full versions of Davinci Resolve and UltraScope, 6G-SDI/HD-SDI live output via a locking BNC connector (not stupid HDMI), balanced audio inputs, and has a built-in uninterruptible power supply battery (a feature that certain other manufacturers charge dearly for).
I received my BMPC-4K yesterday and look forward to using it productively. I'll be using both its 4K and 1080p recording modes to create high-quality, cost-effective 1080p edits for broadcast and the web. Given the camera's low cost, it'll pay for itself in a few days of use.
There's no 1 perfect camera ideally suited to every production.
Posted 01 March 2014 - 07:48 AM
I do hate it when manufacturers try to call a camera a Super 35mm digital camera. That's just annoying and causes all sorts of confusion (which is presumably what they are trying to achieve!) Most blogs etc tho, seem to talk of cameras with Super 35mm sized sensors and I'm totally okay with that.
The worst was a couple of years ago or so, when Sony were trying to promote a camera as being a 3perf digital camera which was fric*** ridiculous! I was like "Can you show me where the holes are?", "er, which holes?", "The 3 perforations you were talking about?"
Posted 01 March 2014 - 08:08 AM
I think a lot of the stuff is as you suggest is ungraded although the stuff I've seen that was graded also looked fairly awful.
The video at the top of this thread is one of the best videos I have seen from this camera so far.
The only other video I have seen come from the 4K Blackmagic that I thought looked nice was the one with the octocopter in the snow! That was really impressive in a lot of ways and surprisingly was ungraded. Maybe it wasn't recorded in log tho?
I'm fairly unimpressed by everything I have seen so far tho, and the main problem for me is the pastely colours and lack of saturation. When I have seen people try and correct for this so far the results have been harsh and nasty.
As Peter suggests, we can't expect a video camera to come anywhere near the look of film. The nearest anyone has ever come to achieving this was the Sony F35! (Followed closely by the original Varicam and the Alexa). I wouldn't say those examples were that close either but everything else is so far off from film it's another world.
I think the 4K blackmagic might be one of those cameras where you have to work to its strengths, although I'm also willing to accept that the problem with the videos out there so far is not the camera but the people operating it!
Also tagged with one or more of these keywords: blackmagic, 4k