gloopglop

[media]http://vimeo.com/77268402[/media]

68GB worth of material was used to get a studio based test this finely tuned, with the cameras matched in post as close as possible. This effort to remove the variables of grading and camera settings leaves behind a truer picture of the differences in hardware capabilities.

The 5D Mark III raw (from Magic Lantern), if it were a film stock, would be Fuji. Warm vivid colours which may need taming a bit in post. The Blackmagic is more Kodak, cooler and more muted, it often requires the opposite treatment in post to the Canon cameras. The 7D is totally back from the dead - with Magic Lantern raw and the Mosaic Engineering VAF-7D tested here, it offers lovely image quality from a Super 35mm sized sensor, at a similar price to the Pocket Cinema Camera. The Panasonic GH3 - best of the standard system cameras out of the box without modifications does a good job keeping up with them.

The scene was lit three ways to test resolution, dynamic range and low light performance.

[url=http://www.eoshd.com/content/11350/depth-test-5d-mark-iii-7d-raw-vs-blackmagic-pocket-vs-gh3]Read the full article here[/url]

I think what Axel is saying, citizenkaden, is that if you want a nice 3d like chromakey, the background wrapping around the actors, you would build the set like that pictured.  The problem with such a set is that the side walls in green will reflect green light onto the actors and that will confuse the chromakey.  So you want the software to differentiate between green on the screen, and green that has been "spilled" onto the actors.  
 
When I first read about this I thought it was stupid.  Until I tried shoot green screen in a small room and it kept spilling onto arms and hair (that is, a green reflection).  If the OP can't place his actors far enough from the green screen he/she will run into this problem.  dishe's link clearly explains why 4:2:0 does not give accurate pixel level chroma for the software to work with.  Even 4:2:2 is not perfect.
 
In practical terms I would say this,
 
SCENE A: Actors are in big sword fight on some crazy tropical island.  The green screen is in a large room with plenty of room for lights, actors, etc.  A G6 would be fine.
 
SCENE B:  A man and woman are having an intimate talk aboard a "starship" with a window overlooking space.  Plenty of closeups.  The woman has fine, flowing hair.  Then xenogears BMCC is, if you ask me, would be the only chance you have of making it look real.

You are not an idiot, I guess my english isn't very good. With 'unless you work in a set like this, you can do with 4:2:0' I was trying to say that it is not good for 4:2:0, because of the overall spill. The sentence may have been grammatically wrong.
 
If there are roughly 4 pixels on the edges where green and foreground mix in 4:2:0 without any spill, you can imagine how many there will be with a lot of spill ...
 
As human being, with eyesight limited to 4:2:0, you probably see purple dots if you close your eyes in this studio. Personally, I have never been on such a set. But I can't believe that color keying will suffice to get a clean matte for the Peter Jackson team. There will still be some trainees who spend hours and hours on rotoscoping additionally. And you can detect another trick in all LOTR movies: The actors are always backlit. Read jghardings post here.

man, ive been trying to wrap my brain around the whole color sampling thing for awhile now, and ive understood it enough to be successful in practice, but conceptually the little illustrations in this article really helped me a lot, thank you dishe! 

It would not be any less accurate than a camera recording that resolution in 4:4:4 color. 
 
The accuracy of the color isn't in question- the resolution of it is. The accuracy of any given pixel is really a side effect of how much resolution each color channel gets. As I understand it, color sampling is reducing the resolution of certain color channels in order to save bandwidth on compressed video. The idea is that your eyes will see the difference in contrast more than the difference in color, therefore not every channel of color needs the full resolution. This is mostly true, as side-by-side the images look identical to the human eye, until you isolate colors.
So when we say 4:2:2, you are saying that for every 4 pixels, cB gets 2 and cR gets 2. If you isolate the red channel in After Effects, and you'll see the resolution is lower than the full image, with the borders around objects much less precise and blocky looking/pixelated Here's an example of this:

Top part shows the full color sample, bottom is the resolution of the actual color channels. 

But like any enlarged picture that looks heavily pixelated, if you scale it down the discrepancy between the lost pixels is gone, resulting in a resampled fine pattern around the borders instead of the blocky pixels. In other words, there would be no difference from a picture taken at that correct resolution.

As far as I can tell, this is why having higher color sampling makes a difference to chroma keying in particular, not because it is hard to see what is "green", but because the borders of color aren't as well defined, and it can wreck havoc on the edges of your key. Again, its not about color accuracy, but resolution of the color channel you are trying to key out. 
 
If this is going over anyone's head, there's a great write up here on DVXUser about it:
http://www.dvxuser.com/articles/colorspace/
 
Yes, shooting full res 4:2:0 and even 4:2:2 won't give as smooth of a key as 4:4:4, as the edges are inherently less defined by the very nature of color sampling. That doesn't mean it isn't possible, and that doesn't mean you can't pull of a great key regardless... especially if you are able to reduce the image being composited to clean it up. 

But this isn't just "smoke and mirrors", Andy. Its science. None of these cameras operate on magic. Understanding how it works in order to make an educated decision vs. just recommending perhaps a great all-around-camera is important! He isn't asking what's the best all-around camera right now! There are cameras around the same price point as the G6 that are capable of recording more color information per frame.
I understand what you are saying about RAW file management being a beast- I agree. But a 4:4:4 compressed image, or heck even a 4:2:2 will be far cleaner than a 4:2:0. And that's my own real-world experience backed up with science!

Video from the A7 - moiré festival.

http://youtu.be/oh3dKWglaXQ

I was going to say maybe they were Sony fanboys not having an objective view.  But after reading the thread I actually don't think that is the case.  I think the people there are videographers in the traditional pre-dslr sense of the word.   They are seeing this cam as a convenient all in one workhorse for ok videoey looking video.   Whereas people here are more interested in ultimate image quality.  We judge dynamic range, detail and tonality to the same standard a still photographer does with stills.  It doesn't matter to us if a camera is twice as convenient: If it doesn't represent a significant increase in image quality for its price point then we aren't interested.

omg....france is covered with chemtrails !

If they are doing on the new full frame sensors the same as on the RX10's new 1" sensor then video is going to be amazing.
 
The RX10 does a FULL READOUT producing 20MP 6K raw stream at 60fps.
 
The new Bionz X as featured in the A7 and A7R is powerful enough to then subsample that 6K down to 1080p before writing it to the card.
 
That means no moire or aliasing, no line skipping on the sensor, no great loss of dynamic range from a poorly subsampled sensor output.
 
If your credit card is not ruined by now it may well be when the RX10 sample videos start trickling out.

I stumbled on this video and it displays perfectly the biggest flaw that I had the displeasure to work with on that camera:
 
https://vimeo.com/76824060
 
*** EDIT WITH WORKAROUND ***
 
There is a clever trick to work around this issue.
Thanks to Axel for digging it up.
 
Since the bmcc records raw, and when shooting raw you keep the actual recorded footage of the sensor, and an adjustment such as iso, is only kept written alongside the file in the metadata, what you can do is lower the iso until your image becomes decent.
 
In post you can then bring it back down to the native 800iso.
 
Not a perfect fix, as we would need to know how much of an increment we would need to exactly counter the screen displayed over exposure, but a nice workaround.
 
*** EDIT WITH WORKAROUND ***
 
 
When you work as a dp on a film,
 
You need the best visual feedback possible through the monitor in order to frame and work on your lighting adequately.
Does not need to be perfectly accurate, but the more the better, especially when working with lighting equipment to adjust your lighting.
You need proper feedback of how soft your fill light is, how strong your rim light and with what fall off etc...
 
And you need to have the vectorscope, as the visual feedback fools you in terms of recorded data, to get the best exposure possible on your camera, and make decisions on what you want to preserve, whether highlights or shadow or a balance of both, if the situation demands it.
 
Now with the bmcc, you have to extremely over expose your image to get best image quality.
The result is a completely washed out image through the monitor, good luck properly fine tuning your different lights with that.
 
And you have to disregard the vectorscope, because if you use it to expose properly you will actually be underexposed.
You have instead to completely over expose until you see zebras, meaning no proper control over exposure.
 
This is a major flaw, no professional camera should disregard the vectorscope and provide such inaccurate visual feedback.
 
Now let me know if I missed something here

Camera threads always seem to get hostile..  It is a mathematical fact.  :rolleyes:
 
I will say this:  (and not to get off topic) ...I've been through - "3" - bmcc 2.5k cams in 4 months... Each replaced by BMD support.. each an acknowledged legitimate problem with the cam.  My ego doesn't have enough in the bank to stand off if someone tells me the problem is operator error, so I'm all about a solution considering the shipping is $50 every time I send it to California..  But 3 in a row is either an issue in the Q.A. area, ...or really shitty luck.  The first one had noise in the green channel so bad, all you could do is barely push it, much less add any post sharpening.  The second had the back focus issue.  My last one produced what looked like bad web compression on the rear display with blocky shadows and smeared details, & "rolling noise" waves that would roll up the scene in dark areas.  (Truly a weird problem to have, but others had it also: http://forum.blackmagicdesign.com/viewtopic.php?f=2&t=12588)
 
Yes the pocket cam is a different camera altogether of course, & I'm not bad mouthing BMD.  Plenty of people have perfectly functional cameras from them.  (just not me, lol.) In these early days of release however, I'm just saying if you get a Blackmagic camera, put it through its paces before your return window expires..  Make sure you get a good one while you can!... Even if it starts feeling like harassment.  They're cheap, everybody wants one, and BMD have a delivery reputation to mend.   

Hello friends, I am new here, long time lurker. Thank you sincerely for the discourse; it's extremely educational to me.
 
After a brief analysis of these .dng files in ACR my reaction is, in a word...pleasure~! These images feel naturalistic to me in a way that the BMPCC prores for the most part does not, not to shit on the prores and say it's worthless––it's very fine. But there is a certain character to these that I far prefer, perhaps because of the added effective dynamic range [when properly processed]? I'm really not sure; I appreciate all of your comments as they help me get a grasp on what is going on here. I would like to understand the 'why' of digital cinema as much as possible and ultimately I am just a beginner.
 
Anyway, again, ACR produced pleasing results for me with minimal effort, and a little noise reduction goes a long way if that's the look you're going for. I didn't find what I would consider to be an unexpected amount of noise in any part of the images based on my perception of the exposure and lighting conditions.
 
I'm looking forward to seeing more raw imagery from the BMPCC and I look forward to seeing the analysis here on this board, thank you guys.
 
Cheers!

You are right. It is a default Photoshop preset applied to every raw file. And setting it from "25" to "0" practically eliminates what is described as "noise". Keep in mind, that camcorders with intelligent compression algorithms also develop raw images under the hood, that they add gain, NR and sharpening. But sharpening must be applied to the BMPCC images, obviously. Perhaps there are better methods (.i.e. in Resolve, as a last step of grading then, there are also discussions everywhere whether to apply NR first or last, plainly the right answer is first, the node model makes it easy to get the order right - only that there is no denoiser in Lite :( ), but I guess it just needs some understanding on how the tools work:
 
 
 
This is from here. Hystery begone!

One thing to note.  ACR applies sharpening by default on dng files.  So if you haven't zeroed it the noise does get lifted a bit.