peederj

    Congratulations but I don't see why you don't liquidate and exit all the m43 stuff while it still has value. No one's going to want moire and two redundant lens systems when they can have a second 5D3 body at the price for B/backup and invest in only one system. The only possible justification would be the use of shorter flange distance glass you might come across, but you could rent e.g. an FS700/Speedbooster for that and get 240fps in the bargain.

Well if you really want to shoot big-screen features with camera systems under $10K, I think most will prefer 4K+ resolution in MJPEG over 1080p RAW. A 4K MJPEG with a log gamma is going to look pretty damn good even with the compression and limited color accuracy...compare 5D3 stills JPEGs with 1080p stills. Then add Neat Video and a LUT and you can go ahead and crop and recompose and stabilize in post like the big boys and still have awesome resolution for screens over 10 meters. Especially if they can do the full 5760 x 3840.   I really can't see anything stopping ML from giving us 1DC or even better quality with the 5D3. ML is open source so someone could do it and it would be impossible for Canon to stamp out. Essentially just run native res JPEG stills out the CF port with the mirror and shutter locked up in live view mode. The camera has to be able to do it, Canon just didn't enable it to be done 25 times per second for less than $10K.   That will be the next big story. 6K for $3K available and working at your local camera shop. And an easier workflow for that than the 1080p RAW.

First of all, they are called "Cinema EOS" cameras not "TV EOS." With "Cinema Lock" for log gamma shooting. They are better in every way to DSLRs for shooting films at least before the RAW hack and still in most ways after it.   Also, 1000x CF cards have only been available a little over a year. So the speed of memory writing needed to capture 1080p RAW or as I predict 4K or even ~6K MJPEG simply wasn't there "for years." Canon didn't put a speed governor on the CF slot. So ML could come in and open Pandora's box.   I'm happy about the RAW hack and will also be happy when they get compressed 4K+ running. I own a 5D3 and have shots that those modes may be best used for. I also think my C100 + Ninja 2 is going to be a better option for filmmaking overall and for very little additional cost compared to kitting out a RAW-shooting 5D3 with dual system audio, NDs, side and top handle, peaking field monitor/EVF, and oh a huge stack of very fast CF cards. The C100 likely shoots better low-light, higher resolution and sharpness, and probably has more dynamic range at ISO 850 than the 5D3 RAW has, even though that will only have 8 bit fidelity in a log gamma as opposed to 14 bit linear on the 5D3 RAW. It also has one of the best built-in IR filters in the business, I almost never have the fan running in typical use, it's gorgeously weighted and balanced for hand-held shooting and is no worse than the 5D3 for mounting on a steadicam. Batteries last for several hours and change without dismounting the camera.   It's a lovely design and it's a lovely image. 5d3 RAW is a decent design and a lovely image too. But starting from zero (and I know many of you already have DSLR rigs you've sunk large money into) I really think most of you would prefer the C100 + Ninja 2 for about the same cost. I think many of you will prefer the 5D3 1.2.1 firmware + Ninja 2 over the ML hack, also at about the same cost (vs. all those fast CF cards and drives to offload to). Sacrilegious to suggest that here on the site that celebrates hacks above all else, but I really think that's true. And I have no interest in saying so other than my preference to keep things real.

Or actually wait. Native res on the 5D3 is 5760 x 3840.   Slay the Dragon.  :o

  ...or left *facepalm*   Hilarious suggestion that the Dominant Order is Threatened by this Game-Changer that must be Put Down At Once for the Love Of All That Is Good And Holy. A Canonic Conspiracy to oppress the lowly pixelpeeper at all costs.   And as for "those who dare win," I understand the cadavers up at the summit of Everest are more bother than they're worth bringing down for a proper burial. Which is what a director will make of you if their project is screwed up by the camera in your hands even if it was their idea to shoot that way.   Anyway to matters at hand. How will the 5D3 RAW work its way into a practical production? I would think the size of the sensor and the high dynamic range is going to be good for landscapes and architectural/establishing shots that need to be done in realtime. DSLRs are already used for timelapse for these reasons, and Luke's "Bodie" example aptly showed the earlier ML HDR video trick doing the same. Those are low-stress, short shots generally, you aren't tying up the talent etc. if you have to re-shoot because your card skipped or died. Shallow DoF is good for the hackneyed rack focus effect if any are still using that, and the 5D3 is a very good portrait camera up at 85mm or so, so having a B cam do closeups with it makes sense, something to cut away to, that you don't need to worry about as it's not the master shot.   But more interesting is if they implement an onboard codec like MJPEG, what resolutions they could send out to the fast CF cards. I think 4k is a near certainty in this case. I mean, if they can crop 1:1 pixel they do have native res RAW access. So why can't they stream out 4k MJPEGS?   You know, Canon may have told ML they could do the RAW thing as a way of killing off BMD. Which it certainly will (who would have bought one of those before this anyway?). But Canon has made clear they don't want the 1DX's turned into 1DC's. I can't see anything else in the way of 4K on the 5D3, and it will be at least as good as 1DC video (which is limited by the same CF interface). And no nothing will burn up, there is already an optimized JPEG compressor onboard, and the stills are limited by shutter/mirror speed and SD compatibility which this won't involve.   So there's your next mission. Get 4K for us. For those who value resolution over color accuracy (I doubt 4K RAW could be done in UDMA 7, but maybe). That will not only kill off BMC it will kill off Scarlet.

Again Andrew, the C100/C300 ergonomics aren't overrated, and I own, paid for, and shoot with both the 5D3 and C100. I shot only with DSLRs until the C100 appeared. The C100 is vastly better for handheld work...there is no comparison. I've posted some of the most thorough critiques of the C100 ergonomics, and even with those complaints (some of which they are fixing, such as moving the focus assist point) it wouldn't come close to my list of complaints with the 5D3 ergonomics. Even with ML and/or a Ninja or SmallHD helping, and various Zacuto and other addons that were contrived to make an unworkable situation a bit more pleasant. The only real ENG weakness for the C100/C300 is the lack of an inbuilt shoulder mount...how's the built-in shoulder mount on your DSLR? Or power zoom for that matter?   Really the RAW capability is impressive technologically but it's a year or two ahead of its time, while the C100/C300/C500 are here and working today and incredibly pleasant tools. One-touch custom white balance with a dedicated button, expose for the highlights in Cinelock using the waveform scopes, use the Ninja with 4-channel audio capability (I use a Rode Stereo VideoMic Pro and the XLRs in the handle if I need them) direct to ProRes, apply a LUT (I use Pomfort's DSLRLog2Video even though that was intended for Cinestyle, it works great with Canon Log too for $29) and Bob's your proverbial uncle. It works. It looks great. You swear by it rather than at it. People using such setups don't post many example videos, they are too busy making money with them and they don't see anything that's a genuine improvement that will net them more money otherwise.   I may well use the ML RAW if it gets worked out, and you can claim bragging rights for being its advocate from the beginning, but pissing on things you don't understand either is the pot-kettle problem. I'm an advocate for the less financially empowered filmmaker like you are, but not at the expense of attacking the better off ones that are making very sound decisions you aren't privy to.  

First of all, Apple requires a license for the use of ProRes, and it's just about unthinkable they will give one to Magic Lantern. Secondly, the whole point of using RAW is to manipulate the curves you desire *prior* to encoding in ProRes or DNx for your cutting. Cinema5D illustrates this well in their 5D3 RAW workflow video. 5D3 RAW is 14 bit linear (how many actual stops the sensor gives has not been determined yet; DxO claims 11.9 stops at base ISO for 8MP resolution, 2MP resolution may have more than that, or perhaps less, depending on how Canon downsampled). That allows you to pull your own custom "gamma" so to speak for encoding into ProRes, you can pull a whole custom look if you want with full color accuracy and without having to LUT a log gamma. A nice advantage though currently the workflow is unfacilitated. The Canon 1.2.1 firmware for the 5D3 does indeed provide uncompressed clean HDMI output at 8 bit 422 for recording onto a Ninja or other external recorder. It's a nice image when used with Cinestyle and run through a LUT, and much better than the internal recording. I hope ML is able to get RAW working simultaneously with all the normal 1.2.1 features so that we can have an internal proxy and an external ProRes. Just like most of the unfortunates who bought into the BMCC hype record ProRes rather than RAW, I think most of the 5D3 owners will use the clean HDMI out for most of their work unless ML implements a better internal codec such as a very high quality MJPEG. But as I mentioned they will most likely never be able to record to ProRes or DNx internally due to licensing encumberances.

FYI http://www.lensrentals.com/blog/2012/09/lensrentals-repair-data-january-july-2012

On the subject of reliability, Roger Cicala reports that the most frequent camera damage coming back from rentals on the 5D3 is bent CF pins inside the camera's slot. If there's a grain of sand caught in those little holes on the card or if you just jam it in there a bit askew your camera is SD-only until repaired. It happened so often on the 5D2 they made it easier for service to repair the pins but they didn't make it harder to break in the first place, that's the CF standard's problem. And the 5D3 has only one CF slot, and it's the only way you're getting RAW out of it...and you're gonna be doing insertions every 15 minutes or so rather than once or twice a day...   I have no conflicts of interest and I've never even been in contact with anyone from Canon. I'm just running through the other sides of this, immune enough to the hype to not have already bought $500 worth of CF cards but not immune enough to not be thinking it all through very carefully. Please resist group polarization...there's a waft of Lord of the Flies in this whole personal allegiances thing.

I am not concerned about the camera burning out as much as I am about the cards. A lot more write cycles and a lot more intense then typical expectation. But I guess you can view them as disposable if your work deserves RAW treatment. However there is no way at all to have redundancy, and you can easily blow a $5000 shooting day if your cards are toast at the end of it. Total failure you might detect with periodic offloads but the worst would be glitches and skips right in your best takes that you don't see till long after wrap.   Simultaneous uncompressed HDMI out to a Ninja would be a nice safety...a proxy on the internal SD card would be at least something. I always record to the internal SD cards on the C100 even though they are only safeties for the Ninja footage. Regardless, we are going to learn everything about the state of Compact Flash in the coming weeks.

Using the Ninja 2 and Neat Video with both the C100 HDMI out (with Canon Log) and the 5D3 HDMI out (with Cinestyle) yields perfectly gradeable and usable footage provided a camera operator able to set a custom white balance and expose for the highlights properly. Those are the best "official" (unhacked) low light options available under $10K and the Ninja is a joy to use and cut the footage from....you can literally shoot all day and not worry about batteries or storage and the ProRes or DNx drops straight into your NLE with scenes and shots and takes all numbered for you. That is absolutely a professional workflow and any operator that can call themselves professional will be able to get a completely broadcast or big screen-ready result that way. There's a bit of pot-meet-kettle with all the criticism of Bloom and Laforet not falling all over themselves publicly over the ML RAW hack. To avoid being guilty of the same unfairness we can't dismiss entirely what's already officially here and working for real productions rather than a few image quality junkies. The RAW image will likely give more latitude and color accuracy, but I'm not yet sure it provides more DR or resolution than the HDMI out with a log gamma. We need scientific tests to determine that and to date ZERO have been posted, just a bunch of image junkie anecdata examples.

  Happy to send you my 5D3 in even trade for your C300.  B)

it's a nice look, and that combo along with the 5D3 RAW should carry me through the 1080p era. Without feeling like I need to wipe my arse across the carpet!  :o

Whatever gain setting you have at the sensor can be compensated for automatically to adjust it to your bit depth. They know the level of the sensor noise floor and they can just keep that near the LSB, meaning the sensels will saturate before any additional headroom in the bit depth can be taken advantage of. An implementation that doesn't do this is just wasting a lot of bandwidth, or wasting the actual DR of its sensor...the sensor DR is the limiting factor and getting the specific gain mapping right once and for all isn't hard.

Best use of a Blackmagic cage to date!

It possible that adding grain will help cope with compression, as a form of dither. Because part of the cost of compression is banding and macroblocking, and the grain might break that up a bit and make it less visible. Yes the extra entropy in the signal will mean the compressor has to work harder but try it and see if it's a bad thing overall.

Well I can tell you if you had a C300 you would most likely be very happy with it and not particularly interested in owning anything else at the moment. Especially if you are a name that is hired for reasons other than your camera (and you can go ahead and cast aspersions to what those may be). I think very few working professionals who are making enough money to save some at the end of the year are hired for their camera. And many of them are more interested in...uh oh...story and composition and lighting and the other things than the camera. The C300 is just about the least pain-in-the-arse camera there is and the picture out of it is very very good 1080p and very gradeable in Canon log. Phenomenal low light. Gorgeous build and ergonomics. Very few quibbles, and some of them are being fixed in firmware. I don't think Philip Bloom's ethics statement is broad enough...he claims he doesn't get paid to review cameras, and that he pays for his own cameras. OK, but he gets paid by manufacturers of those cameras for other things, such as shooting demo videos (as he did for Panasonic GH3), doing speaking engagements, etc. He very likely wouldn't be paid for those things if he wasn't the most famous video camera reviewer in the world. So yes he is a bit interest conflicted. But even so, I don't think that's why he shoots C300. Canon doesn't need to bribe undercover spokespeople as some smaller firms apparently do. I think he just finds that camera fits his needs perfectly and lets him focus on other things. As my C100 (supplemented with the Ninja) does for me (I paid full retail for it, the original $6500, and I have never been paid or otherwise compensated by Canon or Atomos or any other related manufacturer or distributor ever for any reason). I'm sure he will cover the 5D3 ML RAW hack when it settles down, it's not a conspiracy Andrew. I wonder whether you really believe these things or you think the controversial claims make the blog more exciting.

Cinema 5D is finding it faster to use ACR in Photoshop to TIFF and then assemble in QT7Pro like we'd do a time-lapse.   http://***URL not allowed***/?p=18065   Before I order cards I'm currently waiting for:   1) stable build 2) record time only limited by memory (>4GB) 3) no dropped frames with 1000x cards @1080p24   I'd like:   4) Simultaneous H.264 proxy recorded to SD card, for take review, sound, etc. 5) Simultaneous HDMI out, ideally 8 bit 422 uncompressed as an option, or mirroring of onboard LCD 6) I assume the other ML goodies still work like peaking 7) High quality internal codec option instead of RAW

The bit depth does not limit the Dynamic Range. It limits the number of distinguishable luma and chroma levels, which we can call "fidelity."   A "gamma" is used to map a given dynamic range to a given digital pixel code number. In a linear mapping, you need 1 bit for each 1 stop. But in say the standard sRGB mapping you are likely viewing this site with, you can map about 11 stops of DR to 8 bits, with fairly minor loss of distinguishable luma and chroma values. In a log gamma, you can map even more stops to a given bit depth. The more bits you have to work with the more levels you can distinguish and the less you need a log gamma.   Therefore the 8 bits of the HDMI output of the 5D3 and C100, using a log gamma (Cinestyle and Canon Log respectively) can represent the entire 11-12 stops of DR tests show those cameras capable of at their native ISOs. Will that you lose some fidelity, and therefore be susceptible to color banding with an extreme grading curve, compared to a linear mapping used for RAW? Sure, but you are cutting your storage needs per pixel nearly in half with not that much loss in fidelity. 14 bit RAW for an 11 stop sensor is overkill; the extra 3 bits will just be noise or unused headroom anyway. Therefore you shouldn't evaluate bit depth as anything more than the potential for fidelity, not a guarantee of it.

If they shot at f/22 then the different diffraction characteristics of the lenses and their irises will also be an issue. One would hope a comparison would be done at a lens's best settings. And with primes rather than zooms I would think. Again I apologize for complaining about the tests when not being able to run better ones of my own. A/B tests are the most useful but as a result carry the greatest responsibility in their design, execution, reporting and analysis. I want to get someone out there determined to meet that challenge. The manufacturers will fall into line very quickly as they did when DPR got big.

Thanks, and it's sad the Zacuto shootout is about the best there is currently. Stills shooters have DPR and DxO and TDP and therefore we know exactly how the D800 and 1DX etc. compare on an objective, reproducible, apples-to-apples basis, even if we might have quibbles with that basis. Video? Crickets... Another thing to consider when comparing the 5D3 RAW with the BMCC is the crop option in ML. As I understand it, ML will allow 1:1 pixel crops, which will essentially make the full frame 135 5D3 into a little BMCC pocket cam sensor (I generally dismiss the BMCC as a 3lb GoPro with the lens removable but the battery not). I predict there will be poorer false color and aliasing suppression, and more noise in that mode, and it will be more directly comparable with the BMCC. The BMCC won't have the full frame mode anytime soon though. B) On the subject of file time limits, it's clear from the ML forums that the 4GB is just a temporary issue. I predict a practical 1080p/25 limit of 30 minutes continuous on a 128GB card, which our friends in the European gov't should be very happy to hear! :rolleyes:

It would be easier for me to just take your word for it if you were really precise in making and backing up your statements. Shots of boulders? =)    Also I am discounting your reports as I have confirmed quite a dramatic increase in quality using Ninja 2 with 5D3 HDMI out vs. the same settings internal. Look at the noise floor for instance. Look at color edges. There is no comparison.   I'm procrastinating getting a card seeing as the Komputerbay is so much cheaper (perhaps the savings is on poor QA?) and we're only now getting feedback on whether it works or not. The 128GB and 64GB cards are completely different implementations...and what's more, the production runs of these cards can be done in different factories with different components without notice. Furthermore I would rather just get a dependable bin of the ML that has more or less settled than a nightly I have to compile. My primary is the C100 + Ninja 2 and the 5D3 is my B cam and stills so I'm not in a mad rush. I'm also procrastinating getting a second Ninja as they just came out with the Samurai Blade with the far better screen and it shouldn't be long before they have an HDMI version rather than having to buy one of their HDMI converters (which are very nice looking btw, but $300 more, and only mount to their hardware).  

I'm sure you know we are discussing two different lenses feeding two very different size sensors. As a result, the focal plane position and depth of field for the two shots will not be comparable directly. The BMCC will have a much greater DoF and so even if the focal plane is at the same point in the image (I don't see that we've been given confirmation of that, typically these unscientific tests are completely unreproducible because the testers are too lazy to tell us exactly what was done in them) as we move away from that plane in either direction the smaller sensor will retain smaller circles of confusion optically at the sensor. This optical advantage can easily make up for significant disadvantages of the sensor and later processing (visible in the BMCC's false color artifacts on the pebbles for instance, which you had to butcher to tame). Which is why the first comparisons to be made should be test charts (properly focused of course, with lenses of similar scientifically validated performance even if not identical due to mount/sensor size differences). Why we can't get 100% reproducible test chart data first of all I have no explanation to provide you. We can still argue somewhat objectively with this anecdata but we have to be very careful in understanding everything we may be seeing before declaring something "proven."

  But what did the smearing? Everyone agrees that on the 5D3 the internal video recording sucks (regardless of picture style used) and 22MP RAW stills are the best you can do. However we can't get 24 22MP RAW stills per second, at least not unless there is some massive development on the CF card front (some have suggested an SSD adapter of some form). So there is detail that's being smeared away for the ML hack just as there is for the uncompressed HDMI out, which is a fairer thing to compare it with.   Is the ML RAW hack in its full sensor mode using the same 3x3 pixel binning downsampler that the uncompressed HDMI out is? Because if it is, that's the same detail damage. The difference in luma detail should then be minimal, unless Canon is damaging the HDMI image somehow. ML hasn't figured out the HDMI driver yet but they do suggest that some image processing is being done there (it does appear a bit brighter of an image I think). But is Canon intentionally crippling that image as they clearly did for the internal recording (40Mbps H.264 should look a LOT better than what they are giving us...)? I don't know but a fair, scientific, detailed comparison between the two would tell us. We know that chroma detail will be cut in half (422 vs. 444) and we know we will only be mapping the DR into 8 bits rather than 14, but that will only result in gradient accuracy being lost (i.e. banding) rather than detail and acutance/sharpness being lost.   So I want to know how bad the uncompressed HDMI actually is vs. the ML RAW...I predict it won't be noticeably worse when Cinestyle + LUT is used, and with Neat Video applied properly even a strong grade should hold up well. If not, if Canon really is damaging the supposedly uncompressed HDMI out beyond its 1080i60/8 bit 422 that we all know already, then we have a new thing to yell at them for and praise ML for saving us from. Otherwise we have a choice between buying a Ninja 2 and a couple cheap laptop drives and getting a usable field monitor in the bargain, or investing in a stack of the fastest CF cards available and trying to figure out a practical RAW workflow that's worth the hassle. Again I would have already answered this for you if I had the fast CF cards and a reliable ML build...asking if there's anyone out there who is trustworthy enough to do so for us now.

  I completely agree that Neat Video is doing a lot more to restore the image and remove compression artifacts/banding than just NR. Anytime I'm doing a severe grade it's got to be on there...it's basically the best $99 you can spend in post. And far less hassle than RAW.   However I think it's laughable that you suggest ProLost Neutral is better than Cinestyle. And that Cinestyle has poorer highlight handling! Lol!   I will try Canon's new "X" style (released the other day but lost in the ML fuss) to see if that's like a Cinestyle that doesn't raise the blacks so much, saving a bit (update...it isn't but it's a nice look anyway). But for the moment I haven't found anything overall better. RAW will of course be the best we can do, but at what cost? Also note, "you can change the WB in RAW without losing anything"...no, your noise floor is going to shift with a WB shift, not much but some, so there will still be some loss...but the loss won't be any more than the loss would be if you set it properly in-camera. One thing the 5D3 lacks that the C100 has is the ABB (Automatic black balance) function that neutralizes the noise floor for a given ISO and WB setting...RAW will cover that too but again at enormous data and workflow cost.   Just give us a great internal codec, whatever you can, ML!